% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % % Project Gutenberg's Utility of Quaternions in Physics, by Alexander McAulay % % % 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: Utility of Quaternions in Physics % % % % Author: Alexander McAulay % % % % Release Date: August 11, 2008 [EBook #26262] % % Most recently updated: June 11, 2021 % % % % Language: English % % % % Character set encoding: UTF-8 % % % % *** START OF THIS PROJECT GUTENBERG EBOOK UTILITY OF QUATERNIONS IN PHYSICS *** % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \def\ebook{26262} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %% %% %% Packages and substitutions: %% %% %% %% book: Required. %% %% geometry: Enhanced page layout package. Required. %% %% %% %% inputenc: Standard DP encoding. Required. %% %% amsmath: AMS mathematics enhancements. Required. %% %% amssymb: Additional mathematical symbols. Required. %% %% yfonts: Fraktur fonts, fall back to bold. Optional. %% %% alltt: Fixed-width font environment. Required. %% %% footmisc: Enhanced footnote capabilities. Required. %% %% perpage: Restart footnote numbering on each page. Required. %% %% txfonts: PostScript Times fonts. Required. %% %% indentfirst: Indent after headings. Recommended. %% %% %% %% ifthen: Logical conditionals. Required. %% %% %% %% fancyhdr: Enhanced running headers and footers. Required. %% %% extramarks: Part of fancyhdr %% %% graphicx: Standard interface for graphics inclusion. Required. %% %% %% %% hyperref: Hypertext embellishments for pdf output. Required. %% %% %% %% Producer's Comments: The book contains a lot of in-line math and %% %% uses idiosyncratic notation (particularly %% %% an upside-down D symbol), unusual titling, %% %% numbering, and ToC; the pagination should %% %% be straightforward, however. %% %% Custom \Chapter, \Section, and \Paragraph %% %% commands implement the book's divisions, %% %% see "formatting details" below. %% %% %% %% %% %% Things to Check: %% %% %% %% Spellcheck: OK %% %% Smoothreading pool: Yes %% %% lacheck: OK, false positives: %% %% unmatched braces, et. al. (ten instances) %% %% bad character in label (three instances, in macro defs) %% %% missing `\@' before `.' (in commented region) %% %% deliberate space after { (seven instances) %% %% multiple complaints from log file %% %% %% %% Lprep/gutcheck: OK %% %% PDF pages: 134 %% %% PDF page size: US Letter (8.5" x 11") %% %% PDF bookmarks: created, point to ToC entries %% %% PDF document info: filled in %% %% ToC page numbers: OK %% %% Images: 2 PDF %% %% %% %% Summary of log file: %% %% * Six underfull vboxes. %% %% * One overfull hbox (smaller than 2.24pt). %% %% %% %% Formatting details: %% %% %% %% The book's divisions don't map well to plain LaTeX, so the %% %% preamble defines the following commands: %% %% %% %% 1. \Chapter: Primary chapter-like headings, which the book calls %% %% ``Sections''. %% %% %% %% 1a. \FNChapter (footnote chapter): The title of ``Section III'' %% %% contains a footnote in the title, which should not go into the %% %% toc or running head. %% %% %% %% 1b. \HeadChapter: The title of ``Section I'' is preceded by a %% %% half title page. %% %% %% %% 1c. \Preface: Formatted like a \Chapter, but no toc entry. %% %% %% %% %% %% 2. \Paragraph: Numbered divisions formatted like paragraphs, %% %% but referenced internally with the ``section'' symbol \S. By %% %% default these are not listed in the toc, but their numbering %% %% affects the toc. %% %% %% %% The number of \Paragraph{94} is deliberately not indented, to %% %% leave space on the line for an in-line math expression. %% %% %% %% 2a. \Section: \Paragraph having a toc entry, and labeled in the %% %% text with a centered italic title. %% %% %% %% 2b. \SectionTitle: No toc entry, but having a centered italic %% %% title. %% %% %% %% 2c. \TCSection (``ToC section''): A \Section whose title and %% %% toc entry differ (e.g., because of a footnote). %% %% %% %% %% %% 3. Equations are tagged and referenced with \EqnTag and \Eref. %% %% \EqnTag{section}{equation} prints a tag and issues a label %% %% command, \Eref{section}{equation} generates the corresponding %% %% reference. For instance, \EqnTag{IV}{16} makes its equation %% %% referrable with \Eref{IV}{16}. %% %% %% %% %% %% Compile History: %% %% %% %% August, 2008: adhere (Andrew D. Hwang) %% %% texlive2007, GNU/Linux %% %% %% %% August, 2008: dcwilson %% %% MiKTeX 2.7, Windows XP Pro %% %% %% %% pdflatex x3 (Run pdflatex three times) %% %% %% %% %% %% August 2008: pglatex. %% %% Compile this project with: %% %% pdflatex 26262-t.tex ..... THREE times %% %% %% %% pdfeTeX, Version 3.141592-1.30.5-2.2 (Web2C 7.5.5) %% %% %% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%% PACKAGES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \listfiles \documentclass[12pt,letterpaper]{book}[2005/09/16] \usepackage{amsmath}[2000/07/18] % AMS math enhancements \usepackage{amssymb}[2002/01/22] % Extra math symbols \usepackage{ifthen}[2001/05/26] % Logical conditionals \usepackage[utf8]{inputenc}[2006/05/05] % DP standard encoding \usepackage{graphicx}[1999/02/16] % publisher's mark, upside-down D operator % Define \textgoth; document subsequently defines \Mygoth \IfFileExists{yfonts.sty}% {\usepackage{yfonts}[2003/01/08]} % fraktur font (titlepage only) {\providecommand{\textgoth}[1]{\textbf{##1}}} % fallback if no yfonts \usepackage{alltt}[1997/06/16] % for boilerplate, credits, license \usepackage{indentfirst}[1995/11/23] \usepackage[symbol]{footmisc}[2005/03/17] \usepackage{perpage}[2006/07/15] % footnote numbering restarts on each page \usepackage{txfonts}[2005/01/03] % PostScript Times font \usepackage{fancyhdr,extramarks} % running heads %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%%% Interlude: Set up PRINTING (default) or SCREEN VIEWING %%%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % ForPrinting=true (default) false % Asymmetric margins Symmetric margins % Black hyperlinks Blue hyperlinks % Even-odd page headings Uniform page headings % Start Preface, ToC, etc. recto No blank verso pages % % Chapter-like ``Sections'' start both recto and verso in the scanned % book. This behavior has been retained. \newboolean{ForPrinting} %% *** COMMENT the next line for a SCREEN-OPTIMIZED VERSION *** %% \setboolean{ForPrinting}{true} %% Initialize values to ForPrinting=false \newcommand{\Margins}{hmarginratio=1:1} % Symmetric margins \newcommand{\HLinkColor}{blue} % Hyperlink color \newcommand{\PDFPageLayout}{SinglePage} \newcommand{\TransNote}{Transcriber's Note} \newcommand{\TransNoteText}{% This PDF file is formatted for screen viewing, but may be easily recompiled for printing. Please see the preamble of the \LaTeX\ source file for instructions. } %% Re-set if ForPrinting=true \ifthenelse{\boolean{ForPrinting}}{% \renewcommand{\Margins}{hmarginratio=2:3} % Asymmetric margins \renewcommand{\HLinkColor}{black} % Hyperlink color \renewcommand{\PDFPageLayout}{TwoPageRight} \renewcommand{\TransNote}{Transcriber's Notes} \renewcommand{\TransNoteText}{% Table of contents entries, section references, and equation numbers are internal hyperlinks. \bigskip This PDF file is formatted for printing, but may be easily recompiled for screen viewing. Please see the preamble of the \LaTeX\ source file for instructions. } }{% If ForPrinting=false, don't skip to recto \renewcommand{\cleardoublepage}{\clearpage} } %%%% End of PRINTING/SCREEN VIEWING code; back to packages %%%% %%%% Text block size %%%% % The printed book's text block is well-approximated by hscale=0.5, but % the painstakingly-chosen value hscale=0.63 minimizes overfull hboxes % and is not uncomfortable to read, ~65 chars per line. % The vscale value is the package default, specified explicitly to guard % against changes to geometry.sty. % N.B. global letterpaper option above \usepackage[hscale=0.63,vscale=0.7,\Margins]{geometry}[2002/07/08] %% Alternative text block settings for Computer Modern Roman %% %\usepackage[hscale=0.67,vscale=0.7,\Margins]{geometry}[2002/07/08] %%%% PDF and hyperref options %%%% \providecommand{\ebook}{00000} % Overridden during white-washing \usepackage[pdftex, hyperfootnotes=false, pdftitle={The Project Gutenberg eBook \#\ebook: Utility of Quaternions in Physics}, pdfauthor={Alexander McAulay}, pdfkeywords={Joshua Hutchinson, Andrew D. Hwang, Carolyn Bottomley, Project Gutenberg Online Distributed Proofreading Team, Cornell University}, pdfstartview=Fit, % default value pdfstartpage=1, % default value pdfpagemode=UseNone, % default value bookmarks=true, % default value linktocpage=true, pdfdisplaydoctitle, pdfpagelayout=\PDFPageLayout, pdfpagelabels=true, bookmarksopen=true, bookmarksopenlevel=1, colorlinks=true, linkcolor=\HLinkColor]{hyperref}[2007/02/07] % Re-crop screen-formatted version \ifthenelse{\boolean{ForPrinting}} {} {\hypersetup{pdfpagescrop= 65 90 547 752}} %%%%%%%%%%%%%%%%%%%%%%%%%%%%% PREAMBLE %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % No number on first page of Table of Contents \AtBeginDocument{\addtocontents{toc}{\protect\thispagestyle{empty}}} % To prevent underfull hboxes \newcommand{\stretchyspace}{\spaceskip0.5em plus 0.5em minus 0.25em} % To improve spacing on titlepage \newcommand{\Stretchout}{\spaceskip0.85em} %%%% Fixed-width environment to format PG boilerplate %%%% % 9.2pt leaves no overfull hbox at 80 char line width \newenvironment{PGtext}{% \begin{alltt} \fontsize{9.2}{10.5}\ttfamily\selectfont}% {\end{alltt}} %%%% Reset Table of Contents heading %%%% \renewcommand{\contentsname} {\protect\centering\protect\normalfont\protect\scshape{% \protect\large CONTENTS.}} %%%% Footnote numbers restart on each page %%%% \MakePerPage{footnote} %%%% \Chapter and related macros %%%% % Set up page for start of \Chapter \newcommand{\ChapterPrep}[1]{% \clearpage % [sic]; ``Sections'' start both recto and verso \phantomsection \vspace*{0.75in} \pagestyle{fancy} \thispagestyle{empty} \fancyhead{} \fancyfoot{} \renewcommand{\headrulewidth}{0pt} \fancyhead[C]{{\scshape\small\MakeLowercase{#1.}}} \ifthenelse{\boolean{ForPrinting}} {\fancyhead[RO,LE]{\thepage}} {\fancyhead[R]{\thepage}} \setlength{\headheight}{14.5pt} } % increment counter, issue label \newcommand{\ChapterRef}[2]{% \refstepcounter{chapter} \label{chapter:\thechapter} \pdfbookmark[0]{#2}{#2} \addtocontents{toc} {\protect\bigskip\normalfont\scshape\protect\centering % #1.\protect\quad #2\protect\medskip} } %% Macros used in file body %% % Special format, no toc for Preface \newcommand{\Preface}{% \ChapterPrep{Preface} \begin{center} \textsc{\Large Preface}. \end{center} } % For Chapter I, which has a pre-title \newcommand{\HeadChapter}[3]{% \ChapterPrep{#2} \ChapterRef{#1}{#2} \begin{center} \textsc{\Large #3}. \bigskip \textsc{\Large #1}. \bigskip \textsc{\large #2}. \end{center} } % For Chapter III, whose title contains a footnote \newcommand{\FNChapter}[3]{% \ChapterPrep{#2} \ChapterRef{#1}{#2} \begin{center} \textsc{\Large #1}. \bigskip \textsc{\large #3}. \end{center} } % For others. Usage: \Chapter{Section N}{Title} \newcommand{\Chapter}[2]{\FNChapter{#1}{#2}{#2}} %%%% \Paragraph, \Section, and related macros %%%% % Space between Section number and title in ToC \newlength{\Tocskip} \setlength{\Tocskip}{2em} % Add toc line: \Tocline{n}{title} or \Tocline{}{title} \newcommand{\Tocline}[2]{% \phantomsection \ifthenelse{\equal{#1}{}}{% #1 is the null string? \addcontentsline{toc}{section}{% \texorpdfstring{\protect\makebox[\Tocskip]{}% \protect\normalfont\protect\footnotesize{\protect{#2}} }{\protect{#2}} } }{% Otherwise \addcontentsline{toc}{section}{% \texorpdfstring{\protect\makebox[\Tocskip][l]{\protect{#1}.}% add period \protect\normalfont\protect\footnotesize{\protect{#2}} }{\protect{#2}} } } } % For paragraphs 3a, 16a, and 45. Usage: \FNParagraph{label}{tag} \newcommand{\FNParagraph}[2]{% [** PP: Dirty trick -- don't indent \S94] \ifthenelse{\equal{#2}{94}}{\paragraph{#2.}}{\paragraph{\indent#2.}} \phantomsection \label{section:#1} % Set up section numbers in running head \extramarks{\S~#1.}{\S~#1.} \ifthenelse{\boolean{ForPrinting}} {\fancyhead[RE]{[\,\firstleftxmark} % Printed: par. range verso-recto \fancyhead[LO]{\lastrightxmark\,]}} % {\fancyhead[L]{[\,\firstleftxmark\,]}} % Screen: ``initial'' par. % \setlength{\headheight}{14.5pt} % Either case } \newcommand{\Paragraph}[1]{\FNParagraph{#1}{#1}} % Title only; used in \TCSection defn, and four times in the text \newcommand{\SectionTitle}[1]{% \subsubsection*{\centering\normalfont\itshape #1.} % Add dot } % Most general Section-setting command; handles sections whose % title contains a footnote. Usage: \TCSection{tag}{ToC title}{Title} \newcommand{\TCSection}[3]{% \SectionTitle{#3} \Tocline{#1}{#2} } % Most common Section-setting command; title = ToC entry \newcommand{\Section}[2]{\TCSection{#1}{#2}{#2}} %%%% Equation and section cross-referencing %%%% % ``equation tag'', labels equation with Roman section and equation number \newcommand{\EqnTag}[2]{% \tag{#2}% \label{eqn:#1.#2}% e.g. \label{eqn:IV.45} = equation 45 in Section IV } % ``equation reference'', e.g., \Eref{V}{45} -> \EqnTag{V}{45} \newcommand{\Eref}[2]{\hyperref[eqn:#1.#2]{(#2)}} \newcommand{\equationref}[2]{\hyperref[eqn:#1.#2]{equa\-tion~(#2)}} \newcommand{\Equationref}[2]{\hyperref[eqn:#1.#2]{Equa\-tion~(#2)}} % ``section reference'' \newcommand{\Secref}[1]{\hyperref[section:#1]{\S\,#1}} % Pointers to errata \newcommand{\Erratum}[1]{\phantomsection\label{#1}} %%%% Miscellaneous macros %%%% \newcommand{\McAulay}{M\raisebox{0.45ex}{c}AULAY} % Reverts to \textbf if yfonts not available \newcommand{\Mygoth}[1]{\textgoth{#1}} \newcommand{\ds}{d\mathfrak{s}} \newcommand{\smalliint}{{\textstyle\iint}} \newcommand{\smalliiint}{{\textstyle\iiint}} % Upside-down D operator \newcommand{\Dop}{\raisebox{-0.4ex}{\rotatebox[origin=c]{180}{$\mathrm{D}$}}} % set first arg as affix on second arg \newcommand{\Pre}[2]{{}_{#1}{#2}} \newcommand{\Nabla}[1]{\Pre{#1}{\!\nabla}} %%%%%%%%%%%%%%%%%%%%%%%% START OF DOCUMENT %%%%%%%%%%%%%%%%%%%%%%%%%% \begin{document} \pagestyle{empty} \pagenumbering{Roman} \pdfbookmark[0]{Project Gutenberg Boilerplate}{Project Gutenberg Boilerplate} \begin{center} \begin{minipage}{\textwidth} \begin{PGtext} Project Gutenberg's Utility of Quaternions in Physics, by Alexander McAulay 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: Utility of Quaternions in Physics Author: Alexander McAulay Release Date: August 11, 2008 [EBook #26262] Most recently updated: June 11, 2021 Language: English Character set encoding: UTF-8 *** START OF THIS PROJECT GUTENBERG EBOOK UTILITY OF QUATERNIONS IN PHYSICS *** \end{PGtext} \end{minipage} \end{center} \vfill %% -----File: 001.png--- % Formatted, commented production note \iffalse \null\vfill \begin{center} \begin{minipage}{3in} \renewcommand{\baselinestretch}{0.8} \begin{verbatim} Production Note Cornell University Library produced this volume to replace the irreparably deteriorated original. It was scanned using Xerox software and equipment at 600 dots per inch resolution and compressed prior to storage using CCITT Group 4 compression. The digital data were used to create Cornell's replacement volume on paper that meets the ANSI Standard Z39.48-1984. The production of this volume was supported in part by the Commission on Preservation and Access and the Xerox Corporation. 1991. \end{verbatim} \renewcommand{\baselinestretch}{1} % restore default value \end{minipage} \end{center} \vfill \fi %% -----File: 002.png--- % [Blank Page] %% -----File: 003.png--- \iffalse Cornell University Library BOUGHT WITH THE INCOME FROM THE SAGE ENDOWMENT FUND THE GIFT OF Henry W. Sage 1891 A.52629 MATHEMATICS 24/10/93 \fi %% -----File: 004.png--- % [Blank Page] %% -----File: 005.png---Folio i------- %%%%%%%%%%%%%%%%%%%%%%%%%%% FRONT MATTER %%%%%%%%%%%%%%%%%%%%%%%%%% \frontmatter \pagestyle{empty} \pagenumbering{Alph} %% Half title page \null \vfil \begin{center} \textbf{\Large UTILITY}\\[36pt] \textsc{of}\\[36pt] \textbf{\Large\Stretchout QUATERNIONS IN PHYSICS.} \end{center} \vfil \clearpage %% -----File: 006.png---Folio ii------- % [Illustration: *Emblem* M & M Co.] \null \vspace{0pt plus 2fil} \begin{center} \includegraphics[width=1.5in]{./images/pubmark.pdf} \end{center} \vfil \clearpage %% -----File: 007.png---Folio iii------- %% Title page % [** PP: Pagesize-dependent dimension] \enlargethispage{12pt} \noindent\begin{minipage}{\textwidth} \centering \textbf{\huge UTILITY}\\[36pt] \textsc{\small of}\\[36pt] \textbf{\huge\Stretchout QUATERNIONS IN PHYSICS.}\\[144pt] \textsc{\small by}\\[18pt] \textbf{\Stretchout A.~\McAulay, M.A.,} \textsc{\scriptsize{% LECTURER IN MATHEMATICS AND PHYSICS IN THE UNIVERSITY OF TASMANIA}.% }\\[144pt] \large \Mygoth{London}: \textbf{\Stretchout MACMILLAN AND CO.} \textbf{\footnotesize\Stretchout AND NEW YORK.} 1893\\[4pt] [\textit{\footnotesize All Rights reserved}.] \end{minipage} \clearpage %% -----File: 008.png---Folio iv------- \iffalse \null \vspace{0pt plus 2fil} \begin{center} \textbf{\Mygoth{Cambridge}}: \medskip \textsc{\scriptsize PRINTED BY C. J. CLAY, M.A., AND SONS,} \medskip \textsc{\scriptsize AT THE UNIVERSITY PRESS.} \end{center} \vfil \clearpage \fi \begin{center} \begin{minipage}{\textwidth} \begin{PGtext} Produced by Joshua Hutchinson, Andrew D. Hwang, Carolyn Bottomley and the Online Distributed Proofreading Team at http://www.pgdp.net (This ebook was produced from images from the Cornell University Library: Historical Mathematics Monographs collection.) \end{PGtext} \end{minipage} \end{center} \vfill \begin{minipage}{0.8\textwidth} \small \SectionTitle{\TransNote} \raggedright % Behavior of \TransNoteText depends on value of ForPrinting. % If true, explains that page numbers et. al. are hyperlinks. % In either case, explains how to recompile to change the option. \TransNoteText \end{minipage} \null \vspace{0pt plus 2fill} \begin{center} \textbf{\Mygoth{Cambridge}}: \medskip \textsc{\scriptsize PRINTED BY C.~J.~CLAY, M.A., AND SONS,} \medskip \textsc{\scriptsize AT THE UNIVERSITY PRESS.} \end{center} \cleardoublepage %% -----File: 009.png---Folio v------- \pagenumbering{roman} \Preface The present publication is an essay that was sent in (December,~1887) to compete for the Smith's Prizes at Cambridge. To the onlooker it is always a mournful thing to see what he considers splendid abilities or opportunities wasted for lack of knowledge of some paltry common-place truth. Such is in the main my feeling when considering the neglect of the study of Quaternions by that wonderful corporation the University of Cambridge. To the alumnus she is apt to appear as the leader in all branches of Mathematics. To the outsider she appears rather as the leader in \emph{Applied} Mathematics and as a ready welcomer of other branches. If Quaternions were simply a branch of Pure Mathematics we could understand why the study of them was almost confined to the University which gave birth to them, but as the truth is quite otherwise it is hard to shew good reason why they have not struck root also in Cambridge. The prophet on whom Hamilton's mantle has fallen is more than a mathematician and more than a natural philosopher---he is both, and it is to be noted also that he is a Cambridge man. He has preached in season and out of season (if that were possible) that Quaternions are especially useful in Physical applications. Why then has his Alma Mater turned a deaf ear? I cannot believe that she is in her dotage and has lost her hearing. The problem is beyond me and I give it up. But I wish to add my little efforts to Prof.~Tait's powerful advocacy to bring about another state of affairs. Cambridge is the prepared ground on which if anywhere the study of the Physical applications of Quaternions ought to flourish. %% -----File: 010.png---Folio vi------- When I sent in the essay I had a faint misgiving that perchance there was not a single man in Cambridge who could understand it without much labour---and yet it is a straightforward application of Hamilton's principles. I cannot say what transformation scene has taken place in the five years that have elapsed, but an encouraging fact is that one professor at any rate has been imported from Dublin. There is no lack in Cambridge of the cultivation of Quaternions \emph{as an algebra}, but this cultivation is not Hamiltonian, though an evidence of the great fecundity of Hamilton's work. Hamilton looked upon Quaternions as a \emph{geometrical} method, and it is in this respect that he has as yet failed to find worthy followers resident in Cambridge. [The chapter contributed by Prof.~Cayley to Prof.~Tait's 3rd~ed.\ of `Quaternions' deals with quite a different subject from the rest of the treatise, a subject that deserves a distinctive name, say, \emph{Cayleyan Quaternions}.] I have delayed for a considerable time the present publication in order at the last if possible to make it more effective. I have waited till I could by a more striking example than any in the essay shew the immense utility of Quaternions in the regions in which I believe them to be especially powerful. This I believe has been done in the `Phil.~Trans.'\ 1892, p.~685. Certainly on two occasions copious extracts have been published, viz.\ in the P. R. S. E., 1890--1, p.~98, and in the `Phil.~Mag.' June~1892, p.~477, but the reasons are simple. The first was published after the subject of the `Phil.~Trans.'\ paper had been considered sufficiently to afford clear daylight ahead in that direction, and the second after that paper had actually been despatched for publication. At the time of writing the essay I possessed little more than faith in the potentiality of Quaternions, and I felt that something more than faith was needed to convince scientists. It was thought that rather than publish in driblets it were better to wait for a more copious shower on the principle that a well-directed heavy blow is more effective than a long-continued series of little pushes. Perhaps more harm has been done than by any other cause to the study of Quaternions in their Physical applications by a silly superstition with which the nurses of Cambridge are wont to %% -----File: 011.png---Folio vii------- frighten their too timorous charges. This is the belief that the subject of Quaternions is difficult. It is difficult in one sense and in no other, in that sense in which the subject of analytical conics is difficult to the schoolboy, in the sense in which every subject is difficult whose fundamental ideas and methods are different from any the student has hitherto been introduced to. The only way to convince the nurses that Quaternions form a healthy diet for the \emph{young} mathematician is to prove to them that they will ``pay'' in the first part of the Tripos. Of course this is an impossible task while the only questions set in the Tripos on the subject are in the second part and average one in two years. [This solitary biennial question is rarely if ever anything but an exercise in \emph{algebra}. The very form in which candidates are invited, or at any rate were in my day, to study Quaternions is an insult to the memory of Hamilton. The monstrosity ``Quaternions and other % [** PP: Changing ``paralleled'' to ``parallelled''] non-commutative algebras'' can only be parallelled by ``Cartesian Geometry and other commutative algebras.'' When I was in Cambridge it was currently reported that if an answer to a Mathematical Tripos question were couched in Hebrew the candidate would or would not get credit for the answer according as one or more of the examiners did or did not understand Hebrew, and that in this respect Hebrew or Quaternions were strictly analogous.] Is it hopeless to appeal to the charges? I will try. Let me suppose that some budding Cambridge Mathematician has followed me so far. I now address myself to him. Have you ever felt a joy in Mathematics? Probably you have, but it was before your schoolmasters had found you out and resolved to fashion you into an examinee. Even now you occasionally have feelings like the dimly remembered ones. Now and then you forget that you are nerving yourself for that Juggernaut the Tripos. Let me implore you as though your soul's salvation depended on it to let these trances run their utmost course in spite of solemn warnings from your nurse. You will in time be rewarded by a soul-thrilling dream whose subject is the Universe and whose organ to look upon the Universe withal is the sense called Quaternions. Steep yourself in the delirious pleasures. When you wake you will have forgotten the Tripos and in the fulness of time will develop into a %% -----File: 012.png---Folio viii------- financial wreck, but in possession of the memory of that heaven-sent dream you will be a far happier and richer man than the millionest millionaire. To pass to earth---from the few papers I have published it will be evident that the subject treated of here is one I have very much at heart, and I think that the publication of the essay is likely to conduce to an acceptance of the view that \emph{it is now the duty of mathematical physicists to study Quaternions seriously}. I have been told by more than one of the few who have read some of my papers that they prove rather stiff reading. The reasons for this are not in the papers I believe but in matters which have already been indicated. Now the present essay reproduces the order in which the subject was developed in my own mind. The less complete treatment of a subject, especially if more diffuse, is often easier to follow than the finished product. It is therefore probable that the present essay is likely to prove more easy reading than my other papers. Moreover I wish it to be studied by a class of readers who are not in the habit of consulting the proceedings,~\&c., of learned societies. I want the slaves of examination to be arrested and to read, for it is apparently to the rising generation that we must look to wipe off the blot from the escutcheon of Cambridge. And now as to the essay itself. But one real alteration has been made. A passage has been suppressed in which were made uncomplimentary remarks concerning a certain author for what the writer regards as his abuse of Quaternion methods. The author in question would no doubt have been perfectly well able to take care of himself, so that perhaps there was no very good reason for suppressing the passage as it still represents my convictions, but I did not want a side issue to be raised that would serve to distract attention from the main one. To bring the notation into harmony with my later papers $d\nu$ and~$\nabla'$ which occur in the manuscript have been changed throughout to $d\Sigma$ and~$\Delta$ respectively. To facilitate printing the solidus has been freely introduced and the vinculum abjured. Mere slips of the pen have been corrected. A formal prefatory note required by the conditions of competition has been omitted. The Table of Contents was not prefixed to the original essay. It consists of little %% -----File: 013.png---Folio ix------- more than a collection of the headings scattered through the essay. Several notes have been added, all indicated by square brackets and the date (1892 or~1893). Otherwise the essay remains absolutely unaltered. The name originally given to the essay is at the head of p.~1 below. The name on the title-page is adopted to prevent confusion of the essay with the `Phil.~Mag.', % [** PP: Changing ``caligraphy'' to ``calligraphy''] paper referred to above. What in the peculiar calligraphy of the manuscript was meant for the familiar $\smalliiint()\,d\varsigma$ has been consistently rendered by the printer as $\smalliiint()\,\ds$. As the mental operation of substituting the former for the latter is not laborious I have not thought it necessary to make the requisite extensive alterations in the proofs. \bigskip I wish here to express my great indebtedness to Prof.~Tait, not only for having through his published works given me such knowledge of Quaternions as I possess but for giving me private encouragement at a time I sorely needed it. There was a time when I felt tempted to throw my convictions to the winds and follow the line of least resistance. To break down the solid and well-nigh universal scepticism as to the utility of Quaternions in Physics seemed too much like casting one's pearls---at least like crying in the wilderness. But though I recognise that I am fighting under Prof.~Tait's banner, yet, as every subaltern could have conducted a campaign better than his general, so in some details I feel compelled to differ from Professor Tait. Some two or three years ago he was good enough to read the present essay. He somewhat severely criticised certain points but did not convince me on all. Among other things he pointed out that I sprung on the unsuspicious reader without due warning and explanation what may be considered as a peculiarity in symbolisation. I take this opportunity therefore of remedying the omission. In Quaternions on account of the non-commutative nature of multiplication we have not the same unlimited choice of order of the terms in a product as we have in ordinary algebra, and the same is true of certain quaternion operators. It is thus inconvenient in many cases to use the familiar method of indicating the connection between an operator and its operand by placing the former %% -----File: 014.png---Folio x------- immediately before the latter. Another method is adopted. With this other method the operator may be separated from the operand, but it seems that there has been a tacit convention among users of this method that the separated operator is still to be restricted to precedence of the operand. There is of course nothing in the nature of things why this should be so, though its violation may seem a trifle strange at first, just as the tyro in Latin is puzzled by the unexpected corners of a sentence in which adjectives (operators) and their nouns (operands) turn up. Indeed a Roman may be said to have anticipated in every detail the method of indicating the connection now under discussion, for he did so by the similarity of the suffixes of his operators and operands. In this essay his example is followed and therefore no restrictions except such as result from the genius of the language (the laws of Quaternions) are placed on the relative positions in a product of operators and operands. With this warning the reader ought to find no difficulty. One of Prof.~Tait's criticisms already alluded to appears in the third edition of his `Quaternions.' The process held up in \S~500 of this edition as an example of ``how not to do it'' is contained in \Secref{6} below and was first given in the `Mess.\ of Math.,' 1884. He implies that the process is a ``most intensely artificial application of'' Quaternions. If this were true I should consider it a perfectly legitimate criticism, but I hold that it is the exact reverse of the truth. In the course of Physical investigations certain volume integrals are found to be capable of, or by general considerations are obviously capable of transformation into surface integrals. We are led to seek for the correct expression in the latter form. Starting from this we can by a long, and in my opinion, tedious process arrive at the most general type of volume integral which is capable of transformation into a surface integral. [I may remark in passing that Prof.~Tait did not however arrive at quite the most general type.] Does it follow that this is the most natural course of procedure? Certainly not, as I think. It would be the most natural course for the empiricist, but not for the scientist. When he has been introduced to one or two volume integrals capable of the transformation the \emph{natural} course of the mathematician is to ask himself what is the most general volume %% -----File: 015.png---Folio xi------- integral of the kind. By quite elementary considerations he sees that while only such volume integrals as satisfy certain conditions are transformable into surface integrals, yet \emph{any} surface integral which is continuous and applies to the complete boundary of any finite volume can be expressed as a volume integral throughout that volume. He is thus led to start from the surface integral and deduces by the briefest of processes the most general volume integral of the type required. Needless to say, when giving his demonstration he does not bare his soul in this way. He thinks rightly that any mathematician can at once divine the exact road he has followed. Where is the artificiality? Let me in conclusion say that even now I scarcely dare state what I believe to be the proper place of Quaternions in a Physical education, for fear my statements be regarded as the uninspired babblings of a misdirected enthusiast, but I cannot refrain from saying that I look forward to the time when Quaternions will appear in every Physical text-book that assumes the knowledge of (say) elementary plane trigonometry. I am much indebted to Mr~G.~H.~A.~Wilson of Clare College, Cambridge, for helping me in the revision of the proofs, and take this opportunity of thanking him for the time and trouble he has devoted to the work. \bigskip \null\hfill \textsc{ALEX. \McAulay.}\qquad \bigskip \textsc{\footnotesize University of Tasmania,} \qquad\textsc{\footnotesize Hobart.} \qquad\qquad{\footnotesize \emph{March}~26, 1893.} \cleardoublepage %% -----File: 016.png---Folio xii------- %[Blank Page] %% -----File: 017.png---Folio xiii------- % [*** Page number on first ToC page] \pagestyle{fancy} \fancyhead{} \fancyfoot{} \renewcommand{\headrulewidth}{0pt} \fancyhead[C]{{\scshape\small contents.}} \ifthenelse{\boolean{ForPrinting}} {\fancyhead[RO,LE]{\thepage}} {\fancyhead[R]{\thepage}} \pdfbookmark[0]{Table of Contents}{Table of Contents} \thispagestyle{empty} \tableofcontents \cleardoublepage \iffalse %% Scanned text from book's ToC CONTENTS. /* SECTION I. INTRODUCTION. PAGE General remarks on the place of Quaternions in Physics 1 Cartesian form of some of the results to follow 6 SECTION II. QUATERNION THEOREMS. 1. Definitions 12 2. Properties of $\zeta$ 15 5. Fundamental property of~$\Dop$ 18 6. Theorems in Integration 19 9. Potentials 22 SECTION III. ELASTIC SOLIDS. 11. Brief recapitulation of previous work in this branch 25 12. Strain, stress-force, stress-couple 26 14. Stress in terms of strain 27 16. The equations of equilibrium 32 16a. Variation of temperature 35 17. Small strains 37 20. Isotropic bodies 40 22. Particular integral of the equation of equilibrium 42 24. Orthogonal coordinates 44 27. Saint-Venant's torsion problem 46 29. Wires 49 SECTION IV. ELECTRICITY AND MAGNETISM. 34. ELECTROSTATICS---general problem 55 41. The force in particular cases 63 43. Nature of the stress 65 46. MAGNETISM---magnetic potential, force, induction 67 49. Magnetic solenoids and shells 69 54. ELECTRO-MAGNETISM---general theory 71 60. Electro-magnetic stress 74 */ %% -----File: 018.png---Folio xiv------- /* SECTION V. HYDRODYNAMICS. PAGE 61. Preliminary 76 62. Notation 76 63. Euler's equations 77 68. The Lagrangian equations 80 69. Cauchy's integrals of these equations 81 71. Flow, circulation, vortex-motion 82 74. Irrotational motion 84 76. Motion of a solid through a liquid 85 79. The velocity in terms of the convergences and spins 88 83. Viscosity 91 SECTION VI. THE VORTEX-ATOM THEORY. 85. Preliminary 94 86. Statement of Sir Wm. Thomson's and Prof. Hicks's theories 94 87. General considerations concerning these theories 95 88. Description of the method here adopted 96 89. Acceleration in terms of the convergences, their time-fluxes and the spins 97 91. Sir Wm. Thomson's theory 98 93. Prof. Hicks's theory 100 94. Consideration of all the terms except $-\nabla(\sigma^2)/2$ 101 96. Consideration of the term $-\nabla\centerdot(\sigma^2)/2$ 103 */ \fi %% -----File: 019.png---Folio 1------- \mainmatter \pagenumbering{arabic} \HeadChapter{Section I}{Introduction} {quaternions as a practical instrument \\ of physical research} \Tocline{}{General remarks on the place of Quaternions in Physics} It is a curious phenomenon in the History of Mathematics that the greatest work of the greatest Mathematician of the century which prides itself upon being the most enlightened the world has yet seen, has suffered the most chilling neglect. The cause of this is not at first sight obvious. We have here little to do with the benefit provided by Quaternions to Pure Mathematics. The reason for the neglect here \emph{may} be that Hamilton himself has developed the Science to such an extent as to make successors an impossibility. One cannot however resist a strong suspicion that were the subject even studied we should hear more from Pure Mathematicians, of Hamilton's valuable results. This reason at any rate cannot be assigned for the neglect of the Physical side of Quaternions. Hamilton has done but little in this field, and yet when we ask what Mathematical Physicists have been tempted by the bait to win easy laurels (to put the incentive on no higher grounds), the answer must be scarcely one. Prof.~Tait is the grand exception to this. But well-known Physicist though he be, his fellow-workers for the most part render themselves incapable of appreciating his valuable services by studying the subject if at all only as dilettanti. The number who read a small amount in Quaternions is by no means small, but those who get further than what is recommended by Maxwell as imperatively necessary are but a small percentage of the whole. I cannot help thinking that this state of affairs is owing chiefly to a prejudice. This prejudice is well seen in Maxwell's well-known statement---``I am convinced that the introduction of the %% -----File: 020.png---Folio 2------- ideas, \emph{as distinguished from the operations and methods} of Quaternions, will be of great use to us in all parts of our subject.''\footnote{\emph{Elect.\ and~Mag.} Vol.~I. \S~10.} Now what I hold and what the main object of this essay is to prove is that the ``operations and methods'' of Quaternions are as much better qualified to deal with Physics than the ordinary ones as are the ``ideas''. But, what has produced this notion, that the subject of Quaternions is only a pretty toy that has nothing to do with the serious work of practical Physics? It must be the fact that it has hitherto produced few results that appeal strongly to Physicists. This I acknowledge, but that the deduction is correct I strongly disbelieve. As well might an instrument of which nobody has attempted to master the principles be blamed for not being of much use. Workers naturally find themselves while still inexperienced in the use of Quaternions incapable of clearly thinking through them and of making them do the work of Cartesian Geometry, and they conclude that Quaternions do not provide suitable treatment for what they have in hand. The fact is that the subject requires a slight development in order readily to apply to the practical consideration of most physical subjects. The first steps of this, which consist chiefly in the invention of new symbols of operation and a slight examination of their chief properties, I have endeavoured to give in the following pages. I may now state what I hold to be the mission of Quaternions to Physics. I believe that Physics would advance with both more rapid and surer strides were Quaternions introduced to serious study to the almost total exclusion of Cartesian Geometry, except in an insignificant way as a particular case of the former. All the geometrical processes occurring in Physical \emph{theories} and \emph{general} Physical problems are much more graceful in their Quaternion than in their Cartesian garb. To illustrate what is here meant by ``theory'' and ``general problem'' let us take the case of Elasticity treated below. That by the methods advocated not only are the already well-known results of the general \emph{theory} of Elasticity better proved, but more general results are obtained, will I think be acknowledged after a perusal of \Secref{12} to~\Secref{21} below. That Quaternions are superior to Cartesian Geometry in considering the \emph{general problems} of (1)~an infinite isotropic solid, (2)~the torsion and bending of prisms and cylinders, (3)~the general theory of wires, I have endeavoured to shew in %% -----File: 021.png---Folio 3------- \Secref{22}--\Secref{33}. But for \emph{particular} problems such as the torsion problem for a cylinder of given shape, we require of course the various theories specially constructed for the solution of particular problems such as Fourier's theories, complex variables, spherical harmonics,~\&c. It will thus be seen that I do not propose to banish these theories but merely Cartesian Geometry. So mistaken are the common notions concerning the pretensions of advocates of Quaternions that I was asked by one well-known Mathematician whether Quaternions furnished methods for the solution of differential equations, as he asserted that this was all that remained for Mathematics in the domain of Physics! Quaternions can no more solve differential equations than Cartesian Geometry, but the solution of such equations can be performed as readily, in fact generally more so, in the Quaternion shape as in the Cartesian. But that the sole work of Physical Mathematics to-day is the solution of differential equations I beg to question. There are many and important Physical questions and extensions of Physical theories that have little or nothing to do with such solutions. As witness I may call attention to the new Physical work which occurs below. If only on account of the extreme simplicity of Quaternion notation, large advances in the parts of Physics now indicated, are to be expected. Expressions which are far too cumbrous to be of much use in the Cartesian shape become so simple when translated into Quaternions, that they admit of easy interpretation and, what is perhaps of more importance, of easy manipulation. Compare for instance the particular case of \equationref{III}{15$m$} \Secref{16} below when $\mathbf{F} = 0$ with the same thing as considered in Thomson and Tait's \textit{Nat.~Phil.}, App.~C. The Quaternion equation is \[ \rho'_1 S \nabla_{1\Psi}\Dop w\Delta = 0. \] The Cartesian \emph{exact} equivalent consists of Thomson and Tait's equations~(7), viz. \begin{align*} \frac{d}{dx} &\left\{ 2\frac{dw}{dA} \left( \frac{d\alpha}{dx} + 1 \right) + \frac{dw}{db} \frac{d\alpha}{dz} + \frac{dw}{dc} \frac{d\alpha}{dy} \right\} \\ % &{} + \frac{d}{dy} \left\{ 2\frac{dw}{dB} \frac{d\alpha}{dy} + \frac{dw}{da} \frac{d\alpha}{dz} + \frac{dw}{dc} \left( \frac{d\alpha}{dx} + 1 \right) \right\} \\ % &{}+ \frac{d}{dz} \left\{ 2\frac{dw}{dC} \frac{d\alpha}{dz} + \frac{dw}{da} \frac{d\alpha}{dy} + \frac{dw}{db} \left( \frac{d\alpha}{dx} + 1 \right) \right\} = 0, \end{align*} and two similar equations. %% -----File: 022.png---Folio 4------- Many of the equations indeed in the part of the essay where this occurs, although quite simple enough to be thoroughly useful in their present form, lead to much more complicated equations than those just given when translated into Cartesian notation. It will thus be seen that there are two statements to make good:---(1) that Quaternions are in such a stage of development as already to justify the practically complete banishment of Cartesian Geometry from Physical questions of a general nature, and (2) that Quaternions will in Physics produce many \emph{new} results that cannot be produced by the rival and older theory. To establish completely the first of these propositions it would be necessary to go over \emph{all} the ground covered by Mathematical Physical Theories, by means of our present subject, and compare the proofs with the ordinary ones. This of course is impossible in an essay. It would require a treatise of no small dimensions. But the principle can be followed to a small extent. I have therefore taken three typical theories and applied Quaternions to most of the general propositions in them. The three subjects are those of Elastic Solids, with the thermodynamic considerations necessary, Electricity and Magnetism, and Hydrodynamics. It is impossible without greatly exceeding due limits of space to consider in addition, Conduction of Heat, Acoustics, Physical Optics, and the Kinetic Theory of Gases. With the exception of the first of these subjects I do not profess even to have attempted hitherto the desired applications, but one would seem almost justified in arguing that, since Quaternions have been found so applicable to the subjects considered, they are very likely to prove useful to about the same extent in similar theories. Again, only in one of the subjects chosen, viz., Hydrodynamics, have I given the \emph{whole} of the general theory which usually appears in text-books. For instance, in Electricity and Magnetism I have not considered Electric Conduction in three dimensions which, as Maxwell remarks, lends itself very readily to Quaternion treatment, nor Magnetic Induction, nor the Electro-Magnetic Theory of Light. Again, I have left out the consideration of Poynting's theories of Electricity which are very beautifully treated by Quaternions, and I felt much tempted to introduce some considerations in connection with the Molecular Current theory of Magnetism. With similar reluctance I have been compelled to omit many applications in the Theory of Elastic Solids, but the already too large size of the essay admitted of no other course. Notwithstanding these omissions, %% -----File: 023.png---Folio 5------- I think that what I have done in this part will go far to bear out the truth of the first proposition I have stated above. But it is the second that I would especially lay stress upon. In the first it is merely stated that Cartesian Geometry is an antiquated machine that ought to be thrown aside to make room for modern improvements. But the second asserts that the improved machinery will not only do the work of the old better, but will also do much work that the old is quite incapable of doing at all. Should this be satisfactorily established and should Physicists in that case still refuse to have anything to do with Quaternions, they would place themselves in the position of the traditional workmen who so strongly objected to the introduction of machinery to supplant manual labour. But in a few months and synchronously with the work I have already described, to arrive at a \emph{large} number of new results is too much to expect even from such a subject as that now under discussion. There are however some few such results to shew. I have endeavoured to advance each of the theories chosen in at least one direction. In the subject of Elastic Solids I have expressed the stress in terms of the strain in the most general case, i.e.\ where the strain is not small, where the ordinary assumption of no stress-couple is not made and where no assumption is made as to homogeneity, isotropy,~\&c. I have also obtained the equations of motion when there is given an external force and couple per unit volume of the unstrained solid. These two problems, as will be seen, are by no means identical. In Electrostatics I have considered the most general mechanical results flowing from Maxwell's theory, and their explanation by stress in the dielectric. These results are not known, as might be inferred from this mode of statement, for to solve the problem we require to know forty-two independent constants to express the properties of the dielectric at a given state of strain at each point. These are the six coefficients of specific inductive capacity and their thirty-six differential coefficients with regard to the six coordinates of pure strain. But, as far as I am aware, only such particular cases of this have already been considered as make the forty-two constants reduce at most to three. In Hydrodynamics I have endeavoured to deduce certain general phenomena which would be exhibited by vortex-atoms acting upon one another. This has been done by examination of an equation which has not, I believe, been hitherto given. The result of this part of the essay is to lead to a presumption against %% -----File: 024.png---Folio 6------- Sir William Thomson's Vortex-Atom Theory and in favour of Hicks's. As one of the objects of this introduction is to give a bird's-eye view of the merits of Quaternions as opposed to Cartesian Geometry, it will not be out of place to give side by side the Quaternion and the Cartesian forms of most of the new results I have been speaking about. It must be premised, as already hinted, that the usefulness of these results must be judged not by the Cartesian but by the Quaternion form. \TCSection{}{Cartesian form of some of the results to follow}{Elasticity} Let the point $(x, y, z)$ of an elastic solid be displaced to $(x', y', z')$. The strain at any point that is caused may be supposed due to a pure strain followed by a rotation. In Section III. below, this pure strain is called $\psi$. Let its coordinates be $e$, $f$, $g$, $a/2$, $b/2$, $c/2$; i.e.\ if the vector $(\xi, \eta, \zeta)$ becomes $(\xi', \eta', \zeta')$ by means of the pure strain, then \begin{gather*} \xi' = e\xi + \tfrac{1}{2} c\eta + \tfrac{1}{2} b\zeta, \\ \text{ \&c., \&c.} \end{gather*} Thus when the strain is small $e$, $f$, $g$ reduce to Thomson and Tait's $1+e$, $1+f$, $1+g$ and $a$, $b$, $c$ are the same both in their case and the present one. Now let the coordinates of $\Psi$, \Secref{16} below, be $E$, $F$, $G$, $A/2$, $B/2$, $C/2$. \Equationref{III}{15}, \Secref{16} below, viz. \[ \footnotemark\Psi\omega = \psi^2 \omega = \chi'\chi\omega = \nabla_1 S \rho'_1 \rho'_2 S\omega\nabla_2, \] gives in our present notation \Erratum{err6a}%[** PP: Break second equation at equals sign] \begin{align*} E &= e^2 + c^2/4 + b^2/4 = (dx'/dx)^2 + (dy'/dx)^2 + (dz'/dx)^2, \\ & \text{\qquad\qquad\qquad\qquad\qquad\&c., \&c.} \\ % A &= a(f+g) + bc/2 \\ % [** PP: Re-breaking] &= 2\bigl\{(dx'/dy)(dx'/dz) + (dy'/dy)(dy'/dz) + (dz'/dy)(dz'/dz)\bigr\}, \\ & \text{\qquad\qquad\qquad\qquad\qquad\&c., \&c.} \end{align*} which shew that the present $E$, $F$, $G$, $A/2$, $B/2$, $C/2$ are the $A$, $B$, $C$, $a$, $b$, $c$ of Thomson and Tait's \textit{Nat.\ Phil.}, App.~C. \footnotetext{This result is one of Tait's (\textit{Quaternions} \S~365 where he has $\phi'\phi = \overline{\varpi}^2$). It is given here for completeness.} % end of footnote %% -----File: 025.png---Folio 7------- Let us put \begin{align*} J \begin{pmatrix} x' y' z' \\ x\, y\, z \end{pmatrix} &= J \\ % J \begin{pmatrix} y' z' \\ y\, z \end{pmatrix}\phantom{z} &=J_{11},\quad \text{\&c., \&c.,} \\ % J \begin{pmatrix} z' x' \\ y\, z \end{pmatrix}\phantom{z} &= J_{12}, J \begin{pmatrix} y' z' \\ z\, x \end{pmatrix} = J_{21},\quad \text{\&c., \&c., \&c., \&c.} \end{align*} I have shewn in \Secref{14} below that the \emph{stress-couple is quite independent of the strain}. Thus we may consider the stress to consist of two parts---an ordinary stress $PQRSTU$ as in Thomson and Tait's \emph{Nat.~Phil.} and a stress which causes a couple per unit volume $L' M' N'$. The former only of these will depend on strain. The result of the two will be to cause a force (as indeed can be seen from the expressions in \Secref{13} below) per unit area on the $x$-interface $P$, $U+ N'/2$, $T-M'/2$, and so for the other interfaces. If $L$, $M$, $N$ be the external couple per unit volume of the \emph{unstrained} solid we shall have \[ L' = -L/J, \quad M' = -M/J, \quad N' = -N/J, \] for the external couple and the stress-couple are \emph{always} equal and opposite. Thus the force on the $x$-interface becomes \[ P, \quad U-N/2J, \quad T+M/2J \] and similarly for the other interfaces. To express the part of the stress ($P$~\&c.)\ which depends on the strain in terms of that strain, consider $w$ the potential energy per unit volume of the unstrained solid as a function of $E$~\&c. In the general thermodynamic case $w$ may be defined by saying that \Erratum{err7a}% [** PP: Displaying this equation for readability] \begin{align*} w &\times (\text{\small the element of volume}) \\ &= (\text{the intrinsic energy of the element}) \\ &\quad- (\text{\small the entropy of the element} \times \text{\small its absolute temperature} \times \text{\small Joule's coefficient}). \end{align*} Of course~$w$ may be, and indeed is in \Secref{14}, \Secref{15} below, regarded as a function of~$e$~\&c. The equation for stress is~\Eref{III}{15$b$} \Secref{16} below, viz., \[ J\overline{\phi}\omega = 2\chi \Pre{\Psi}{\Dop} w\chi'\omega = 2\rho_1' S \rho_2'\omega S\nabla_{1\Psi}\Dop w\nabla_2. \] The second of the expressions is in terms of the strain and the third in terms of the displacement and its derivatives. In our present notation this last is %% -----File: 026.png---Folio 8------- \begin{align*} \frac{JP}{2} &= \left(\frac{dx'}{dx}\right)^2 \frac{dw}{dE} + \left(\frac{dx'}{dy}\right)^2 \frac{dw}{dF} + \left(\frac{dx'}{dz}\right)^2 \frac{dw}{dG} \\[1ex] & \quad + 2\frac{dx'}{dy} \frac{dx'}{dz} \frac{dw}{dA} + 2\frac{dx'}{dz} \frac{dx'}{dx} \frac{dw}{dB} + 2\frac{dx'}{dx} \frac{dx'}{dy} \frac{dw}{dC}, \\ &\qquad\qquad\qquad\qquad\qquad\text{\&c., \&c.} \\ \frac{JS}{2} &= \frac{dy'}{dx} \frac{dz'}{dx} \frac{dw}{dE} + \frac{dy'}{dy} \frac{dz'}{dy} \frac{dw}{dF} + \frac{dy'}{dz} \frac{dz'}{dz} \frac{dw}{dG} \\[1ex] & \quad + \left(\frac{dy'}{dy} \frac{dz'}{dz} + \frac{dy'}{dz} \frac{dz'}{dy}\right)\frac{dw}{dA} + \left(\frac{dy'}{dz} \frac{dz'}{dx} + \frac{dy'}{dx} \frac{dz'}{dz}\right)\frac{dw}{dB} \\[1ex] & \quad % skip right by width of first summand... \phantom{+ \left(\frac{dy'}{dx} \frac{dz'}{dy} + \frac{dy'}{dy} \frac{dz'}{dx}\right)\frac{dw}{dC}} % plus a bit \qquad + \left(\frac{dy'}{dx} \frac{dz'}{dy} + \frac{dy'}{dy} \frac{dz'}{dx}\right)\frac{dw}{dC}, \\ &\qquad\qquad\qquad\qquad\qquad\text{\&c., \&c.} \end{align*} In \Secref{14} I also obtain this part of the stress explicitly in terms of $e$, $f$, $g$, $a$, $b$, $c$, of $w$ as a function of these quantities and of the axis and amount of rotation. But these results are so very complicated in their Cartesian shape that it is quite useless to give them. To put down the equations of motion let $X_x$, $Y_x$, $Z_x$ be the force due to stress on what before strain was unit area perpendicular to the axis of $x$. Similarly for $X_y$, \&c. Next suppose that $X$, $Y$, $Z$ is the external force per unit volume of the unstrained solid and let $D$ be the original density of the solid. Then the equation of motion~\Eref{III}{15$n$} \Secref{16$a$} below, viz.\ \[ D\ddot{\rho}' = \mathbf{F} + \tau\Delta, \] gives in our present notation \[ X + dX_x/dx + dX_y/dy + dX_z/dz = \ddot{x}'D,\quad \text{\&c., \&c.} \] It remains to express $X_x$~\&c.\ in terms of the displacement and~$LMN$. This is done in \equationref{III}{15$l$} \Secref{16} below, viz.\ \[ \tau\omega = -2\rho_1' S \nabla_{1\Psi}\Dop w \omega + 3V\mathbf{M} V\rho_1' \rho_2' S\omega\nabla_1 \nabla_2 /2 S\nabla_1 \nabla_2 \nabla_3 S\rho_1'\rho_2'\rho_3'.\footnotemark \] \footnotetext{The second term on the right contains \emph{in full} the nine terms corresponding to $(J_{12}N-J_{13}M)/2J$. Quaternion notation is therefore here, as in nearly all cases which occur in Physics, considerably more compact even than the notations of determinants or Jacobians.} In our present notation this consists of the following nine equations: %% -----File: 027.png---Folio 9------- \begin{align*} X_x &= 2\left( \frac{dw}{dE} \frac{dx'}{dx} + \frac{dw}{dC} \frac{dx'}{dy} + \frac{dw}{dB} \frac{dx'}{dz}\right) + \frac{J_{12}N - J_{13}M}{2J}, \\ % Y_x &= 2\left( \frac{dw}{dE} \frac{dy'}{dx} + \frac{dw}{dC} \frac{dy'}{dy} + \frac{dw}{dB} \frac{dy'}{dz}\right) + \frac{J_{13}L - J_{11}N}{2J}, \\ % Z_x &= 2\left( \frac{dw}{dE} \frac{dz'}{dx} + \frac{dw}{dC} \frac{dz'}{dy} + \frac{dw}{dB} \frac{dz'}{dz}\right) + \frac{J_{11}M - J_{12}L}{2J}, \\ \end{align*} and six similar equations. We thus see that in the case where $LMN$ are zero, our present $X_x$, $X_y$, $X_z$ are the $PQR$ of Thomson and Tait's \emph{Nat.~Phil}.\ App.~C (d), and therefore equations~(7) of that article agree with our equations of motion when we put both the external force and the acceleration zero. These are some of the new results in Elasticity, but, as I have hinted, there are others in \Secref{14}, \Secref{15} which it would be waste of time to give in their Cartesian form. \SectionTitle{Electricity} In Section~IV. below I have considered, as already stated, the most general mechanical results flowing from Maxwell's theory of Electrostatics. I have shewn that here, as in the particular cases considered by others, the forces, whether per unit volume or per unit surface, can be explained by a stress in the dielectric. It is easiest to describe these forces by means of the stress. Let the coordinates of the stress be $PQRSTU$. Then $F_1 F_2 F_3$ the mechanical force, due to the field per unit volume, exerted upon the dielectric where there is no discontinuity in the stress, is given by \[ F_1 = dP/dx + dU/dy + dT/dz,\quad \text{\&c., \&c.} \] and $(l, m, n)$ being the direction cosines of the normal to any surface, pointing away from the region considered \[ F_1' = - [lP + mU + nT]_a - [\ ]_b,\quad \text{\&c., \&c.,} \] where $a$, $b$ indicate the two sides of the surface and ${F_1}'$, ${F_2}'$, ${F_3}'$ is the force due to the field per unit surface. It remains to find $P$~\&c. Let $X$, $Y$, $Z$ be the electro-motive force, $\alpha$, $\beta$, $\gamma$ the displacement, $w$ the potential energy per unit volume and $K_{xx}$, $K_{yy}$, $K_{zz}$, $K_{yz}$, $K_{zx}$, $K_{xy}$ the coefficients of specific %% -----File: 028.png---Folio 10------- inductive capacity. Let $1+e$, $1+f$, $1+g$, $a/2$, $b/2$, $c/2$ denote the pure part of the strain of the medium. The $K$'s will then be functions of $e$~\&c.\ and we must suppose these functions known, or at any rate we must assume the knowledge of both the values of the $K$'s and their differential coefficients at the particular state of strain in which the medium is when under consideration. The relations between the above quantities are \begin{gather*} 4\pi\alpha = K_{xx}X + K_{xy}Y + K_{zx}Z,\quad \text{\&c., \&c.} \\ \begin{aligned} w &= (X\alpha + Y\beta + Z\gamma)/2 \\ &= (K_{xx} X^2 + K_{yy} Y^2 + K_{zz} Z^2 + 2K_{yz} YZ + 2K_{zx} ZX + 2K_{xy} XY)/8\pi. \end{aligned} \end{gather*} It is the second of these expressions for~$w$ which is assumed below, and the differentiations of course refer only to the~$K$'s. The equation expressing $P$~\&c.\ in terms of the field is~\Eref{IV}{21} \Secref{40} below, viz.\ \[ \phi\omega = -\tfrac{1}{2} V\mathbf{D}\omega\mathbf{E} - \Pre{\Psi}{\Dop} w\,\omega, \] which in our present notation gives the following six equations \begin{align*} P &= -\tfrac{1}{2}(-\alpha X + \beta Y + \gamma Z) - dw/de,\quad \text{\&c., \&c.}, \\ S &= \tfrac{1}{2} (\beta Z + \gamma Y) - dw/da,\quad \text{\&c., \&c.} \end{align*} I have shewn in \Secref{41}--\Secref{45} below that these results agree with particular results obtained by others. \SectionTitle{Hydrodynamics} The new work in this subject is given in Section~VI.---``The Vortex-Atom Theory.'' It is quite unnecessary to translate the various expressions there used into the Cartesian form. I give here only the principal equation in its two chief forms, \equationref{VI}{9} \Secref{89} and \equationref{VI}{11} \Secref{90}, viz.\ \begin{align*} & P+v - \sigma^2/2 + (4\pi)^{-1} \smalliiint (S\sigma\tau \nabla u + u\partial m/\partial t)\, \ds = H, \\ & P+v - \sigma^2/2 + (4\pi)^{-1} \smalliiint \{\ds S \nabla u (V\sigma\tau - m\sigma) + ud(m\ds)/dt\} = H. \end{align*} In Cartesian notation these are \begin{align*} &\int dp/\rho + V + q^2/2 \\ &\qquad - (4\pi)^{-1} \smalliiint \bigl\{2[(x'-x)(w\eta - v\zeta) + \cdots + \cdots]/r^3 \\ &\phantom{\qquad - (4\pi)^{-1} \smalliiint} + (\partial c/\partial t)/r \bigr\}\,dx'\,dy'\,dz' = H.\\ &\int dp/\rho + V + q^2/2 \\ &\qquad - (4\pi)^{-1} \smalliiint \bigl\{(x'-x)[2(w\eta - v\zeta) - cu] + \cdots + \cdots \bigr\}/r^3. \,dx'\,dy'\,dz' \\ &\qquad - (4\pi)^{-1} \smalliiint \bigl\{d(cdx'\,dy'\,dz')/dt\bigr\}/r = H. \end{align*} %% -----File: 029.png---Folio 11------- The fluid here considered is one whose motion is continuous from point to point and which extends to infinity. The volume integral extends throughout space. The notation is as usual. It is only necessary to say that~$H$ is a function of the time only, $r$~is the distance between the points $x', y', z'$ and~$x, y, z$; \[ c = du/dx + dv/dy + dw/dz; \] $d/dt$ is put for differentiation which follows a particle of the fluid, and $\partial/\partial t$ for that which refers to a fixed point. The explanation of the unusual length of this essay, which I feel is called for, is contained in the foregoing description of its objects. If the objects be justifiable, so must also be the length which is a necessary outcome of those objects. %% -----File: 030.png---Folio 12------- \Chapter{Section II}{Quaternion Theorems} \Section{1}{Definitions} \Paragraph{1} As there are two or three symbols and terms which will be in constant use in the following pages that are new or more general in their signification than is usual, it is necessary to be perhaps somewhat tediously minute in a few preliminary definitions and explanations. A \emph{function} of a variable in the following essay is to be understood to mean anything which depends on the variable and which can be subjected to mathematical operations; the variable itself being anything capable of being represented by a mathematical symbol. In Cartesian Geometry the variable is generally a single scalar. In Quaternions on the other hand a general quaternion variable is not infrequent, a variable which requires 4 scalars for its specification, and similarly for the function. In both, however, either the variable or the function may be a mere symbol of operation. In the following essay we shall frequently have to speak of variables and functions which are neither quaternions nor mere symbols of operation. For instance~$K$ in \Secref{40} below requires 6~scalars to specify it, and it is a function of~$\psi$ which requires 6~scalars and~$\rho$ which requires 3~scalars. When in future the expression ``any function'' is used it is always to be understood in the general sense just explained. We shall frequently have to deal with functions of many independent vectors, and especially with functions which are linear in each of the constituent vectors. These functions merely require to be noticed but not defined. Hamilton has defined the meaning of the symbolic vector $\nabla$ thus:--- \[ \nabla = i \frac{d}{dx} + j\frac{d}{dy} + k\frac{d}{dz}, \] %% -----File: 031.png---Folio 13------- where $i$, $j$, $k$ are unit vectors in the directions of the mutually perpendicular axes $x$, $y$, $z$. I have found it necessary somewhat to expand the meaning of this symbol. When a numerical suffix $1$, $2$, \ldots\ is attached to a~$\nabla$ in any expression it is to indicate that the differentiations implied in the~$\nabla$ are to refer to and only to other symbols in the same expression which have the same suffix. After the implied differentiations have been performed the suffixes are of course removed. Thus $Q(\alpha, \beta, \gamma, \delta)$ being a quaternion function of any four vectors $\alpha$, $\beta$, $\gamma$, $\delta$, linear in each \begin{align*} Q(\lambda_1 \mu_2 \nabla_1 \nabla_2) & \equiv Q\left( \frac{d\lambda}{dx} \frac{d\mu}{dx} ii \right) + Q\left( \frac{d\lambda}{dy} \frac{d\mu}{dx} ji \right) + Q\left( \frac{d\lambda}{dz} \frac{d\mu}{dx} ki \right) \\ % &+ Q\left( \frac{d\lambda}{dx} \frac{d\mu}{dy} ij \right) + Q\left( \frac{d\lambda}{dy} \frac{d\mu}{dy} jj \right) + Q\left( \frac{d\lambda}{dz} \frac{d\mu}{dy} kj \right) \\ % &+ Q\left( \frac{d\lambda}{dx} \frac{d\mu}{dz} ik \right) + Q\left( \frac{d\lambda}{dy} \frac{d\mu}{dz} jk \right) + Q\left( \frac{d\lambda}{dz} \frac{d\mu}{dz} kk \right) \end{align*} and again \[ Q_1(\lambda_1, \mu_2, \nabla_1, \nabla_2) \equiv Q_3(\lambda_1, \mu_2, \nabla_1 + \nabla_3, \nabla_2). \] It is convenient to reserve the symbol~$\Delta$ for a special meaning. It is to be regarded as a particular form of~$\nabla$, but its differentiations are to refer to \emph{all} the variables in the term in which it appears. Thus~$Q$ being as before \[ Q(\lambda_1, \mu, \Delta, \nabla_1) = Q_2(\lambda_1, \mu_2, \nabla_1 + \nabla_2, \nabla_1)\footnotemark. \] \footnotetext{% These meanings for $\nabla_1, \nabla_2\ldots\Delta$ I used in a paper on ``Some General Theorems in Quaternion Integration,'' in the \textit{Mess.\ of Math.}\ Vol.~\textsc{xiv}.\ (1884), p.~26. The investigations there given are for the most part incorporated below. [Note added, 1892, see preface as to the alteration of $\nabla'$ into $\Delta$.]} If in a linear expression or function $\nabla_1$ and~$\rho_1$ ($\rho$ being as usual $\equiv ix +jy + kz$) occur once each they can be interchanged. Similarly for $\nabla_2$ and~$\rho_2$. So often does this occur that I have thought it advisable to use a separate symbol~$\zeta_1$ for each of the two $\nabla_1$ and~$\rho_1$, $\zeta_2$ for each of the two $\nabla_2$ and~$\rho_2$ and so for $\zeta_3$,~\&c. If only one such pair occur there is of course no need for the suffix attached to $\zeta$. Thus $\zeta$ may be looked upon as a symbolic vector or as a single term put down instead of three. For $Q(\alpha, \beta)$ being linear in each of the vectors $\alpha$, $\beta$ \[ \EqnTag{II}{1} Q(\zeta, \zeta) = Q(\nabla_1, \rho_1) = Q(i, i) + Q(j, j) + Q(k, k). \] There is one more extension of the meaning of $\nabla$ to be given. %% -----File: 032.png---Folio 14------- $u$, $v$, $w$ being the rectangular coordinates of any vector $\sigma$, $\Nabla{\sigma}$ is defined by the equation \[ \Nabla{\sigma} = i\frac{d}{du} + j\frac{d}{dv} + k\frac{d}{dw}. \] To $\Nabla{\sigma}$ of course are to be attached, when necessary, the suffixes above explained in connection with~$\nabla$. Moreover just as for $\nabla_1$, $\rho_1$ we may put $\zeta$, $\zeta$ so also for $\Nabla{\sigma}_1$, $\sigma_1$ may we put the same. With these meanings one important result follows at once. \emph{The $\nabla_1$'s, $\nabla_2$'s, \&c.,\ obey all the laws of ordinary vectors whether with regard to multiplication or addition}, for the coordinates $d/dx$, $d/dy$, $d/dz$ of any~$\nabla$ obey with the coordinates of any vector or any other~$\nabla$ all the laws of common algebra. Just as $\Nabla{\sigma}$ may be defined as a symbolic vector whose coordinates are $d/du$, $d/dv$, $d/dw$ so $\phi$ being a linear vector function of any vector whose coordinates are \[ (a_1b_1c_1\; a_2b_2c_2\; a_3b_3c_3)\quad(\text{i.e.\ } \phi i = a_1i + b_1j + c_1k,\quad \text{\&c.}). \] $\Pre{\phi}{\Dop}$\footnote{% I have used an inverted~$D$ to indicate the analogy to Hamilton's inverted~$\Delta$.} is defined as a symbolic linear vector function whose coordinates are \[ (d/da_1,\, d/db_1,\, d/dc_1,\ d/da_2,\, d/db_2,\, d/dc_2,\ d/da_3,\, d/db_3,\, d/dc_3), \] and to $\Pre{\phi}{\Dop}$ is to be applied exactly the same system of suffixes as in the case of~$\nabla$. Thus $q$ being any quaternion function of~$\phi$, and~$\omega$ any vector \begin{align*} \Pre{\phi}{\Dop}_1\omega\centerdot q_1 = & -(idq/da_1 + jdq/db_1 + kdq/dc_1) Si\omega \\ & -(idq/da_2 + jdq/db_2 + kdq/dc_2) Sj\omega \\ & -(idq/da_3 + jdq/db_3 + kdq/dc_3) Sk\omega. \end{align*} The same symbol $\Pre{\phi}{\Dop}$ is used without any inconvenience with a slightly different meaning. If the independent variable~$\phi$ be a \emph{self-conjugate} linear vector function it has only six coordinates. If these are~$PQRSTU$ (i.e.\ $\phi i = Pi + Uj + Tk$,~\&c.)\ $\Pre{\phi}{\Dop}$ is defined as a self-conjugate linear vector function whose coordinates are \[ (d/dP,\, d/dQ,\, d/dR,\, \tfrac{1}{2}d/dS,\, \tfrac{1}{2}d/dT,\, \tfrac{1}{2}d/dU). \] We shall frequently have to compare volume integrals with integrals taken over the bounding surface of the volume, and again surface integrals with integrals taken round the boundary of the surface. For this purpose we shall use the following notations for linear, surface and volume integrals respectively~$\int Q\,d\rho$, %% -----File: 033.png---Folio 15------- $\smalliint Q\,d\Sigma$, $\smalliiint Q\,\ds$ where~$Q$ is any function of the position of a point. Here $d\rho$~is a vector element of the curve, $d\Sigma$ a vector element of the surface, and~$\ds$ an element of volume. When comparisons between line and surface integrals are made we take~$d\Sigma$ in such a direction that~$d\rho$ is in the direction of positive rotation round the element~$d\Sigma$ close to it. When comparisons between surface and volume integrals are made $d\Sigma$ is always taken in the direction \emph{away from} the volume which it bounds. \Section{2}{Properties of \texorpdfstring{$\zeta$}{zeta}} \Paragraph{2} The property of~$\zeta$ on which nearly all its usefulness depends is that if~$\sigma$ be any vector \[ \sigma = -\zeta S\zeta \sigma, \] which is given at once by \equationref{II}{1} of last section. This gives a useful expression for the conjugate of a linear vector function of a vector. Let~$\phi$ be the function and~$\omega$, $\tau$ any two vectors. Then $\phi'$ denoting as usual the conjugate of~$\phi$ we have \[ S\omega\phi\tau = S\tau\phi'\omega, \] whence putting on the left $\tau =-\zeta S\zeta\tau$ we have \[ S\tau(-\zeta S\omega\phi\zeta) = S\tau\phi'\omega, \] or since $\tau$ is quite arbitrary \[ \EqnTag{II}{2} \phi'\omega = -\zeta S\omega\phi\zeta. \] From this we at once deduce expressions for the pure part~$\overline{\phi}\omega$ and the rotational part~$V\epsilon\omega$ of~$\phi\omega$ by putting \[ \EqnTag{II}{3} \left. \begin{aligned} (\phi + \phi')\omega &= -\phi\zeta S\omega\zeta - \zeta S\omega\phi\zeta = 2\overline{\phi}\omega, \\ % (\phi - \phi')\omega &= -\phi\zeta S\omega\zeta + \zeta S\omega\phi\zeta = VV\zeta\phi\zeta.\omega = 2V\epsilon\omega. \end{aligned}\quad % [** PP: Need a bit of space before brace] \right\} \] And all the other well-known relations between $\phi$ and~$\phi'$ are at once given e.g.\ $S\zeta\phi\zeta = S\zeta\phi'\zeta$, i.e.\ the ``convergence'' of~$\phi =$ the ``convergence'' of~$\phi'$. \Paragraph{3} Let $Q (\lambda, \mu)$ be any function of two vectors which is linear in each. Then if~$\phi\omega$ be any linear vector function of a vector~$\omega$ given by \[ \EqnTag{II}{4} \left. \begin{aligned} && &\phi\omega = -\Sigma\beta S\omega\alpha && \\ &\text{we have\qquad\qquad}& & Q(\zeta, \phi\zeta) = \Sigma Q(-\zeta S\zeta\alpha,\beta) = \Sigma Q(\alpha,\beta), && \end{aligned} \right\} \] %% -----File: 034.png---Folio 16------- or more generally \[ \EqnTag{II}{4$a$} Q(\zeta, \phi\chi\zeta) = \Sigma Q(\chi'\alpha, \beta). \] To prove, it is only necessary to observe that \begin{flalign*} && \phi\chi\zeta = -\Sigma \beta S\alpha\chi\zeta &= -\Sigma \beta S\zeta\chi'\alpha, && \\ &\text{and that}& -\zeta S\zeta\chi'\alpha &= \chi'\alpha. && \end{flalign*} As a particular case of \hyperref[eqn:II.4]{eq.~(4)} let~$\phi$ have the self-conjugate value \Erratum{err16a}% [** PP: Added period after first equation] \[ \EqnTag{II}{5} \left. \begin{aligned} && &\phi\omega = -\tfrac{1}{2}\Sigma(\beta S\omega\alpha + \alpha S\omega\beta). && \\ &\text{Then}\qquad& &Q(\zeta, \phi\zeta) = \tfrac{1}{2}\Sigma\{Q(\alpha,\beta) + Q(\beta,\alpha)\}, && \end{aligned}\right\} \] or if $Q(\lambda, \mu)$ is symmetrical in $\lambda$ and~$\mu$ \[ \EqnTag{II}{6} Q(\zeta, \phi\zeta) = \Sigma Q(\alpha, \beta). \] The application we shall frequently make of this is to the case when for~$\alpha$ we put~$\nabla_1$ and for~$\beta$, $\sigma_1$, where~$\sigma$ is any vector function of the position of a point. In this case the first expression for~$\phi$ is the strain function and the second expression the pure strain function resulting from a small displacement~$\sigma$ at every point. As a simple particular case put $Q(\lambda, \mu) = S\lambda\mu$ so that $Q$ is symmetrical in $\lambda$ and~$\mu$. Thus~$\phi$ being either of these functions \[ S\zeta\phi\zeta = S\nabla\sigma. \] Another important equation is \[ \EqnTag{II}{6$a$} \footnotemark Q(\zeta, \phi\zeta) = Q(\phi'\zeta, \zeta). \] \footnotetext{% [** PP: Keeping ``practice''] [Note added, 1892. For practice it is convenient to remember this in words:---\emph{A term in which $\zeta$ and~$\phi\zeta$ occur is unaltered in value by changing them into $\phi'\zeta$ and~$\zeta$ respectively.}]} This is quite independent of the form of~$\phi$. To prove, observe that by \equationref{II}{2} \[ \phi'\zeta = -\zeta_1 S\zeta\phi\zeta_1, \] and that $-\zeta S\zeta\phi\zeta_1 = \phi\zeta_1$. Thus we get rid of~$\zeta$ and may now drop the suffix of~$\zeta_1$ and so get \hyperref[eqn:II.6$a$]{eq.~(6$a$)}. [Notice that by means of \Eref{II}{6$a$}, \Eref{II}{4$a$} may be deduced from~\Eref{II}{4}; for by~\Eref{II}{6$a$} \[ Q(\zeta, \phi\chi\zeta) = Q(\chi'\zeta, \phi\zeta) = \Sigma Q(\chi'\alpha, \beta)\quad \text{by~\Eref{II}{4}}.] \] \FNParagraph{3$a$}{3a} A more important result is the expression for $\phi^{-1}\omega$ in terms of~$\phi$. We assume that \[ S\phi\lambda\, \phi\mu\, \phi\nu = mS\lambda\mu\nu, \] where $\lambda$, $\mu$, $\nu$ are any three vectors and~$m$ is a scalar independent %% -----File: 035.png---Folio 17------- of these vectors. Substituting $\zeta_1$, $\zeta_2$, $\zeta_3$ for $\lambda$, $\mu$, $\nu$ and multiplying by~$S\zeta_1\zeta_2\zeta_3$ \[ \EqnTag{II}{6$b$} S\zeta_1\,\zeta_2\,\zeta_3\, S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3 = 6m, \] which gives $m$ in terms of~$\phi$. That $S\zeta_1\,\zeta_2\,\zeta_3\, S\zeta_1\,\zeta_2\,\zeta_3 = 6$ is seen by getting rid of each pair of $\zeta$'s in succession thus:--- \[ S\zeta_1\, (S\zeta_1\, V\zeta_2\,\zeta_3)\, \zeta_2\,\zeta_3 = -SV\zeta_2\zeta_3 \centerdot \zeta_2\zeta_3 = S(\zeta_2^2\zeta_3 - \zeta_2 S\zeta_2\,\zeta_3)\,\zeta_3 = -2 \zeta_3\zeta_3 = 6. \] Next observe that \[ S\phi\omega\, \phi\zeta_1\, \phi\zeta_2 = mS\omega\,\zeta_1\,\zeta_2. \] Multiplying by $V\zeta_1\zeta_2$ and again on the right getting rid of the $\zeta$s we have \[ \EqnTag{II}{6$c$} V\zeta_1\zeta_2\, S\phi\omega\, \phi\zeta_1\, \phi\zeta_2 = -2m\omega, \] whence from \equationref{II}{6$b$} \[ \omega S\zeta_1\zeta_2\zeta_3\, S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3 = -3V\zeta_1\zeta_2\, S\phi\omega\, \phi\zeta_1\, \phi\zeta_2, \] or changing $\omega$ into $\phi^{-1}\omega$ \[ \EqnTag{II}{6$d$} \phi^{-1}\omega = -\frac{3V\zeta_1\zeta_2\, S\omega\, \phi\zeta_1\, \phi\zeta_2} {S\zeta_1\zeta_2\zeta_3\, S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3}. \] By \equationref{II}{6$a$} of last section we can also put this in the form \[ \EqnTag{II}{6$e$} \phi^{-1}\omega = -\frac{3V\phi'\zeta_1\, \phi'\zeta_2\, S\omega\zeta_1\zeta_2} {S\zeta_1\zeta_2\zeta_3\, S\phi'\zeta_1\, \phi'\zeta_2\, \phi'\zeta_3}, \] so that $\phi^{-1}\omega$ is obtained explicitly in terms of $\phi$ or~$\phi'$. \Equationref{II}{6$c$} or~\Eref{II}{6$d$} can be put in another useful form which is more analogous to the ordinary cubic and can be easily deduced therefrom, or \footnotemark{}less easily from~\Eref{II}{6$d$}, viz.\ \Erratum{err6f}% [** PP: Re-breaking] \[ \EqnTag{II}{6$f$} \begin{split} S\zeta_1\zeta_2\zeta_3 (\phi^3\omega\, S\zeta_1\zeta_2\zeta_3 &- 3\phi^2\omega\, S\zeta_1\zeta_2 \phi\zeta_3 \\ &+ 3\phi\omega\, S\zeta_1\, \phi\zeta_2\, \phi\zeta_3 - \omega S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3) = 0. \end{split} \] \footnotetext{[Note added, 1892. The cubic may be obtained in a more useful form from the equation $\omega S\zeta_1\zeta_2\zeta_3\, S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3 = -3V\zeta_1\zeta_2\, S\phi\omega\, \phi\zeta_1\, \phi\zeta_2$ thus \begin{align*} V\zeta_1\zeta_2\, S\phi\omega\, \phi\zeta_1\, \phi\zeta_2 &= \phi\omega S \centerdot \phi\zeta_1 \phi\zeta_2 V\zeta_1\zeta_2 - \phi\zeta_1 S \centerdot \phi\omega \phi\zeta_2 V\zeta_1\zeta_2 + \phi\zeta_2 S \centerdot \phi\omega \phi\zeta_1 V\zeta_1\zeta_2 \\ % &= \phi\omega S \centerdot \phi\zeta_1\, \phi\zeta_2\, V\zeta_1\zeta_2 - 2\phi\zeta_1\, S \centerdot \phi\omega\, \phi\zeta_2\, V\zeta_1\zeta_2. \end{align*} % \begin{flalign*} &\text{\indent Again}& \phi\zeta_1\, S \centerdot \phi\omega\, \phi\zeta_2\, V\zeta_1\zeta_2 &= \phi\zeta_1\, S \centerdot \phi\omega V \centerdot \phi\zeta_2\, V\zeta_1\zeta_2 && \\ % [** PP: Adding line break] && &= \phi\zeta_1\, S\phi\omega (-\zeta_1 S\zeta_2\, \phi\zeta_2 + \zeta_2\, S\zeta_1\, \phi\zeta_2) && \\ && &= -\phi(\zeta_1\, S\zeta_1\, \phi\omega) S\zeta_2\, \phi\zeta_2 + \phi(\zeta_1\, S\zeta_1\, \phi\zeta_2)\, S\zeta_2\, \phi\omega && \\ && &= \phi^2\omega\, S\zeta\, \phi\zeta - \phi^2\zeta\, S\zeta\, \phi\omega = \phi^2\omega\, S\zeta\, \phi\zeta + \phi^3\omega. && \\ % &\text{\indent Hence}& &\phi^3\omega - m''\phi^2\omega + m'\phi\omega - m\omega = 0, && \\ &\text{where}& 6m &= S\zeta_1\,\zeta_2\,\zeta_3\, S\phi\zeta_1\, \phi\zeta_2\, \phi\zeta_3 && \\ && 2m' &= -SV\zeta_1\,\zeta_2\, V\phi\zeta_1\, \phi\zeta_2 && \\ && m'' &= -S\zeta\, \phi\zeta.] && \end{flalign*}} % [** end of footnote] %% -----File: 036.png---Folio 18------- As a useful particular case of \equationref{II}{6$d$} we may notice that by \equationref{II}{4} \Secref{3} if \Erratum{err18a}% [** PP: Added comma after next equation] \begin{flalign*} && \phi\omega &= -\sigma_1 S\omega\nabla_1, && \\ % \EqnTag{II}{6$g$} && \phi^{-1}\omega &= -3V \nabla_1\nabla_2 S\omega\, \sigma_1\sigma_2 / S\nabla_1\nabla_2\nabla_3\, S\sigma_1\sigma_2\sigma_3, && \\ % \EqnTag{II}{6$h$} &\text{and}& \phi'^{-1}\omega &= -3V \sigma_1\sigma_2\, S\omega\,\nabla_1\nabla_2 / S\nabla_1\nabla_2\nabla_3\, S\sigma_1\sigma_2\sigma_3. && \end{flalign*} \Paragraph{4} Let $\phi$, $\psi$ be two linear vector functions of a vector. Then if \[ S\chi\zeta\, \phi\zeta = S\chi\zeta\, \psi\zeta, \] where $\chi$ is a \emph{quite arbitrary} linear vector function \[ \phi \equiv \psi, \] for we may put $\chi\zeta = \tau S\zeta\omega$ where $\tau$ and~$\omega$ are arbitrary vectors, so that \begin{flalign*} && S\tau\, \phi\omega &= S\tau\,\psi\omega, && \\ &\text{or}& \phi\omega &= \psi\omega. && \end{flalign*} Similarly\footnote{[Note added, 1892. The following slightly more general statement is a practically much more convenient form of enunciation: if $S\chi\zeta\, \phi\zeta = S\chi\zeta\, \psi\zeta$, where~$\chi$ is a perfectly arbitrary self-conjugate and $\phi$ and~$\psi$ are not necessarily self-conjugate then $\overline{\phi} = \overline{\psi}$].} if $\phi$ and~$\psi$ are both self-conjugate and $\chi$~is a quite arbitrary self-conjugate linear vector function the same relation holds as can be seen by putting \[ \chi\zeta = \tau S\zeta\omega + \omega S\zeta\tau. \] \Section{5}{Fundamental Property of \texorpdfstring{$\protect\Dop$}{the D operator}} \Paragraph{5} Just as the fundamental property of $\Nabla{\sigma}$ is that, $Q$~being any function of~$\sigma$ \[ \delta Q = -Q_1 S\, \delta\sigma\, \Nabla{\sigma}_1, \] so we have a similar property of~$\Dop$. $Q$~being any function of~$\phi$ a linear vector function \[ \EqnTag{II}{7} \delta Q = -Q_1 S\,\delta\phi\zeta\, \Pre{\phi}{\Dop}_1 \zeta. \] The property is proved in the same way as for~$\nabla$, viz.\ by expanding $S\, \delta\phi\, \zeta \Pre{\phi}{\Dop}_1 \zeta$ in terms of the coordinates of $\Pre{\phi}{\Dop}$. First let~$\phi$ be not self-conjugate, and let its nine coordinates be \[ (a_1 b_1 c_1\, a_2 b_2 c_2\, a_3 b_3 c_3). \] %% -----File: 037.png---Folio 19------- Thus \begin{align*} - Q_1\, S\,\delta\phi\,\zeta\,\Pre{\phi}{\Dop}_1 \zeta &= - Q_1\, S\, \delta\phi i\, \Pre{\phi}{\Dop}_1 i - Q_1\, S\, \delta\phi j\, \Pre{\phi}{\Dop}_1 j - Q_1\, S\, \delta\phi k\, \Pre{\phi}{\Dop}_1 k, \\ &= \delta a_1\, dQ/da_1 + \delta b_1\, dQ/db_1 + \delta c_1\, dQ/dc_1, \\ &+ \delta a_2\, dQ/da_2 + \delta b_2\, dQ/db_2 + \delta c_2\, dQ/dc_2, \\ &+ \delta a_3\, dQ/da_3 + \delta b_3\, dQ/db_3 + \delta c_3\, dQ/dc_3 = \delta Q. \end{align*} The proposition is exactly similarly proved when~$\phi$ is self-conjugate. \Section{6}{Theorems in Integration} \Paragraph{6} Referring back to \Secref{1} above for our notation for linear surface and volume integrals we will now prove that if $Q$ be \emph{any} linear function of a vector\footnote{These two propositions are generalisations of what Tait and Hicks have from time to time proved. They were first given in the present form by me in the article already referred to in \Secref{1} above. In that paper the necessary references are given.} \begin{align*} \EqnTag{II}{8} \smallint Q\, d\rho &= \smalliint Q\, (V\,d\Sigma\, \Delta), \\ % \EqnTag{II}{9} \smalliint Q\, d\Sigma &= \smalliiint Q\Delta\, \ds. \end{align*} To prove the first divide the surface up into a series of elementary parallelograms by two families of lines---one or more members of one family coinciding with the given boundary,---apply the line integral to the boundary of each parallelogram and sum for the whole. The result will be the linear integral given in \equationref{II}{8}. Let the sides of one such parallelogram taken in order in the positive direction be $\alpha$, $\beta+\beta'$, $-\alpha-\alpha'$, $-\beta$; so that $\alpha'$ and $\beta'$ are infinitely small compared with $\alpha$ and~$\beta$, and we have the identical relation \[ 0 = \alpha + \beta + \beta' - \alpha -\alpha' - \beta = \beta' - \alpha'. \] The terms contributed to $\int Q\,d\rho$ by the sides $\alpha$ and $-\alpha - \alpha'$ will be (neglecting terms of the third and higher orders of small quantities) \[ Q\alpha - Q\alpha - Q\alpha' + Q_1\alpha\, S\beta \nabla_1 = -Q\alpha' + Q_1\alpha\, S\beta\nabla_1. \] Similarly the terms given by the other two sides will be \[ Q\beta' - Q_1\beta\, S \alpha \nabla_1 \] %% -----File: 038.png---Folio 20------- so that remembering that $\beta' - \alpha' = 0$ and therefore $Q\beta' - Q\alpha' = 0$ we have for the whole boundary of the parallelogram \[ Q_1 \alpha\, S\beta\nabla_1 - Q_1 \beta\, S\alpha\nabla_1 = Q_1(VV\alpha\beta \centerdot \nabla_1) = QV\, d\Sigma \Delta, \] where~$d\Sigma$ is put for~$V\alpha\beta$. Adding for the whole surface we get \equationref{II}{8}. \Equationref{II}{9} is proved in an exactly similar way by splitting the volume up into elementary parallelepipeda by three families of surfaces one or more members of one of the families coinciding with the given boundary. If $\alpha$, $\beta$, $\gamma$ be the vector edges of one such parallelepiped we get a term corresponding to $Q\beta' - Q\alpha'$ viz.\ \[ Q\text{(vector sum of surface of parallelepiped)} \equiv 0, \] and we get the sum of three terms corresponding to \[ Q_1 \alpha\, S\beta\nabla_1 - Q_1 \beta\, S\alpha\nabla_1 \] above, viz.\ \[ -Q_1(V\beta\gamma)\, S\alpha\nabla_1 -Q_1(V\gamma\alpha)\, S\beta\nabla_1 -Q_1(V\alpha\beta)\, S\gamma\nabla_1 = -Q_1 \nabla_1\, S\alpha\beta\gamma, \] whence putting $S\alpha\beta\gamma = -\ds$ we get \equationref{II}{9}. \Paragraph{7} It will be observed that the above theorems have been proved only for cases where we can put $dQ = -Q_1 S\,d\rho\nabla_1$ i.e.\ when the space fluxes of~$Q$ are finite. If at any isolated point they are not finite this point must be shut off from the rest of the space by a small closed surface or curve as the case may be and this surface or curve must be reckoned as part of the boundary of the space. If at a surface (or curve) $Q$ has a discontinuous value so that its derivatives are there infinite whereas on each side they are finite, this surface (or curve) must be considered as part of the boundary and each element of it will occur twice, i.e.\ once for the part of the space on each side. In the case of the isolated points, if the surface integral or line integral round this added boundary vanish, we can of course cease to consider these points as singular. Suppose $Q$ becomes infinite at the point $\rho = \alpha$. Draw a small sphere of radius~$a$ and also a sphere of unit radius with the point~$\alpha$ for centre, and consider the small sphere to be the added boundary. Let~$d\Sigma'$ be the element of the unit sphere cut off by the cone which has $\alpha$ for vertex and the element~$d\Sigma$ of the small sphere for base. Then $d\Sigma = a^2 d\Sigma'$ and we get for the part of the surface integral considered $a^2 \smalliint Q_{d\Sigma}\,d\Sigma'$ where $Q_{d\Sigma}$ is the value of~$Q$ at the element~$d\Sigma$. If then \[ \text{Lt}_{T(\rho-\alpha) = 0}\ T^2(\rho-\alpha) Q_{d\Sigma} U(\rho-\alpha) = 0 \] %% -----File: 039.png---Folio 21------- the point may be regarded as not singular. If the limiting expression is finite the added surface integral will be finite. If the expression is infinite the added surface integral will be generally but not always infinite. Similarly in the case of an added line integral if $\text{Lt}_{T(\rho-\alpha)=0} T(\rho-\alpha) QU(\rho-\alpha)$ is zero or finite, the added line integral will be zero or finite respectively (of course including in the term finite a possibility of zero value). If this expression be infinite, the added line integral will generally also be infinite. This leads to the consideration of potentials which is given in~\Secref{9}. \Paragraph{8} Some particular cases of equations \Eref{II}{8} and~\Eref{II}{9} which (except the last) have been proved by Tait, are very useful. First put $Q =$ a simple scalar~$P$. Thus \begin{align*} \EqnTag{II}{10} \smallint P\,d\rho &= \smalliint V\,d\Sigma\, \nabla P, \\ % \EqnTag{II}{11} \smalliint P\,d\Sigma &= \smalliiint\nabla P\, \ds. \end{align*} If $P$ be the pressure in a fluid $-\smalliint P\,d\Sigma$ is the force resulting from the pressure on any portion and \equationref{II}{11} shews that $-\nabla P$~is the force per unit volume due to the same cause. Next put $Q\omega = S\omega\sigma$ and $V\omega\sigma$. Thus \begin{align*} \EqnTag{II}{12} \smallint S\, d\rho\sigma &= \smalliint S\, d\Sigma\, \nabla\sigma, \\ % \EqnTag{II}{13} \smallint V\, d\rho\sigma &= \smalliint V(V\, d\Sigma\, \nabla\centerdot \sigma) = \smalliint d\Sigma\, S\nabla\sigma - \smalliint\nabla_1 S\, d\Sigma\sigma_1, \\ % \EqnTag{II}{14} \smalliint S\,d\Sigma\sigma &= \smalliiint S\nabla\sigma\, \ds, \\ % \EqnTag{II}{15} \smalliint V\,d\Sigma\sigma &= \smalliiint V\nabla\sigma\, \ds. \end{align*} Equations \Eref{II}{12} and~\Eref{II}{14} are well-known theorems, and \Eref{II}{13} and~\Eref{II}{15} will receive applications in the following pages. Green's Theorem with Thomson's extension of it are, as indeed has been pointed out by Tait particular cases of these equations. Equations \Eref{II}{14} and~\Eref{II}{15} applied to an element give the well-known physical meanings for $S\nabla\sigma$ and~$V\nabla\sigma$\footnote{% [Note added, 1892. Let me disarm criticism by confessing that what follows concerning $V\nabla\sigma$ is nonsense.]}. The first is obtained by applying~\Eref{II}{14} to any element, and the second (regarding $\sigma$ as a velocity) is obtained by applying~\Eref{II}{15} to the element contained by the following six planes each passing infinitely near to the point considered---(1) two planes containing the instantaneous axis of rotation, (2) two planes at right angles to this axis, and (3) two planes at right angles to these four. %% -----File: 040.png---Folio 22------- One very frequent application of \equationref{II}{9} may be put in the following form:---$Q$ being any linear function (varying from point to point) of $R_1$ and~$\nabla_1$, $R$ being a function of the position of a point \[ \EqnTag{II}{16} \smalliiint Q(R_1, \nabla_1)\, \ds = - \smalliiint Q_1(R, \nabla_1)\, \ds + \smalliint Q(R, d\Sigma). \] \Section{9}{Potentials} \Paragraph{9} We proceed at once to the application of these theorems in integration to Potentials. Although the results about to be obtained are well-known ones in Cartesian Geometry or are easily deduced from such results it is well to give this quaternion method if only for the \emph{collateral} considerations which on account of their many applications in what follows it is expedient to place in this preliminary section. If $R$ is some function of $\rho - \rho'$ where $\rho'$ is the vector coordinate of some point under consideration and $\rho$ the vector coordinate of any point in space, we have \[ \Nabla{\rho} R = - \Nabla{\rho'} R. \] Now let $Q(R)$ be any function of~$R$, the coordinates of~$Q$ being functions of~$\rho$ only. Consider the integral $\smalliiint Q(R)\, \ds$ the variable of integration being~$\rho$ ($\rho'$~being a constant so far as the integral is concerned). It does not matter whether the integral is a volume, surface or linear one but for conciseness let us take it as a volume integral. Thus we have \[ \Nabla{\rho'}\smalliiint Q(R)\, \ds = \smalliiint \Nabla{\rho'} Q(R)\, \ds = -\smalliiint \Nabla{\rho}_1 Q(R_1)\, \ds. \] Now $\Nabla{\rho}$ operating on the whole integral has no meaning so we may drop the affix to the~$\nabla$ outside and always understand~$\rho'$. Under the integral sign however we must retain the affix $\rho$ or~$\rho'$ unless a convention be adopted. It is convenient to adopt such a convention and since~$Q$ will probably contain some $\Nabla{\rho}$ but cannot possibly contain a~$\Nabla{\rho'}$ we must assume that when~$\nabla$ appears without an affix under the integral sign the affix~$\rho$ is understood. With this understanding we see that when~$\nabla$ crosses the integral sign it must be made to change sign and refer only to the part we have called~$R$. Thus \[ \EqnTag{II}{17} \nabla\smalliiint Q(R)\, \ds = -\smalliiint \nabla_1Q(R_1)\, \ds. \] %% -----File: 041.png---Folio 23------- Generally speaking~$R$ can and will be put as a function of~$T^{-1} (\rho'-\rho)$ and for this we adopt the single symbol~$u$. Both this symbol and the convention just explained will be constantly required in all the applications which follow. \Paragraph{10} Now let~$Q$ be any function of the position of a point. Then the potential of the volume distribution of~$Q$, say~$q$, is given by:--- \[ \EqnTag{II}{18} q = \smalliiint uQ\, \ds, \] the extent of the volume included being supposed given. We may now prove the following two important propositions \begin{align*} \EqnTag{II}{19} \nabla^2q &= 4\pi Q, \\ % \EqnTag{II}{20} \smalliint S\, d\Sigma\, \nabla\centerdot q &= 4\pi\smalliiint Q\, \ds. \end{align*} The latter is a corollary of the former as is seen from \equationref{II}{9} \Secref{6} above. \Equationref{II}{19} may be proved thus. If $P$~be any function of the position of a point which is finite but not necessarily continuous for all points \[ \nabla\smalliiint uP\, \ds, \] is always finite and if the volume over which the integral extends is indefinitely diminished, so also is the expression now under consideration, and this for the point at which is this remnant of volume. This in itself is an important proposition. The expression, by \equationref{II}{17}, $=-\smalliiint\nabla uP\, \ds$ and both statements are obviously true except for the point~$\rho'$. For this point we have merely to shew that the part of the volume integral just given contributed by the volume indefinitely near to~$\rho'$ vanishes with this volume. Divide this near volume up into a series of elementary cones with~$\rho'$ for vertex. If~$r$ is the (small) height and~$d\omega$ the solid vertical angle of one of these cones, the part contributed by this cone is approximately $U (\rho-\rho') P_{\rho'}r\,d\omega/3$ where $P_{\rho'}$ is the value of~$P$ at the point~$\rho'$. The proposition is now obvious. Now since \[ \nabla^2q = \nabla^2\smalliiint uQ\, \ds = \smalliiint\nabla^2 uQ\, \ds, \] we see that the only part of the volume integral $\smalliiint uQ\, \ds$ which need be considered is that given by the volume in the immediate neighbourhood of~$\rho'$, for at all points except~$\rho'$, $\nabla^2u = 0$. Consider then our volume and surface integrals only to refer to a small %% -----File: 042.png---Folio 24------- sphere with~$\rho'$ for centre and so small that no point is included at which~$Q$ is discontinuous and therefore $\nabla Q$~infinite. (This last assumes that~$Q$ is not discontinuous at~$\rho'$. In the case when $Q$ is discontinuous at~$\rho'$ no definite meaning can be attached to the expression~$\nabla^2 q$.) We now have \begin{align*} \nabla^2 q &= \nabla^2\smalliiint uQ\, \ds = -\nabla \smalliiint \nabla u\, Q\, \ds \quad \text{[by \Secref{9}]} \\ &= \nabla \smalliiint u\, \nabla Q\, \ds - \nabla \smalliint u\, d\Sigma\, Q, \end{align*} \equationref{II}{9} \Secref{6} being applied and the centre of the sphere being not considered as a singular point since the condition of~\Secref{7} is satisfied, viz.\ that $\text{Lt}_{T(\rho-\rho')=0}\ T^2(\rho-\rho') uU(\rho-\rho') = 0$. Now putting $P~\text{ above} = \nabla Q$ we see that the first expression, viz.\ $\nabla\smalliiint u\, \nabla Q\, \ds$ can be neglected and the second gives \[ \nabla^2 q = \smalliint\nabla u\,d\Sigma\, Q = 4\pi \overline{Q}, \] where~$\overline{Q}$ is the mean value of~$Q$ over the surface of the sphere and therefore in the limit~$= Q$. Thus \equationref{II}{19} has been proved. When~$Q$ has a simple scalar value all the above propositions, and indeed processes, become well-known ones in the theory of gravitational potential. We do not propose to go further into the theory of Potentials as the work would not have so direct a bearing on what follows as these few considerations. %% -----File: 043.png---Folio 25------- \FNChapter{Section III}{Elastic Solids}{\protect\footnotemark Elastic Solids} \footnotetext{[Note added, 1892. It would be better to head this section ``Elastic \emph{bodies}'' since except when the strains are assumed small the equations are equally true of solids and fluids. I may say here that I have proved in the \textit{Proc.~R.~S.~E.} 1890--91, pp.~106 \textit{et seq.},\ most of the general propositions of this section somewhat more neatly though the processes are essentially the same as here.]} \Section{11}{Brief recapitulation of previous work in this branch} \Paragraph{11} As far as I am aware the only author who has applied Quaternions to Elasticity is Prof.\ Tait. In the chapter on Kinematics of his treatise on Quaternions, \S\S~360--371, he has considered the mathematics of strain with some elaboration and again in the chapter on Physical Applications, \S\S~487--491, he has done the same with reference to stress and also its expression in terms of the displacement at every point of an elastic body. In the former he has very successfully considered various useful decompositions of strain into pure and rotational parts and so far as strain alone is considered, i.e.\ without reference to what stress brings it about he has left little or nothing to be done. In the latter he has worked out the expressions for stress by means of certain vector functions at each point, which express the elastic properties of the body at that point. But as far as I can see his method will not easily adapt itself to the solution of problems which have already been considered by other methods, or prepare the way for the solution of fresh problems. To put Quaternions in this position is our present object. I limit myself to the statical aspect of Elasticity, but I believe that Quaternions can be as readily, or nearly so, applied to the Kinetics of the subject. For the sake of completeness I shall repeat in my own notation a small part of the work that Tait has given. %% -----File: 044.png---Folio 26------- Tait shews (\S~370 of his \textit{Quaternions}) that in any small portion of a strained medium the strain is homogeneous and (\S~360) that a homogeneous strain function is a linear vector one. He also shews (\S~487) that the stress function is a linear vector one and he obtains expressions (\S\S~487--8) for the force per unit volume due to the stress, in terms of the space-variation of the stress. \Section{12}{Strain, Stress-force, Stress-couple} \Paragraph{12} This last however I give in my own notation. His expression in \S~370 for the strain function I shall throughout denote by~$\chi$. Thus \[ \EqnTag{III}{1} \chi\omega = \omega - S\omega\,\nabla\centerdot \eta, \] where $\eta$ is the displacement that gives rise to the strain. Let $\chi$ consist of a pure strain~$\psi$ followed by a rotation~$q(\;)q^{-1}$ as explained in Tait's \textit{Quaternions}, \S~365 where he obtains both $q$ and~$\psi$ in terms of~$\chi$. Thus \[ \EqnTag{III}{2} \chi\omega = q\,\psi\omega\, q^{-1}. \] When the strain is small $\psi$ takes the convenient form~$\overline{\chi}$ where~$\chi$ stands for the pure part of~$\chi$ so that \[ \EqnTag{III}{3} \overline{\chi}\omega = \omega - \tfrac{1}{2} S\omega\nabla\centerdot \eta - \tfrac{1}{2} \nabla_1 S\omega\eta_1, \] by \equationref{II}{3} \Secref{2} above. Similarly $q(\;)q^{-1}$ becomes $V\theta(\;)$ where $2\theta = V\nabla\eta$. The truth of these statements is seen by putting in \equationref{III}{2} for~$q$, $1+\theta/2$ and therefore for~$q^{-1}$, $1-\theta/2$ for~$\psi$, $\overline{\chi}$ and then neglecting all small quantities of an order higher than the first. \Paragraph{13} Next let us find the force and couple per unit volume due to a stress which varies from point to point. Let the stress function be~$\phi$. Then the force on any part of the body, due to stress, is \[ \smalliint \phi\, d\Sigma = \smalliiint \phi\Delta\, \ds, \] by \equationref{II}{9} \Secref{6}. Thus \emph{the force per unit volume $= \phi\Delta$}, for the volume considered in the equation may be taken as the element~$\ds$. Again the moment round any arbitrary origin is \[ \smalliint V\rho\phi\, d\Sigma = \smalliiint V\rho_1\phi \nabla_1\, \ds + \smalliiint V\rho\phi_1 \nabla_1\, \ds, \] %% -----File: 045.png---Folio 27------- by \equationref{II}{9} \Secref{6}. The second term on the right is that due to the force~$\phi\Delta$ just considered, and the first shews that in addition to this \emph{there is a couple per unit volume ${}=V\zeta\phi\zeta={}$ twice the ``rotation'' vector} of~$\phi$. Let then~$\overline{\phi}$ be the pure part and~$V\epsilon(\;)$ the rotatory part of~$\phi$. Thus \emph{Force} per unit volume \[ \EqnTag{III}{4} = \phi\Delta = \overline{\phi}\Delta - V\nabla\epsilon, \] \emph{Couple} per unit volume \[ \EqnTag{III}{5} = V\zeta \phi\zeta = 2\epsilon. \] These results are of course equivalent to those obtained by Tait, \textit{Quaternions}, \S\S~487--8. The meanings just given to $\eta$, $\chi$, $\overline{\chi}$, $\psi$, $q$, $\phi$, $\overline{\phi}$ and~$\epsilon$ will be retained throughout this Section. In all cases of small strain as we have seen we may use $\psi$ or~$\overline{\chi}$ indifferently and whenever we wish to indicate that we are considering the physical phenomenon of pure strain we shall use $\psi$, $\overline{\chi}$ being regarded merely as a function of~$\chi$. We shall soon introduce a function~$\varpi$ which will stand towards~$\phi$ somewhat as~$\psi$ towards~$\chi$ and such that when the strain is small $\varpi = \overline{\phi}$. It is to be observed that~$\phi\omega$ is the force exerted on a vector area, which \emph{when strained} is~$\omega$, not the stress on an area which before strain is~$\omega$. Similarly in equations \Eref{III}{4} and~\Eref{III}{5} the independent variable of differentiation is~$\rho + \eta$ so that strictly speaking in~\Eref{III}{4} we should put $\overline{\phi}_{\rho+\eta}\Delta - V_{\rho+\eta}\nabla\epsilon$. In the case of small strain these distinctions need not be made. \Section{14}{Stress in terms of strain} \Paragraph{14} To express stress in terms of strain we assume any displacement and consequent strain at every point of the body and then give to every point a small additional displacement $\delta\eta$ and find in terms of $\psi$ and~$\phi$ the increment $\smalliiint\delta w\, \ds_0$ in the potential energy of the body, $w\, \ds_0$ being the potential energy of any element of the body whose volume before strain was $\ds_0$. Thus \Erratum{err27a}% [** PP: Removed comma after first term] \begin{align*} \smalliiint \delta w\, \ds_0 &= \text{(work done by stresses on surface of portion considered)} \\ &- \text{(work done by stresses throughout volume of same portion).} \end{align*} %% -----File: 046.png---Folio 28------- Thus, observing that by \Secref{12} the rotation due to the small displacement $\delta\eta$ is $V_{\rho+\eta} \nabla\delta\eta/2$, we have \[ \smalliiint \delta w\, \ds_0 = - \smalliint S\, \delta\eta\, \phi\, d\Sigma + \smalliiint S\, \delta\eta\, \phi_{1\rho+\eta} \nabla_1\, \ds + \smalliiint S\epsilon_{\rho+\eta} \nabla\delta\eta\, \ds. \] The first of the terms on the right is the work done on the surface of the portion of the body considered; the second is $-$(work done by stress-forces $\phi_{\rho+\eta}\Delta$); and the third is $-$(work done by stress-couples $2\epsilon$). Thus converting the surface integral into a volume integral by \equationref{II}{9} \Secref{6} above \begin{align*} \smalliiint \delta w\, \ds_0 &= -\smalliiint S\, \delta\eta_1\, \phi_{\rho+\eta} \nabla_1\, \ds + \smalliiint S\, \delta\eta_1\, V\epsilon_{\rho+\eta} \nabla_1\, \ds \\ &= -\smalliiint S\, \delta\eta_1\, \overline{\phi}_{\rho+\eta} \nabla_1\, \ds. \end{align*} Limiting the portion of the body considered to the element $\ds$ we get \[ \EqnTag{III}{6} \delta w = -mS\,\delta\eta_1\, \overline{\phi}_{\rho+\eta} \nabla_1 \] where $m$ is put for~$\ds/\ds_0$ and therefore may be put by \Secref{3$a$} and~\Secref{3} above in the various forms \begin{align*} \EqnTag{III}{6$a$} 6m &= S\zeta_1\zeta_2\zeta_3\, S\chi\zeta_1 \chi\zeta_2 \chi\zeta_3 \\ % \EqnTag{III}{6$b$} &= S\zeta_1\zeta_2\zeta_3\, S\chi'\zeta_1 \chi'\zeta_2 \chi'\zeta_3 \\ % \EqnTag{III}{6$c$} &= S\zeta_1\zeta_2\zeta_3\, S\psi\zeta_1 \psi\zeta_2 \psi\zeta_3 \\ % \EqnTag{III}{6$d$} &= S\nabla_1\nabla_2\nabla_3\, S(\rho+\eta)_1(\rho+\eta)_2(\rho+\eta)_3. \end{align*} It is to be observed that since the rotation-vector $\epsilon$~of $\phi$ does not occur in \equationref{III}{6}, $\epsilon$ and therefore the stress-couple are quite arbitrary so far as the strain and potential energy are concerned. I do not know whether this has been pointed out before. Of course other data in the problem give the stress-couple. In fact it can be easily shewn that in all cases whether there be equilibrium or not the external couple per unit volume balances the stress-couple. [Otherwise the angular acceleration of the element would be infinite.] Thus if~$\mathbf{M}$ be the external couple per unit volume of the \emph{unstrained} solid we have always \[ \EqnTag{III}{7} \mathbf{M} + 2m\epsilon = 0. \] In the particular case of equilibrium $\mathbf{F}$ being the external force per unit volume of the unstrained solid we have \[ \EqnTag{III}{8} \mathbf{F} + m(\overline{\phi}_{\rho+\eta} \Delta - V_{\rho+\eta} \nabla\epsilon) = 0. \] The mathematical problem is then the same as if for~$\mathbf{F}/m$ we substituted $\mathbf{F}/m + V_{\rho+\eta} \nabla(\mathbf{M}/2m)$ and for $\mathbf{M}$,~zero. In the case of %% -----File: 047.png---Folio 29------- small strains $m$~may be put $= 1$. In this case then the mathematical problem is the same as if for~$\mathbf{F}$ we substituted $\mathbf{F} + V \nabla \mathbf{M}/2$ and for $\mathbf{M}$~zero. Returning to \equationref{III}{6} observe that \[ \delta \chi \omega = - S \chi \omega_{\rho+\eta} \nabla\centerdot \delta \eta \] [which is established just as is the equation $\chi\omega = -S\omega \nabla\centerdot (\rho +\eta)$], so that changing~$\omega$ which is \emph{any} vector into~$\chi^{-1}\omega$ \[ \delta \chi\centerdot \chi^{-1} \omega = - S \omega_{\rho+\eta} \nabla \centerdot \delta \eta. \] Therefore by \equationref{III}{6} of this section and \equationref{II}{4} \Secref{3} above we have \[ \delta w = - mS\delta \chi\, \chi^{-1}\,\zeta \overline{\phi} \zeta. \] We must express the differential on the right of this equation in terms of $\delta \psi$ and~$\delta q$, the latter however disappearing as we should expect. Now $\chi\omega = q \psi \omega q^{-1}$ [\equationref{III}{2} \Secref{12}] so that remembering that $\delta\centerdot q^{-1} = - q^{-1}\,\delta q\, q^{-1}$ \begin{flalign*} && \delta \chi \omega &= \delta q\, \psi \omega q^{-1} - q \psi \omega q^{-1}\, \delta q\, q^{-1} + q\,\delta\psi\, \omega q^{-1} && \\ && &= 2VV\,\delta q\, q^{-1}\centerdot \chi \omega + q\, \delta\psi\, \omega q^{-1}, && \\ &\therefore& \delta w / m &= -2 S\centerdot V\,\delta q\, q^{-1}\centerdot \zeta \overline{\phi} \zeta - Sq\,\delta\psi\, \chi^{-1} \zeta\, q^{-1}\overline{\phi}\zeta, && \end{flalign*} or since $V\,\zeta \overline{\phi} \zeta = 0$, $\overline{\phi}$ being self-conjugate \[ \delta w = - m S\, \delta\psi\, \chi^{-1}\, \zeta q^{-1}\overline{\phi}\, \zeta q. \] This can be put into a more convenient shape for our present purpose. First put for~$\chi^{-1}$ its value~$\psi^{-1}q^{-1}(\;)q$ and then apply \equationref{II}{6$a$} \Secref{3} above, putting for the~$\phi$ of that equation~$q^{-1}(\;)q$ and therefore for the~$\phi'$, $q(\;)q^{-1}$. Thus \begin{flalign*} \EqnTag{III}{9} && \delta w &= - m S\, \delta\psi\, \psi^{-1}\, \zeta \varpi \zeta, && \\ % \EqnTag{III}{10} &\text{where}& \varpi \omega &= q^{-1} \overline{\phi} (q\omega q^{-1})q. && \end{flalign*} The physical meaning of this last equation can easily be shewn. Suppose when there is no rotation that $\overline{\phi}=\varpi'$. Then it is natural\footnote{% Observe that we \emph{do not} make this assumption. We really shew that it is true.} to assume---in fact it seems almost axiomatic---that the superimposed rotation~$q(\;)q^{-1}$ should merely so to speak rotate the stress along with it. Thus if~$\omega$ is some vector area before the rotation which becomes~$\omega'$ by means of the rotation \[ q\varpi' \omega\centerdot q^{-1} = \overline{\phi} \omega'. \] %% -----File: 048.png---Folio 30------- But $\omega' = q\omega q^{-1}$, so that \[ \varpi'\omega = q^{-1}\overline{\phi}(q\omega q^{-1})q = \varpi\omega. \] Thus we see that $\varpi$ is as it were~$\overline{\phi}$ with the rotation undone. Before proceeding further with the calculation let us see what we have assumed and what \equationref{III}{9} teaches us. The one thing we have assumed is that the potential energy of the body can be taken as the sum of the potential energies of its elements, in other words that no part of the potential energy depends conjointly on the strains at $P$ and~$Q$ where $P$ and~$Q$ are points separated by a finite distance. This we must take as an axiom. By it we are led to the expression for~$\delta w$ in \equationref{III}{9}. This only involves the variation of the pure strain~$\psi$ \emph{but not the space differential coefficients of~$\psi$}. This is not an obvious result as far as I can see but it is I believe always assumed without proof. We may \emph{now} regard~$w$ as a function of $\psi$~only. Therefore by \equationref{II}{7} \Secref{5} above \[ \delta w = - S\, \delta \psi\, \zeta\, \Pre{\psi}{\Dop} w\, \zeta, \] therefore by \equationref{III}{9} of this section \[ S\, \delta\psi\, \zeta\,\Pre{\psi}{\Dop} w\, \zeta = mS\, \delta\psi\, \psi^{-1}\,\zeta\varpi\zeta\footnotemark. \] \footnotetext{By putting $\delta\psi = \omega S(\;)\omega' + \omega' S(\;)\omega$ in this equation, \equationref{III}{11} can be deduced but as this method has already been applied in~\Secref{4} I give the one in the text to shew the variety of Quaternion methods. [Note added,~1892. If we use the theorem in the foot-note of~\Secref{4}, \equationref{III}{11} follows at once.]}% [** PP: End of footnote, no paragraph] % In \equationref{II}{6$a$} \Secref{3} above putting $\phi = \psi^{-1}$ the right-hand member of this equation becomes $mS\,\delta\psi\,\zeta\,\varpi\psi^{-1}\,\zeta$. Now putting in \equationref{II}{6$a$} \Secref{3} $\phi = \varpi\psi^{-1}$ this member becomes $mS\zeta\,\delta\psi\,\psi^{-1}\,\varpi\zeta$ or $mS\,\delta\psi\,\zeta\,\psi^{-1}\varpi\,\zeta$. Thus we have \[ 2S\,\delta\psi\, \zeta\,\Pre{\psi}{\Dop} w\, \zeta = mS\,\delta\psi\, \zeta(\varpi\psi^{-1} + \psi^{-1}\varpi)\zeta. \] Now $\varpi\psi^{-1} + \psi^{-1}\varpi$ is self-conjugate. Hence by~\Secref{4} above \[ \EqnTag{III}{11} m(\varpi\psi^{-1} + \psi^{-1}\varpi) = 2 \Pre{\psi}{\Dop} w. \] This equation can be looked upon as giving~$\varpi$ in terms of the strain. We can obtain~$\varpi$ however explicitly for $\psi^{-1}\varpi$ is the conjugate of~$\varpi\psi^{-1}$. Hence from \equationref{III}{11} (because $(\psi^{-1}\varpi + \varpi\psi^{-1})/2$ is the pure part of~$\varpi\psi^{-1}$) \[ m\, \varpi \psi^{-1}\omega = \Pre{\psi}{\Dop} w\, \omega + V\theta\,\omega \] where $\theta$ is a vector to be found. Changing $\omega$ into $\psi\omega$ \[ m\,\varpi\omega = \Pre{\psi}{\Dop} w\, \psi\omega + V \theta\, \psi\omega. \] %% -----File: 049.png---Folio 31------- Now $\varpi$ being self-conjugate $V\zeta\varpi\zeta = 0$. Hence \begin{align*} V\zeta \Pre{\psi}{\Dop} w\, \psi\zeta &= -V\zeta\, V\theta\, \psi\zeta = \theta S \zeta\psi\zeta - \psi\zeta\, S\theta\zeta \\ &= \theta S \zeta\psi\zeta + \psi\theta, \end{align*} whence % [** PP: A bit too far left, but leaving alone] \begin{flalign*} \EqnTag{III}{12} \left.\begin{aligned} && \theta & = (\psi + S\zeta_1\psi\zeta_1)^{-1} V\zeta \Pre{\psi}{\Dop} w\, \psi\zeta \\ % &\text{or\qquad\qquad}& \theta & = (\psi + S\zeta_1\psi\zeta_1)^{-1} V\psi\zeta \Pre{\psi}{\Dop} w\, \zeta \end{aligned}\; % [** PP: Need some extra space] \right\} && \end{flalign*} by \equationref{II}{6$a$} \Secref{3} above. Thus finally \begin{flalign*} \EqnTag{III}{13} \left.\begin{aligned} && m\varpi\omega &= \Pre{\psi}{\Dop} w\, \psi\omega + V\theta\psi\omega && \\ % &\text{or\qquad\qquad}& m\varpi\omega &= \Pre{\psi}{\Dop} w\, \psi\omega - V\psi\omega(\psi + S\zeta_1\psi\zeta_1)^{-1} V\psi\zeta \Pre{\psi}{\Dop} w\, \zeta. \end{aligned}\!\!\right\} && \end{flalign*} This completely solves the problem of expressing stress in terms of strain in the most general case. \Paragraph{15} $\varpi\omega + V\epsilon'\omega$, where, as we saw in last section, $\epsilon'$ is perfectly arbitrary so far as the strain is concerned, is the force on the strained area~$\omega$ due to the pure strain~$\psi$. And again \[ \overline{\phi}\omega + V\epsilon''\omega \quad\text{or}\quad q(\varpi q^{-1}\ \varpi q)q^{-1} + V\epsilon''\omega \] is that due to the strain $q\psi(\;)q^{-1}$ or~$\chi$. To find the force on an area which before strain was~$\omega_0$ let its strained value after~$\psi$ has taken place be~$\omega$. Then by equation~(4), \S~145 of Tait's \textit{Quaternions}, $m\omega_0 = \psi\omega$. Hence \[ \text{Required force} = \varpi\omega + V\epsilon'\omega = \Pre{\psi}{\Dop} w\, \omega_0 + V\theta\omega_0 + mV\epsilon'\psi^{-1}\omega_0 \] by \equationref{III}{13} of last section. If the rotation now take place this force rotates with it so that the force on the area which was originally~$\omega_0$ is after the strain $\chi$ or~$q\psi(\;)q^{-1}$ \[ \EqnTag{III}{14} \tau\omega_0 = q \Pre{\psi}{\Dop} w\,\omega_0 q^{-1} + qV\theta\omega_0\centerdot q^{-1} + V(\epsilon q\psi^{-1}\omega_0\centerdot q^{-1}) \] where $\epsilon = m\,q\epsilon'q^{-1}$ and \[ \text{$\therefore\qquad \epsilon = \tfrac{1}{2}$ the stress-couple per un.\ vol.\ of unstrained body}. \] This force then is a linear vector function of~$\omega_0$, but in general even when $\epsilon = 0$ it is not self-conjugate. When both $\epsilon$ and~the rotation are zero we see that the rotation vector of~$\tau$ is~$\theta$ given by \equationref{III}{12} of last section. The stress-force can be shewn as in~\Secref{13} to be~$\tau\Delta$ per unit volume of the unstrained body. Thus since the corresponding stress-couple is~$2\epsilon$ the moment exerted by the stresses on any portion of the body round an arbitrary origin is \[ \smalliiint V(\rho + \eta)\tau_1\nabla_1\, \ds_0 + 2\smalliiint\epsilon\, \ds_0. \] %% -----File: 050.png---Folio 32------- But this moment may also be put in the form \begin{flalign*} && &\smalliint V(\rho+\eta)\tau\, d\Sigma_0, && \\ % &\text{or\qquad}& \smalliiint V\zeta\tau\zeta \, \ds_0 &+ \smalliiint V(\rho+\eta)\tau_1\nabla_1\, \ds_0 + \smalliiint V\eta_1\tau\nabla_1\, \ds_0, && \end{flalign*} by \equationref{II}{9} \Secref{6} above. Comparing these results \begin{flalign*} \EqnTag{III}{14$a$} &\text{or}& \left. \begin{aligned} 2\epsilon &= V\zeta\tau\zeta + V\eta_1\tau\nabla_1 \\ 2\epsilon &= V\rho'_1\tau\nabla_1, \end{aligned}\; % [** PP: Need some extra space] \right\} && \end{flalign*} in the notation of next section. This equation may also be deduced from \equationref{III}{15$l$} below but not so naturally as above. \Section{16}{The equations of equilibrium} \Paragraph{16} We require \equationref{III}{9} \Secref{14} above to prove the statement that no space-variations of $\psi$ or~$q$ are involved in~$w$. It is also required to shew that the fact that~$q$ also is not involved in~$w$ is a mathematical sequence of the assumption that the potential energy of a solid is the sum of the potential energies of its elements. Assuming these facts however we can arrive at the equation of stress~\Eref{III}{11} \Secref{14} in a different way from the above. We shall also obtain quite different expressions for $\varpi$, $\tau$,~\&c.\ and most important of all we shall obtain the equations of equilibrium by obtaining~$\tau$ explicitly in terms of the displacement and its space derivatives. From \Secref{12} above we have \begin{gather*} \chi\omega = q\psi\omega q^{-1},\qquad \chi'\omega = \psi(q^{-1}\omega q), \\ % \EqnTag{III}{15} \therefore\qquad \chi'\chi = \psi^2 = \Psi, \end{gather*} (say) as in Tait's \textit{Quaternions}, \S~365. Thus $\Psi\omega = \nabla_1 S\omega\, \nabla_2 S\rho'_1\rho'_2$ where~$\rho'$ is put as it will be throughout this section for $\rho + \eta$, the vector coordinate after displacement of the point~$\rho$. From this as we have seen in the Introduction we deduce that the coordinates of~$\Psi$ are the $A$, $B$, $C$, $a$, $b$, $c$ of Thomson and Tait's \textit{Nat.\ Phil.}\ App.~C. We may as do those authors regard~$w$ as a function of~$\Psi$. Thus:--- \begin{align*} \delta w &= -S\, \delta\Psi\, \zeta \Pre{\Psi}{\Dop} w\, \zeta, \\ &= -S\, \delta\chi'\chi\zeta \Pre{\Psi}{\Dop} w\, \zeta -S\chi'\,\delta\chi\, \zeta \Pre{\Psi}{\Dop} w\, \zeta. \end{align*} By \equationref{II}{4$a$} \Secref{3} above, each of the terms in this last expression %[** PP: Next two eqns aligned in text, but conceptually *shouldn't* be] \[ = -S\,\delta\eta_1\chi \Pre{\Psi}{\Dop} w\, \nabla_1, \] \[ \EqnTag{III}{15$a$} \therefore\qquad \delta w = -2S\,\delta\eta_1\chi \Pre{\Psi}{\Dop} w\, \nabla_1, \] %% -----File: 051.png---Folio 33------- Comparing this with \equationref{III}{6} \Secref{14} and putting in both $\delta\eta_1 = \omega'S\omega\rho'$ where $\omega'$, $\omega$ are arbitrary constant vectors, we get \[ mS\omega'\,\overline{\phi}\omega = 2S\omega'\chi \Pre{\Psi}{\Dop} w\, \chi'\omega. \] Hence, since~$\omega'$ is quite arbitrary \[ \EqnTag{III}{15$b$} m\overline{\phi} = 2\chi \Pre{\Psi}{\Dop} w\, \chi'. \] From \equationref{III}{10} \Secref{14} above which \emph{defines} $\varpi$ we see that \[ \EqnTag{III}{15$c$} m\varpi = 2\psi \Pre{\Psi}{\Dop} w\, \psi, \] which we should expect since we have already seen that~$\varpi$ is the value of~$\overline{\phi}$ when there is no rotation and therefore $\chi = \chi' = \psi$. Now since \[ S\, \delta\psi\, \zeta \Pre{\psi}{\Dop} \zeta = S\, \delta\Psi\, \zeta \Pre{\Psi}{\Dop} \zeta = S(\delta\psi\,\psi + \psi\,\delta\psi) \zeta \Pre{\Psi}{\Dop} \zeta, \] we deduce by any one of the processes already exemplified that \[ \Pre{\psi}{\Dop} = \Pre{\Psi}{\Dop}\, \psi + \psi \Pre{\Psi}{\Dop}, \] where of course the differentiations of $\Pre{\Psi}{\Dop}$ must not refer to~$\psi$. We see then from \equationref{III}{15$c$} that \[ m(\varpi\psi^{-1} + \psi^{-1}\varpi) = 2 \Pre{\psi}{\Dop} w, \] which is \equationref{III}{11} \Secref{14}. Our present purpose however is to find the equations of equilibrium. Let~$\omega_0$ be the vector area which by the strain~$\chi$ becomes~$\omega$. Thus\footnote{See \Secref{83} below.} as in last section \[ \EqnTag{III}{15$e$} m\omega_0 = \chi'\omega. \] Further let~$2\epsilon$ be the stress-couple per unit volume of the unstrained solid so that~$2\epsilon/m$ is the same of the strained solid. As we know, $\epsilon$~is quite independent of the strain. By \Secref{13} we see that the force~$\tau\omega_0$ on the area which before strain was~$\omega_0$ is~$\overline{\phi}\omega + V\epsilon\omega/m$. Therefore \[ \EqnTag{III}{15$f$} \tau\omega_0 = 2\chi \Pre{\Psi}{\Dop} w\,\omega_0 + V\epsilon\chi'^{-1}\omega_0. \] \Erratum{err33a}%[** PP: Added missing ``the''] We saw in the last section that the force per unit volume of the unstrained solid is~$\tau\Delta$ and the couple~$2\epsilon$. Hence \begin{align*} \EqnTag{III}{15$g$} \mathbf{F} + \tau\Delta &= 0, \\ % \EqnTag{III}{15$h$} \mathbf{M} + 2\epsilon &= 0, \end{align*} %% -----File: 052.png---Folio 34------- are the equations of equilibrium, where $\mathbf{F}$ and~$\mathbf{M}$ are the external force and couple per unit volume of the unstrained solid. All that remains to be done then is to express~$\tau$ in terms of~$\epsilon$ and the displacement. Putting \[ \EqnTag{III}{15$i$} \rho + \eta = \rho', \] as already mentioned, we have \[ \EqnTag{III}{15$j$} \chi \omega = - \rho'_1 S \omega \nabla_1, \] so that by \equationref{II}{6$h$} \Secref{3$a$} above we have \[ \EqnTag{III}{15$k$} \chi'^{-1}\omega = -3V \rho'_1 \rho'_2\, S \omega \nabla_1 \nabla_2 / S\nabla_1\nabla_2 \nabla_3\, S \rho'_1 \rho'_2 \rho'_3. \] Therefore by \equationref{III}{15$f$} \[ \EqnTag{III}{15$l$} \tau \omega = -2\rho'_1 S \nabla_1 \Pre{\Psi}{\Dop} w\, \omega -3V\epsilon\, V \rho'_1 \rho'_2\, S \omega \nabla_1 \nabla_2 / S \nabla_1 \nabla_2 \nabla_3\, S \rho'_1 \rho'_2 \rho'_3. \] It is unnecessary to write down what \equationref{III}{15$g$} becomes when we substitute for~$\tau$, changing~$\epsilon$ into~$-\mathbf{M}/2$ by \equationref{III}{15$h$}. In the important case however when $\mathbf{M} = 0$, the equation is quite simple, viz.\ \[ \EqnTag{III}{15$m$} \mathbf{F} = 2\rho'_1 S \nabla_1 \Pre{\Psi}{\Dop} w\, \Delta. \] \textit{Addition to \Secref{16}, Dec.}, 1887 (sent in with the Essay). [The following considerations occurred just before I was obliged to send the essay in, so that though I thought them worth giving I had not time to incorporate them in the text. It is interesting to consider the case of an isotropic body. Here~$w$ is a function of the three principal elongations only and therefore we may in accordance with \Secref{14} and~\Secref{15} suppose it a function of $a$, $b$, $c$ or in accordance with \Secref{16} of $A$, $B$, $C$ where \begin{align*} \EqnTag{III}{$A$} a &= -S\zeta \psi \zeta, \quad& b &= -S\zeta \psi^2\zeta/2, \quad& c &= -S\zeta \psi^3\zeta/3. \\ % \EqnTag{III}{$B$} A &= -S\zeta \Psi \zeta, \quad& B &= -S\zeta \Psi^2\zeta/2, \quad& C &= -S\zeta \Psi^3\zeta/3. \end{align*} Let us use $x$, $y$, $z$, $X$, $Y$, $Z$ for the differential coefficients of~$w$ with respect to $a$, $b$, $c$, $A$, $B$, $C$ respectively. Thus \[ dw = x\,da + y\,db +z\,dc = -S\,d\psi\, \zeta (x + y\psi + z\psi^2)\zeta \] as can be easily proved by means of \equationref{II}{6$a$} \Secref{3} above. But \begin{align*} dw &= -S\,d\psi \zeta \Pre{\psi}{\Dop} w\, \zeta. \\ % \EqnTag{III}{$C$} \therefore\qquad \Pre{\psi}{\Dop} w &= x + y\psi + z\psi^2, \\ % \intertext{by \Secref{4} above. Similarly we have} \EqnTag{III}{$D$} \Pre{\Psi}{\Dop} w &= X + Y\Psi + Z \Psi^2. \end{align*} %% -----File: 053.png---Folio 35------- (Notice in passing that to pass to small strains is quite easy for $\Pre{\psi}{\Dop} w$ is linear and homogeneous in $\psi-1$ so that $\Pre{\psi}{\Dop} w = x + y \psi$ where~$y$ is a constant and $x + y$ a multiple of~$a-3$.)\footnote{[Note added, 1892. In the original essay there was a slip here which I have corrected. It was caused by assuming that~$\psi$ instead of~$\psi-1$ was small for small strains. In the original I said ``where $y$ is a constant and~$x$ is a multiple of~$a$.'']} From \equationref{III}{$C$} we see that~$\theta$ in \equationref{III}{12} \Secref{14} is zero and therefore that from \equationref{III}{13} \begin{align*} \EqnTag{III}{$E$} m\varpi &= \Pre{\psi}{\Dop} w\, \psi = x\psi + y\psi^2 + z\psi^3. \\ \intertext{Again from \equationref{III}{14} \Secref{15}} % \EqnTag{III}{$F$} \tau\omega &= xq \omega q^{-1} + y\chi\omega + z\chi\psi\omega + V\epsilon\chi'^{-1}\omega. \\ \intertext{Similarly from equations \Eref{III}{15$b$} and~\Eref{III}{15$f$} \Secref{16} we may prove by \equationref{III}{$D$} that} % \EqnTag{III}{$G$} \tfrac{1}{2}m\overline{\phi} &= \mathbf{X} \chi\chi' + \mathbf{Y}(\chi\chi')^2 + \mathbf{Z}(\chi\chi')^3, \end{align*} \begin{flalign*} \EqnTag{III}{$H$} \text{and} && \tau = 2X\chi + 2Y\chi\chi'\,\chi + 2Z\chi\chi'\,\chi\chi'\,\chi + V\epsilon\chi'^{-1}(\;). && \end{flalign*} If we wish to neglect all small quantities above a certain order the present equations pave the way for suitably treating the subject. I do not however propose to consider the problem here as I have not considered it sufficiently to do it justice.] \TCSection{16$a$}{Variation of temperature}{Variation of Temperature} \FNParagraph{16$a$}{16a} The~$w$ which appears in the above sections is the same as the~$w$ which occurs in Tait's \emph{Thermo-dynamics}, \S~209, and therefore all the above work is true whether the solid experience change of temperature or not. $w$~will be a function then of the temperature as well as~$\psi$. To express the complete mathematical problem of the physical behaviour of a solid we ought of course instead of the above equations of equilibrium, to have corresponding equations of motion, viz.\ equations~\Eref{III}{15$h$}, \Eref{III}{15$l$} and $\bigl($instead of~\Eref{III}{15$g$}$\bigr)$ \[ \EqnTag{III}{15$n$} D \ddot{\rho}' = \mathbf{F} + \tau \Delta, \] where~$D$ is the original density of the solid at the point considered. Further we ought to put down the equations of conduction of heat and lastly equations \Eref{III}{16$f$} and~\Eref{III}{16$d$} below. We do not propose to consider the conduction of heat, but it will be well to shew how the thermo-dynamics of the present question are treated by Quaternions. %% -----File: 054.png---Folio 36------- Let~$t$ be the temperature of the element which was originally $\ds_0$ and $E\, \ds_0$ its intrinsic energy. Let \[ \delta H = - S\, \delta\psi\, \zeta M \zeta + N\, \delta t, \] where $M, N$ are a linear self-conjugate function and a scalar respectively, both functions of $\psi$ and~$t$. Here $\delta H\, \ds_0$ is the heat required to be put into the element to raise its temperature by~$\delta t$ and its pure strain by~$\delta\psi$. Now when $t$~is constant $\delta H$ must be a perfect differential so that we may put \[ M = t \Pre{\psi}{\Dop} f, \] where $f$ is some function of $\psi$ and~$t$. Thus \[ \EqnTag{III}{16} \delta H = -tS\,\delta\psi\, \zeta \Pre{\psi}{\Dop} f\, \zeta + N\, \delta t. \] Now we have seen in \Secref{14} that the work done on the element during the increment~$\delta\psi$, divided by~$\ds_0$ \[ = - m S\, \delta\psi\, \psi^{-1}\, \zeta\varpi\zeta. \] Thus by the first law of Thermo-dynamics \begin{align*} \delta E &= J\, \delta H - m S\, \delta\psi\, \psi^{-1}\, \zeta\varpi\zeta \\ &= JN\, \delta t - Jt S\, \delta\psi\, \zeta \Pre{\psi}{\Dop} f\, \zeta - mS\, \delta\psi\, \psi^{-1}\, \zeta\varpi\zeta, \end{align*} where $J$ is Joule's mechanical equivalent. Thus \begin{flalign*} \EqnTag{III}{16$a$} && JN &= dE/dt, && \\ % \EqnTag{III}{16$b$} &\text{and}& \tfrac{1}{2}m(\varpi\psi^{-1} &+ \psi^{-1}\varpi) = \Pre{\psi}{\Dop} w, && \\ % \EqnTag{III}{16$c$} &\text{where}& w &= E - Jtf. && \end{flalign*} To apply the second law we go through exactly the same cycle as does Tait in his \emph{Thermo-dynamics}, \S~209, viz.\ \[ (\psi, t) (\psi + \delta \psi, t) (\psi + \delta \psi, t + \delta t) (\psi, t + \delta t) (\psi, t). \] We thus get\footnote{[By assuming from the second law that the work done \emph{by} the element in the cycle, i.e.\ the sum of the works done by it during the first and third steps is~$J \delta t/t$ multiplied by the heat absorbed by the element in the third step. Note added,~1893.]} \[ J \Pre{\psi}{\Dop} f = -\tfrac{1}{2} m \left( \frac{d\varpi}{dt} \psi^{-1} + \psi^{-1} \frac{d\varpi}{dt} \right) = -\frac{d \Pre{\psi}{\Dop} w}{dt} = -\Pre{\psi}{\Dop}\frac{dw}{dt} \] \begin{flalign*} \EqnTag{III}{16$d$} &\text{or}& Jf &= - dw/dt && \end{flalign*} the arbitrary function of~$t$ being neglected as not affecting any physical phenomenon. Substituting for~$w$ from \equationref{III}{16$c$}, \[ \EqnTag{III}{16$e$} Jt\, df/dt = dE/dt = JN \] by \equationref{III}{16$a$}. Thus from \equationref{III}{16} \[ \EqnTag{III}{16$f$} \delta H = t\, \delta f, \] %% -----File: 055.png---Folio 37------- so that in elastic solids as in gases we have a convenient function which is called the ``entropy''. Thus the intrinsic energy~$E\, \ds_{0}$, the entropy~$f\, \ds_{0}$ and the stress~$\varpi$ have all been determined in terms of one function~$w$ of $\psi$ and~$t$ which function is therefore in this general mathematical theory supposed to be known. If instead of regarding~$w$ (which with the generalised meaning it now bears may still conveniently be called the ``potential energy'' per unit volume) as the fundamental function of the substance we regard the intrinsic energy or the entropy as such it will be seen that one other function of~$\psi$ must also be known. For suppose~$f$ the entropy be regarded as known. Then since $dw/dt = -Jf$ \[ \EqnTag{III}{16$g$} w = W - J\smallint f\, dt, \] where the integral is any \emph{particular} one and~$W$ is a function of $\psi$~only, supposed known. Again \begin{flalign*} && E &= w + Jtf && \\ % \EqnTag{III}{16$h$} &\text{or}& E &= W + J(tf - \smallint f\, dt). && \end{flalign*} Thus all the functions are given in terms of $f$ and~$W$. Similarly if~$E$ be taken as the fundamental function \begin{align*} \EqnTag{III}{16$i$} w &= t(W' - \smallint E\, dt/t^2), \\ % \EqnTag{III}{16$j$} Jf &= E/t + \smallint E\, dt/t^2 - W', \end{align*} where as before the integral is some particular one and~$W'$ is a function of $\psi$~only. \Section{17}{Small strains} \Paragraph{17} We now make the usual assumption that the strains are so small that their coordinates can be neglected in comparison with ordinary quantities such as the coefficients of the linear vector function $\Pre{\psi}{\Dop} w$. We can deduce this case from the above more general results. To the order considered $q = 1$ so that by \equationref{III}{10} \Secref{14} above \[ \overline{\phi} = \varpi. \] We shall use the symbol~$\varpi$ rather than~$\overline{\phi}$ for the same reasons explained in \Secref{13} above as induce us to use~$\psi$ rather than~$\overline{\chi}$. Remembering that $\omega_0$ and~$\omega$ are now the same and that~$\psi$ may be put $= 1$ so that $V\zeta \Pre{\psi}{\Dop} w\,\psi\zeta = V\zeta \Pre{\psi}{\Dop} w\,\zeta = 0$ and therefore~$\theta$ of \equationref{III}{12} $= 0$, we have from \equationref{III}{13}, \[ \EqnTag{III}{17} \varpi\omega = \Pre{\psi}{\Dop} w\,\omega. \] %% -----File: 056.png---Folio 38------- Of course we do not require to go through the somewhat complicated process of \Secref{14}, \Secref{15} to arrive at this result. In fact in \equationref{III}{6} \Secref{14}, we may put $m = 1$ and $\Nabla{\rho+\eta} = \nabla$ so that \[ \delta w = -S\, \delta\eta_1\, \overline{\phi}\nabla_1 = -S\, \delta\eta_1\, \varpi\nabla_1, \] and therefore by \Secref{3} above \[ \delta w = -S\, \delta\overline{\chi}\, \zeta\varpi\zeta = -S\, \delta\psi\, \zeta\varpi\zeta. \] But by \Secref{5} above \[ \delta w = -S\, \delta\psi\, \zeta \Pre{\psi}{\Dop} w\, \zeta, \] and therefore by \Secref{4} we get \equationref{III}{17}. \Paragraph{18} It is convenient here to slightly change the notation. For $\psi$, $\chi$ we shall now substitute $\psi+1$, $\chi+1$ respectively. This leads to no confusion as will be seen. With this notation the strain being small the stress is linear in~$\psi$ i.e.\ $\Pre{\psi}{\Dop} w$ is linear and therefore~$w$ quadratic. Now for any such quadratic function \begin{align*} \EqnTag{III}{18} w &= -S\psi\, \zeta \Pre{\psi}{\Dop} w\,\zeta/2, \\ \intertext{for we have by \Secref{5},} dw &= -S\, d\psi\, \zeta \Pre{\psi}{\Dop} w\, \zeta. \end{align*} Put now $\psi = n\psi'$. Then because $\Pre{\psi}{\Dop} w$ is linear in~$\psi$ \[ \Pre{\psi}{\Dop} w = n\Pre{\psi}{\Dop}' w, \] where $\Pre{\psi}{\Dop}' w$ is put for the value of $\Pre{\psi}{\Dop} w$ when the coordinates of~$\psi'$ are substituted for those of~$\psi$. Thus keeping $\psi'$~constant and varying~$n$, \[ dw = -n\,dn\,S\psi'\, \zeta \Pre{\psi}{\Dop}' w\,\zeta, \] whence integrating from $n = 0$ to $n = 1$ and changing $\psi'$~into $\psi$ we get \equationref{III}{18}. From \equationref{III}{18} we see that for small strains we have \[ \EqnTag{III}{19} w = -S\psi\, \zeta\varpi\zeta/2. \] Now $w$ is quadratic in~$\psi$ and therefore also quadratic in~$\varpi$, so that regarding~$w$ as a function of~$\varpi$ we have as in \equationref{III}{18} \[ w = -S\varpi\zeta \Pre{\varpi}{\Dop} w\,\zeta/2, \] so that by \equationref{III}{19} and \Secref{4} above \[ \EqnTag{III}{20} \psi = \Pre{\varpi}{\Dop} w. \] All these results for small strains are well-known in their Cartesian form, but it cannot be bias that makes these quaternion %% -----File: 057.png---Folio 39------- proofs appear so much more natural and therefore more simple and beautiful than the ordinary ones. \Paragraph{19} Let us now consider (as in \Secref{16} is really done) $w$ as a function of the displacement. Now $w$ is quadratic in~$\psi$, and~$\psi$ is linear and symmetrical in $\nabla_1$ and~$\eta_1$. In fact from \equationref{III}{3} \Secref{12} above, remembering that the~$\overline{\chi}$ of that equation is our present $\psi + 1$, we have \[ 2 \psi \omega = -\eta_1\, S\omega\, \nabla_1 - \nabla_1\, S\omega\, \eta_1. \] Therefore we may put \[ \EqnTag{III}{21} w = w(\eta_1, \nabla_1, \eta_2, \nabla_2), \] where $w(\alpha, \beta, \gamma, \delta)$ is linear in each of its constituents, is symmetrical in $\alpha$ and~$\beta$, and again in $\gamma$ and~$\delta$, and is also such that the pair $\alpha$, $\beta$ and the pair $\gamma$, $\delta$ can be interchanged. [This last statement can be \emph{made} true if not so at first, by substituting for $w(\alpha, \beta, \gamma, \delta)$, $w(\alpha, \beta, \gamma, \delta)/2 + w(\gamma, \delta, \alpha, \beta)/2$ as this does not affect \equationref{III}{21}.] Such a function can be proved to involve 21~independent scalars, which is the number also required to determine an arbitrary quadratic function of~$\psi$, since~$\psi$ involves six scalars. Thus we have the two following expressions for $\delta w$, which we equate \begin{align*} -S\, \delta\psi\, \zeta\varpi\zeta &= w (\delta\eta_1, \nabla_1, \eta_2, \nabla_2) + w (\eta_1, \nabla_1, \delta\eta_2, \nabla_2) \\ &= 2w(\delta\eta_1, \nabla_1, \eta_2, \nabla_2), \end{align*} \Erratum{err39a}%[** PP: Added closing square brace at end of footnote] or\footnote{[Note added, 1892. Better thus:---by \Secref{3} above, \[ -S\, \delta\psi\, \zeta\varpi\zeta = 2w (\delta\psi\,\zeta, \zeta, \eta_1, \nabla_1) = -2S\zeta_1\, \delta\psi\, \zeta w (\zeta_1, \zeta, \eta_1, \nabla_1) \] therefore by \Secref{4},\hfil $\varpi\omega = 2\zeta w(\zeta, \omega, \eta_1, \nabla_1)$\hfil \\ for $\zeta w(\zeta, \omega, \eta_1, \nabla_1)$ regarded as a function of $\omega$ is clearly self conjugate.]} %[** Closing brace] by \Secref{3} above, \[ -S\,\delta\eta_1\, \varpi\nabla_1 = 2w(\delta\eta_1, \nabla_1, \eta_2, \nabla_2). \] Now let us put $\delta\eta = \omega'S\omega\rho$ where $\omega'$ and~$\omega$ are arbitrary constant vectors. We thus get \[ S\omega'\, \varpi\omega = -2w(\omega', \omega, \eta_1, \nabla_1) = 2S\omega'\,\zeta w(\zeta, \omega, \eta_1, \nabla_1). \] Whence since $\omega'$ is quite arbitrary, \[ \EqnTag{III}{22} \varpi\omega = 2\zeta w(\zeta, \omega, \eta_1, \nabla_1). \] The statical problem can now be easily expressed. As we saw in \Secref{14}, equations \Eref{III}{7} and~\Eref{III}{8}, it is simply \[ \EqnTag{III}{22$a$} \mathbf{F} + V \nabla \mathbf{M}/2 + \varpi\Delta = 0, \] %% -----File: 058.png---Folio 40------- throughout the mass; and at the surface \[ \EqnTag{III}{22$b$} \textbf{F}_S - VU\, d\Sigma\, \textbf{M}/2 - \varpi U\, d\Sigma = 0, \] where $\mathbf{F}$, $\mathbf{M}$ are the given external force and couple per unit volume and $\mathbf{F}_S$ is the given external surface traction per unit surface. Substituting for~$\varpi$ from \equationref{III}{22} \[ \EqnTag{III}{23} \left. \begin{aligned} \mathbf{F} + V \nabla \mathbf{M}/2 + 2\zeta w(\zeta, \Delta, \eta_1, \nabla_1) = 0\phantom{.} \\ \mathbf{F}_S - V U\, d\Sigma\,\mathbf{M}/2 - 2\zeta w(\zeta, U\,d\Sigma, \eta_1, \nabla_1) = 0. \end{aligned} \right\} \] \Section{20}{Isotropic Bodies} \Paragraph{20} The simplest way to treat these bodies is to consider the (linear) relations between $\varpi$ and~$\psi$. In the first place notice that~$\psi$ can always be decomposed into three real elongations (contractions being of course considered as negative elongations). Thus $i$ being the unit vector in the direction of such an elongation, \[ \psi\omega = -\Sigma\, ei\, Si\,\omega. \] The elongation $-ei\,Si\,\omega$ will cause a stress symmetrical about the vector~$i$, i.e.\ a tension~$Ae$ in the direction of~$i$ and a pressure~$Be$ in all directions at right angles; $A$ and~$B$ being constants (on account of the linear relation between $\varpi$ and~$\psi$) independent of the direction of~$i$ (on account of the isotropy of the solid). This stress may otherwise be described as a tension $(A + B)e$ in the direction of~$i$ and a hydrostatic pressure~$Be$. Thus \begin{align*} \varpi\omega &= -(A+B)\Sigma\, ei\, Si\,\omega - B\omega\,\Sigma e \\ &= (A+B)\psi\omega + B\omega\, S\zeta\, \psi\zeta. \end{align*} To obtain the values of $A$ and~$B$ in terms of Thomson and Tait's coefficients $k$ and~$n$ of cubical expansion and rigidity respectively; first put \begin{align*} \psi\omega = e\omega\quad &\text{and}\quad \varpi\omega = 3k\, e\omega, \\ % \intertext{and then put} \psi\omega = V\lambda\omega\mu\quad &\text{and}\quad \varpi\omega = 2nV\lambda\omega\mu, \end{align*} $\lambda$ and~$\mu$ being any two vectors perpendicular to each other. We thus get \[ A + B = 2n,\qquad B = -(k-2n/3), \] whence \[ \EqnTag{III}{24} \varpi\omega = 2n\,\psi\omega - (k-2n/3)\omega S\, \zeta\psi\zeta. \] %% -----File: 059.png---Folio 41------- From this we have \[ \psi\omega = \varpi\omega/2n + \omega S\, \zeta\psi\zeta(k - 2n/3)/2n, \] but from the same equation \begin{gather*} S\,\zeta\varpi\zeta = 3 k\, S\, \zeta \psi \zeta; \\ \therefore\qquad \psi \omega = \frac{1}{2n} \varpi \omega + \left(\frac{1}{6n} - \frac{1}{9k}\right) \omega S\, \zeta\varpi\zeta. \EqnTag{III}{25} \end{gather*} \Equationref{III}{24} gives stress in terms of strain and~\Eref{III}{25} the converse. \Paragraph{21} We can now give the various useful forms of~$w$ for isotropic bodies for from \equationref{III}{19} \Secref{18}, \[ w = - S\psi\, \zeta\varpi\zeta/2. \] Therefore from equations \Eref{III}{24} and~\Eref{III}{25} respectively \begin{align*} \EqnTag{III}{26} w &= -n(\psi \zeta)^2 + \tfrac{1}{2}(k-2n/3) S\, \zeta_1\psi\zeta_1\, S\, \zeta_2\psi\zeta_2, \\ % \EqnTag{III}{27} w &= -\frac{1}{4n}(\varpi \zeta)^2 - \tfrac{1}{2}\left(\frac{1}{6n} - \frac{1}{9k}\right) S\, \zeta_1\varpi\zeta_1\, S\, \zeta_2\varpi\zeta_2. \end{align*} Therefore again from \Secref{3} above and from \equationref{III}{26}, \[ w = -n S\eta_1 \psi \nabla_1 + \tfrac{1}{2}(k - 2n/3)(S \nabla \eta)^2 \] or \[ \EqnTag{III}{28} \begin{split} 2w(\eta_1, \nabla_1, \eta_2, \nabla_2) = n S \nabla_1 \eta_2\, S \nabla_2 \eta_1 + n S \nabla_1 \nabla_2\, S \eta_1 \eta_2 \\ \shoveright{+ (k-2n/3) S \nabla_1 \eta_1 S \nabla_2 \eta_2.} \end{split} \] Hence from \equationref{III}{22} \[ \EqnTag{III}{29} \varpi\omega = - n S \omega \nabla\centerdot \eta - n \nabla_1 S \eta_1 \omega - (m-n) \omega S \nabla \eta, \] where $m$ is put for $k + n/3$. This last could have been deduced at once from \equationref{III}{24} by substituting for~$\psi \omega$. Thus the \hyperref[eqn:III.23]{equations~(23)} for the statical problem are \begin{align*} \EqnTag{III}{30} \mathbf{F} + V \nabla \mathbf{M}/2 &= n \nabla^2 \eta + m \nabla S \nabla \eta, \\ % \EqnTag{III}{31} -\mathbf{F}_S + V U\, d \Sigma\, \mathbf{M}/2 &= n S U\, d \Sigma\, \nabla \centerdot \eta + n \nabla_1 S \eta_1 U\, d \Sigma \\ &\quad + (m-n) U\, d \Sigma\, S \nabla \eta. \end{align*} We now proceed to apply these results for small strains in isotropic bodies to particular cases. These particular cases have all been worked out by the aid of Cartesian Geometry and they are given to illustrate the truth of the assertion made in the Introduction that the consideration of \emph{general problems} is made simpler by the use of Quaternions instead of the ordinary methods. %% -----File: 060.png---Folio 42------- \TCSection{22}{Particular integral of the equation of equilibrium} {Particular integral of equation {\upshape(30)}\protect\footnotemark} \footnotetext{[Note added, 1892. For a neater quaternion treatment of this problem see \textit{Phil.~Mag.}\ June,~1892, p.~493.]} \Paragraph{22} Since from \equationref{III}{30} ($\mathbf{F}$~being put for simplicity instead of $\mathbf{F} + V\nabla \mathbf{M}/2$) we have \[ n\nabla^2\eta = \mathbf{F} - m\nabla S\nabla\eta \] we obtain as a particular case by equations \Eref{II}{18} and~\Eref{II}{19} \Secref{10}, \[ 4\pi n\eta = \smalliiint u(\mathbf{F} - m\nabla S \nabla \eta)\, \ds, \] where $u$ has the meaning explained in \Secref{9}, and the volume integral extends over any portion (say the whole) of the body we may choose to consider. To express $\smalliiint u \nabla S \nabla \eta\, \ds$ as a function of~$\textbf{F}$ put in this term $u = -\frac{1}{2}U\rho_1\nabla_1$ where~$\rho$ is taken for the $\rho-\rho'$ of \Secref{9}, and apply \equationref{II}{9} \Secref{6}. Thus \begin{align*} 4\pi n \eta &= \smalliiint u \mathbf{F}\, \ds + \tfrac{1}{2} m \smalliiint U \rho_1 \nabla_1 \nabla S\nabla\eta\, \ds \\ &= \smalliiint u \mathbf{F}\, \ds - \tfrac{1}{2} m \smalliiint U \rho \nabla^2 S \nabla \eta \, \ds + \text{ a surf.\ int.} \\ &= \smalliiint u \mathbf{F}\, \ds - \dfrac{m}{2(m+n)} \smalliiint U \rho S \nabla \mathbf{F}\, \ds + \text{ the surf.\ int.} \end{align*} for by \equationref{III}{30} $S\nabla \mathbf{F} = (m + n)\nabla^2 S\nabla \eta$. Now (in order to get rid of any infinite terms due to any discontinuity in~$\mathbf{F}$) apply \equationref{II}{9} \Secref{6} to the second volume integral. Thus \[ 4\pi n \eta = \smalliiint u \mathbf{F}\, \ds + \frac{m}{2(m+n)} \smalliiint S \mathbf{F} \nabla \centerdot U \rho\, \ds + \text{ a surf.\ int.} \] The surface integral may be neglected as we may thus verify. Call the volume integral $4\pi n \eta'$. Thus \begin{gather*} 4\pi n \nabla^2 \eta' = 4 \pi \mathbf{F} + \dfrac{m}{2(m+n)} \smalliiint S\mathbf{F} \nabla \centerdot \nabla^2 U \rho\, \ds \\ 4\pi m \nabla S \nabla \eta' = \dfrac{m}{n} \smalliiint S \mathbf{F} \nabla \centerdot \nabla u\, \ds + \dfrac{m^2}{2n(m+n)}\smalliiint S \mathbf{F} \nabla \centerdot \nabla S \nabla U \rho\, \ds, \end{gather*} so that putting $\nabla U \rho = - 2u$ we get \[ n \nabla^2 \eta' + m \nabla S \nabla \eta' \equiv \mathbf{F}, \] whence we have as a particular solution of \equationref{III}{30} $\eta = \eta'$ or \[ \EqnTag{III}{32} \eta = \dfrac{1}{4 \pi n} \smalliiint u \mathbf{F}\, \ds + \dfrac{m}{8\pi n (m+n)} \smalliiint S \mathbf{F} \nabla \centerdot U \rho\, \ds. \] %% -----File: 061.png---Folio 43------- This is generally regarded as a solution of the statical problem for an infinite isotropic body. In this case some law of convergence must apply to~$\mathbf{F}$ to make these integrals convergent. Thomson and Tait (\textit{Nat.\ Phil.}\ \S~730) say that this law is that~$\mathbf{F}r$ converges to zero at infinity. This I think can be disproved by an example. Put\footnote{% [Note added, 1892. This is not legitimate since it makes $\mathbf{F} = \infty$ for $r = 0$. The reasoning is rectified in the \textit{Phil.\ Mag.}\ paper just referred to by putting $\mathbf{F} = 0$ from $r = 0$ to $r = b$ and $\mathbf{F}r = r^{-a}\lambda$ from $r = b$ to $r = \infty$.]} $\mathbf{F}r = r^{-a}\lambda$ where~$\lambda$ is a constant vector and~$a$ a positive constant less than unity. \Equationref{III}{32} then gives for the displacement at the origin due to the part of the integral extending throughout a sphere whose centre is the origin and radius~$R$ \[ \eta = \dfrac{m+3n}{3n(m+n)} \dfrac{R^{1-a}}{1-a} \lambda. \] Putting $R = \infty$, $\eta$ also becomes~$\infty$. The real law of convergence does not seem to me to be worth seeking as the practical utility of \equationref{III}{32} is owing to the fact that it is a particular integral. The present solution of the problem has only to be compared with the one in Thomson and Tait's \textit{Nat.\ Phil.}\ \S\S~730--1 to see the immense advantage to be derived from Quaternions. It is easy to put our result in the form given by them. We have merely to express $S\mathbf{F}\nabla \centerdot U\rho$ in terms of $\mathbf{F}$ and~$r^2 S\mathbf{F}\nabla \centerdot \nabla u$ where~$r$ is put for the reciprocal of~$u$. Noting that \[ \nabla u = -u^3\rho,\quad U\rho = u\rho,\quad S\mathbf{F}\nabla \centerdot \rho = -\mathbf{F}, \] we have at once \begin{align*} S\mathbf{F}\nabla \centerdot U\rho &= -u \mathbf{F} - u^3\rho S\mathbf{F}\rho \\ S\mathbf{F}\nabla \centerdot \nabla u &= u^3\mathbf{F} + 3u^5\rho S\mathbf{F}\rho \end{align*} therefore eliminating $\rho S\mathbf{F}\rho$, \begin{gather*} S\mathbf{F}\nabla \centerdot U\rho = -u\mathbf{F}2/3 - r^2 S\mathbf{F}\nabla \centerdot \nabla u/3, \\ %[** PP: Not pushing to left margin] \therefore\qquad \eta = \{ 24\pi n(m+n) \}^{-1} \smalliiint \ds\, \{ 2(2m+3n)u\mathbf{F} - mr^2 S\mathbf{F}\nabla \centerdot \nabla u \}, \EqnTag{III}{33} \end{gather*} which is the required form. \Paragraph{23} Calling the particular solution~$\eta'$ as before and putting \[ \eta = \eta' + \eta'' \] the statical problem is reduced to finding $\eta''$ to satisfy \[ n\nabla^2 \eta'' + m\nabla S\nabla \eta'' = 0 \] %% -----File: 062.png---Folio 44------- and the surface equation either \begin{flalign*} && \eta' + \eta'' &= \text{ given value}, && \\ &\text{i.e. }& \eta'' &= \text{ given value,} && \\ &\text{or } & \varpi U d \Sigma &=\text{ given surface traction,} && \end{flalign*} i.e.\ by \equationref{III}{29} \Secref{21} above, \[ nSU\, d\Sigma \nabla\centerdot \eta'' + n \nabla_1 S \eta_1'' U\, d \Sigma + (m-n) U\, d\Sigma S\nabla \eta'' = \text{ known value.} \] This general problem for the spherical shell, the only case hitherto solved, I do not propose to work out by Quaternions, as the \emph{methods} adopted are the same as those used by Thomson and Tait in the same problem. But though each step of the Cartesian proof would be represented in the Quaternion, the saving in mental labour which is effected by using the peculiarly happy notation of Quaternions can only be appreciated by him who has worked the whole problem in both notations. The only remark necessary to make is that we may just as easily use vector, surface or solid, harmonics or indeed quaternion harmonics as ordinary scalar harmonics. \Section{24}{Orthogonal coordinates} \Paragraph{24} It is usual to find what equations~\Eref{III}{22$a$} of \Secref{19} and~\Eref{III}{3}~of \Secref{12} become when expressed in terms of any orthogonal coordinates. This can be done much more easily by Quaternions than Cartesian Geometry. Compare the following investigation with the corresponding one in Ibbetson's \textit{Math.\ Theory of Elasticity}, Chap.~V. Let $x$, $y$, $z$ be any orthogonal coordinates, i.e.\ let $x = \text{const.}$, $y = \text{const.}$, $z = \text{const.}$, represent three families of surfaces cutting everywhere at right angles. Particular cases are of course the ordinary Cartesian coordinates, the spherical coordinates $r$, $\theta$, $\phi$ and the cylindrical coordinates $r$, $\phi$, $z$. Let $D_x$, $D_y$, $D_z$ stand for differentiations \emph{per unit length} perpendicular to the three coordinate surfaces and let $\lambda$, $\mu$, $\nu$ be the unit vectors in the corresponding directions. Thus \[ \nabla = \lambda D_x + \mu D_y + \nu D_z. \] Thus, using the same system of suffixes for the $D$'s as was explained in connection with~$\nabla$ in \Secref{1}, \[ \EqnTag{III}{34} \phi \Delta = D_{x1}\phi_1\lambda + D_{y1}\phi_1\mu + D_{z1}\phi_1\nu, \] or \[ \EqnTag{III}{35} \phi \Delta = D_x(\phi \lambda) + D_y(\phi \mu) + D_z(\phi \nu) - \phi(D_x\lambda + D_y\mu + D_z\nu). \] %% -----File: 063.png---Folio 45------- \Paragraph{25} Now to put \equationref{III}{22$a$} \Secref{19} into the present coordinates all that is required is to express~$\varpi\Delta$ in terms of those coordinates. Let the coordinates of~$\varpi$ be~$PQRSTU$. Thus from \equationref{III}{35} we have \[ \begin{split} \varpi\Delta &= D_x(P\lambda + U\mu + T\nu) + D_y(U\lambda + Q\mu + S\nu) + D_z(T\lambda + S\mu + R\nu) \\ &\qquad\qquad - \varpi(D_x\lambda + D_y\mu + D_z\nu). \end{split} \] The first thing then is to find $D_x\lambda$, $D_x\mu$, $D_y\lambda$~\&c. Let $p_2$, $p_3$ be the principal curvatures normal to $x=\text{const.}$, i.e.\ (by a well-known property of orthogonal surfaces) the curvatures along the lines of intersection of $x=\text{const.}$, with $z=\text{const.}$, and $y=\text{const.}$\ $p_2$, $p_3$ will be considered positive when\footnote{% [Note added, 1892. This is contrary to the usual convention.]} the positive value of~$dx$ is on the convex side of the corresponding curvatures. Similarly for $q_3q_1r_1r_2$. Thus for the coordinates $r$, $\theta$, $\phi$; \[ p_2 = p_3 = 1/r,\quad q_3 = \cot\theta/r,\quad q_1 = r_1 = r_2 = 0. \] Again for $r$, $\phi$, $z$; $p_2 = 1/r$ and the rest are each zero. With these definitions we see geometrically that \[ \EqnTag{III}{36} D_x\lambda = -\mu q_1 - \nu r_1, \quad D_x\mu = \lambda q_1, \quad D_x\nu = \lambda r_1. \] Similarly for $D_y\lambda$, $D_y\mu$, $D_z\lambda$~\&c. Thus \begin{align*} \varpi\Delta &= \lambda(D_x P + D_y U + D_z T) + \mu(\;) + \nu(\;) \\ &+ \lambda(-Qp_2 - Rp_3 + Tr_1 + Uq_1) + \mu(\;) + \nu(\;) \\ &+ \lambda\{ (p_2+p_3)P + (q_3+q_1)U + (r_1+r_2)T \} + \mu\{\;\} + \nu\{\;\}, \end{align*} or \begin{align*} \varpi\Delta &= \lambda\{D_x P + D_y U + D_z T + P(p_2+p_3) - Qp_2 - Rp_3 \notag \\ \EqnTag{III}{37} &+ T(2r_1 + r_2) + U(q_3 + 2q_1)\} + \mu\{\;\} + \nu\{\;\}. \end{align*} \Paragraph{26} The other chiefly useful thing in transformation of coordinates in the present subject is the expression for the strain function~$\psi$ in terms of the coordinates of displacement. Let $u$, $v$, $w$ be these coordinates. Now by \equationref{III}{3} \Secref{12}, remembering (\Secref{18}) that $\psi = \overline{\chi} - 1$ we have \[ -2\psi\omega = S\omega\nabla \centerdot \eta + \nabla_1 S\omega\eta_1, \] \begin{flalign*} &\text{whence} & -2\psi\lambda = \lambda S\lambda D_x\eta + \mu S\lambda D_y\eta + \nu S\lambda D_z\eta - D_x\eta. && \end{flalign*} \begin{flalign*} &\text{But} & D_x\eta &= \lambda D_x u + \mu D_x v + \nu D_x w + uD_x\lambda + vD_x\mu + wD_x\nu && \\ && &= \lambda(D_x u + vq_1 + wr_1) + \mu(D_x v - uq_1) + \nu(D_x w - ur_1). && \end{flalign*} %% -----File: 064.png---Folio 46------- Similarly \begin{align*} D_y\eta &= \lambda(D_y u-vp_2) + \mu(D_y v+wr_2+up_2) + \nu(D_y w-vr_2) \\ D_z\eta &= \lambda(D_z u-wp_3) + \mu(D_z v-wq_3) + \nu(D_z w+up_3+vq_3) \\ \therefore\qquad 2\psi\lambda &= 2\lambda(D_x u + v q_1 + wr_1) + \mu (D_y u + D_x v - uq_1 - vp_2) \\ % \EqnTag{III}{38} & \phantom{{}=2\lambda(D_x u + v q_1 + wr_1)} + \nu (D_z u + D_x w - ur_1 - wp_3). \end{align*} But with Thomson and Tait's notation for pure small strain \[ 2\psi\lambda = 2\lambda e + \mu c + \nu b, \] \[ \EqnTag{III}{39} \left. \begin{aligned} \therefore\qquad \footnotemark e &= D_xu + vq_1 + wr_1 \\ f &= D_yv + wr_2 + up_2 \\ g &= D_zw + up_3 + vq_3 \\ a &= D_yw + D_zv - wq_3 - vr_2 \\ b &= D_zu + D_xw - ur_1 - wp_3 \\ c &= D_xv + D_yu - vp_2 - uq_1. \end{aligned} \right\} \] \footnotetext{% [Note added, 1892. In the \textit{Phil.~Mag.}\ June, 1892, p.~488, there is a mistake in the equation just preceding equation~31 and there are two mistakes in equation~31. In the first of these $I(2D \Pre{\xi}{u} - \Pre{\eta}{\varpi} \Pre{\zeta}{v} - \Pre{\zeta}{\varpi} \Pre{\eta}{w})$ should be $2I(D \Pre{\xi}{u} - \Pre{\eta}{\varpi} \Pre{\zeta}{v} - \Pre{\zeta}{\varpi} \Pre{\eta}{w})$. In \equationref{III}{31} all the 2's should be dropped.]} Thus we have $efgabc$ in terms of the displacement and we have already in \equationref{III}{24} \Secref{20}, which expresses~$\varpi$ in terms of~$\psi$, found the values of $PQRSTU$ in terms of $e$,~\&c. Finally the expression in \equationref{III}{37} for $\varpi\Delta$ gives us the equations of equilibrium in terms of~$P$, \&c. Thus we have all the materials for considering any problem with the coordinates we have chosen. All these results can be at once applied to spherical and cylindrical coordinates, but as this has nothing to do with our present purpose---the exemplification of Quaternion methods---we leave the matter here. Let us as an example of particular coordinates to which this section forms a suitable introduction consider St Venant's Torsion Problem by means of cylindrical coordinates. \TCSection{27}{Saint-Venant's torsion problem}{Saint-Venant's Torsion Problem} \Paragraph{27} In this problem we consider the equilibrium of a cylinder with any given cross-section, subjected to end-couples, but to no bodily forces and no stress on the curved surface. %% -----File: 065.png---Folio 47------- We shall take $r, \phi, z$ as our coordinates, the axis of~$z$ being parallel to the generating lines of the cylinder. Let $\lambda, \mu, \nu$ be the unit vectors in the directions of $dr, d\phi, dz$ respectively and let \[ \eta = u\lambda + v\mu +w \nu \] as before. We shall follow Thomson and Tait's lines of proof---i.e.\ we shall first find the effect of a simple torsion and then add another displacement and so try to get rid of stress on the curved surface. Holding the section $z = 0$ fixed let us give the cylinder a small torsion of magnitude~$\tau$, i.e.\ let us put \[ \EqnTag{III}{40} \eta = \tau z r \mu, \] for all points for which $\tau z$ is small. The practical manipulation of such expressions as this is almost always facilitated by considering the general value of $Q(\nabla_1, \eta_1)$ where~$Q$ is any function linear in each of its constituents. Thus in the present case \[ Q (\nabla_1, \eta_1) = \tau \bigl\{ zQ (\lambda, \mu) - zQ(\mu, \lambda) + rQ(\nu,\mu)\bigr\}. \] [If~$Q$ is symmetrical in its constituents, e.g.\ in the case of stress below this reduces to the simple form $Q(\nabla_1, \eta_1) = \tau r Q(\nu, \mu)$.] From this we at once see that~$\eta$ satisfies the equation of internal equilibrium \[ n \nabla^2 \eta + m \nabla S \nabla \eta = 0, \] for putting $Q(\alpha, \beta) = \alpha\beta$ \[ \nabla\eta = \tau(2 z \nu - r\lambda ) = \tau \nabla(z^2-r^2/2) \] so that both $S\nabla \eta$ and~$\nabla^2\eta = 0$. Again the value for~$Q$ at once gives us the stress for \begin{flalign*} && \varpi\omega = - n S\omega \nabla & \centerdot\eta - n \nabla_1 S \omega \eta_1 - (m-n)\omega S \nabla \eta, && \\ % \EqnTag{III}{41} &\text{or}& \varpi\omega &= - n\tau r (\mu S \omega \nu +\nu S\omega \mu), && \end{flalign*} which is a shearing stress $n\tau r$ on the interfaces perpendicular to $\mu$ and~$\nu$. Putting $\omega = {}$ the unit normal of the curved surface we have for the surface traction \[ \varpi\omega = - n\tau r \nu\, S\omega \mu. \] In the figure let the plane of the paper be $z = 0$, $O$ the origin, $P$~a point on the curved surface and~$OM$ the perpendicular from~$O$ %% -----File: 066.png---Folio 48------- on the tangent at~$P$. Thus $-rS\omega\mu = OP \cos OPM = PM$, $PM$ being reckoned positive or negative according as it is in the positive or negative direction of rotation round~$Oz$. Thus we see that the surface traction is parallel to~$Oz$ and~$= n \tau PM$. %[Illustration] \begin{center} \includegraphics{./images/cylinder.pdf} \end{center} Hence in the case of a circular cylinder a torsion round the axis satisfies all the conditions of our original problem, but this is true in no other case. The surface traction at any point on the plane ends necessary to produce this strain is $\varpi \nu = n \tau r\mu$ by \equationref{III}{41} so that its moment round the origin is $n \tau \smalliint r^2\, dA$, where~$dA$ is an element of area and the integral extends over the whole cross-section. \Paragraph{28} Let us now assume a further displacement \[ \EqnTag{III}{42} \eta = w\nu, \] where $w$ is a function of $r, \phi$~only, and let us try to determine~$w$ so that there is still internal equilibrium and so that the stress on the curved surface due to~$w$ shall neutralise the surface traction already considered. In the present case \[ Q(\nabla_1,\eta_1) = Q(\nabla w, \nu). \] Thus $S \nabla \eta = 0$ (since~$\nu$ is perpendicular to~$\nabla w$) and therefore the equation of internal equilibrium gives \[ \nabla^2 w = 0. \] \begin{flalign*} &\text{Again}& \varpi \omega &= -n S \omega \nabla \centerdot \eta - n \nabla_1 S \omega \eta_1, && \\ % % [** N.B. No Equations 44 or 45 in this section] \EqnTag{III}{43} &\text{or}& \varpi \omega &= -n (\nu S \omega \nabla w + \nabla w S \omega \nu), && \end{flalign*} a shear $= nT\nabla w$ on the interfaces perpendicular to $\nabla w$ and~$\nu$. %% -----File: 067.png---Folio 49------- Thus putting~$\omega$ for the unit normal to the curved surface, the % [** PP: Keeping ``neutralize''] present surface traction will neutralize the former if \begin{flalign*} && S \omega \nabla w &= - \tau r S \omega \mu = -\tau S \omega \nu (\lambda r) && \\ && &= \tfrac{1}{2} \tau S \nu \omega \nabla (r^2), && \\ &\text{i.e.}& dw &/ dn = d (\tau r^2 /2) / ds, && \end{flalign*} where $d/dn$ represents differentiation along the normal outwards and~$d/ds$ differentiation along the positive direction of the bounding curve. We leave the problem here to the theory of complex variables and Fourier's Theorem. % [** PP: Changing opening ``='' to ``is equal to'' twice; % was ``...on the plane end ${} = \varpi \nu...'' and % ``...total couple ${} = -n\smallint ...$'', which both reads % awkwardly and breaks badly] Observe however that the surface traction at any point on the plane end is equal to\Erratum{err49a} $\varpi \nu = n \nabla w$ by \equationref{III}{43}, and therefore that the total couple is equal to $- n \smalliint r S \mu \nabla w\, dA = n \smalliint (dw/d\phi)\, dA$. This leads to the usual expression for torsional rigidity. \Section{29}{Wires} \Paragraph{29} In the following general treatment of Wires some of the processes are merely Thomson and Tait's translated into their shorter Quaternion forms; others are quite different. The two will be easily distinguished by such as are acquainted with Thomson and Tait's \emph{Nat.\ Phil.} The one thing to be specially careful about is the notation and its exact meaning. This meaning we give at the outset. The wires we consider are not necessarily naturally straight but we assume some definite straight condition of the wire as the ``geometrically normal'' condition. The variable in terms of which we wish to express everything is~$s$ the distance along the wire from some definite point on it. Any element of the wire, since it is only \emph{slightly} strained, may be assumed to have turned as a rigid body from its geometrically normal position. This rotation is expressed as usual (Tait's \emph{Quaternions}, \S~354) by the quaternion~$q$; the axis of~$q$ being the axis of rotation, and the angle of~$q$, half the angle turned through. $\omega$ is taken so that the rate of this turning per unit length of the wire is $q \omega q^{-1}$ so that~$\omega$ is the rate of turning per unit of length when the whole wire is moved as a rigid body so as to bring the %% -----File: 068.png---Folio 50------- element under consideration back to its geometrically normal position. Of course~$\omega$ is a function of~$q$ and its derivative with reference to~$s$. This function we shall investigate later. The resolved part of~$q \omega q^{-1}$ parallel to the wire is the vector twist and the resolved part perpendicular to the wire is in the direction of the binormal and equal to the curvature. In fact~$\omega$ is the vector whose coordinates are the $\kappa, \lambda, \tau$ of Thomson and Tait's \emph{Nat.\ Phil.}\ \S~593. When we are given $q$ or~$\omega$ for every point we know the strain of the wire completely. $\omega_0$~is defined as the naturally normal value of~$\omega$, i.e.\ the value of~$\omega$ when the wire is unstressed. As usual we take~$\rho$ as the coordinate vector of any point of the wire, $\rho$ like the rest of the functions being considered as a function of~$s$. We shall denote (after Tait, \emph{Quaternions}, Chap.~\textsc{ix}.)\ differentiations with regard to~$s$ by dashes. We now come to the dynamical symbols. $\mathbf{F}$ and~$\mathbf{M}$ are the force and couple respectively exerted across any normal section of the wire on the part of the wire which is on the negative side of the section by the part on the positive side. Finally let $\mathbf{X, L}$ be the external force and couple per unit length exerted upon the wire. \Paragraph{30} When the wire is strained in any way let us impose a small additional strain represented by an increment~$\delta \omega$ in~$\omega$ and an increment~$\delta e$ in the elongation at any point. Then the work done on the element~$ds$ by the stress-force $= -\delta e S \mathbf{F} \rho'\,ds$ and that done by the stress-couple $= -Sq \delta \omega q^{-1} \mathbf{M}\, ds$. If (as we assume, though the assumption is not justified in some useful applications of the general theory of wires) $\mathbf{F}$ and~$\mathbf{M}$ to be of the same order of magnitude the former of these expressions can be neglected in comparison with the latter for~$\delta e$ is a quantity small compared with~$\delta\omega$. Now the work done on the element by the stress $=$ the increment of the element's potential energy $= \delta w ds$ where~$w$ is some function of the strain. Hence \begin{align*} \delta w &= -S\, \delta \omega\, q^{-1} \mathbf{M} q. \\ \intertext{Thus $w$ is a function of~$\omega$ only and} \therefore\qquad \delta w &= -S\, \delta\omega\, \Nabla{\omega} w, \\ % \intertext{whence we see that} \EqnTag{III}{46} q^{-1} \mathbf{M} q &= \Nabla{\omega} w. \end{align*} Notice that $\mathbf{M}$ is Thomson and Tait's $\xi$, $\eta$, $\zeta$ and~$q^{-1} \mathbf{M} q$ their~$KLM$ (\emph{Nat.\ Phil}.\ \S\S~594,~614). %% -----File: 069.png---Folio 51------- \Paragraph{31} Now since the strain is small, $q^{-1}\mathbf{M}q$ is linear in terms of the strain and therefore in terms of $\omega$. Hence we see that~$w$ is quadratic in terms of $\omega$. Let us then put \begin{align*} w = w_2(\omega - \omega_0, \omega-\omega_0) &+ w_1 (\omega-\omega_0) \\ &+ \text{ a function of temperature only,} \end{align*} $w_2$ and~$w_1$ being linear and homogeneous in each of their constituents. This is the most general quadratic function of~$\omega$. Now \[ \Nabla{\omega} w = \zeta w_2(\zeta, \omega - \omega_0) + \zeta w_2(\omega - \omega_0, \zeta) + \zeta w_1 \zeta. \] Putting then $\omega=\omega_0$ and $\Nabla{\omega} w=0$ we get $\zeta w_1\zeta=0$. Operating on the last by $S\sigma(\;)$ where~$\sigma$ is any vector we see that $w_1=0$. Thus putting \[ \Nabla{\omega} w = \phi(\omega - \omega_0), \] where $\phi$ has the value given by the last equation and is therefore self-conjugate we get the two following equations \[ \EqnTag{III}{47} w = -S(\omega - \omega_0) \phi(\omega - \omega_0)/2 + w_0, \] [as can be seen by a comparison of the last three equations $w_0$ being the function of the temperature] and \[ \EqnTag{III}{48} q^{-1}\mathbf{M}q = \phi(\omega-\omega_0). \] When the natural shape of the wire is straight these become \begin{gather*} \EqnTag{III}{49} w = -S \omega\phi\omega/2 + w_0, \\ % \EqnTag{III}{50} q^{-1}\mathbf{M}q = \phi\omega, \end{gather*} and when further the wire is truly uniform $\phi$ and~$w_0$ are constant along the wire. \Paragraph{32} Assuming the truth of these restrictions let us conceive a rigid body moving about a fixed point which, when placed in a certain position which we shall call the normal position, has, if then rotating with any vector angular velocity~$\omega$, a moment of momentum $= \phi\omega$ where~$\phi$ has the meaning just given. If the rigid body be made to take a finite rotation $q(\;)q^{-1}$ and then to move with angular velocity~$q\omega q^{-1}$ its moment of momentum will be~$q\phi\omega q^{-1}$. Now let a point move along the wire with unit velocity and let the rigid body so move in unison with it that when the moving point reaches the point~$s$ the rigid body shall have made the rotation represented by~$q$ (\Secref{29} above). Thus by the definition of $\omega$ and~$q$ its angular velocity at any instant is~$q\omega q^{-1}$ and its moment of momentum therefore $q\phi\omega q^{-1}$ or~$\mathbf{M}$. %% -----File: 070.png---Folio 52------- Now consider the equilibrium of the wire when no external force or couple acts except at its ends. In this case $\mathbf{F}$~is constant throughout and it is easy to see (what indeed is a particular case of \equationref{III}{53} below) that \[ \EqnTag{III}{51} \mathbf{M}' + V \rho' \mathbf{F} = 0. \] Interpreting this equation for our rigid body we get as the law which governs its motion \[ d (\text{vect.\ mom.\ of mom.})/dt = -V \rho' \mathbf{F}. \] Thus the rigid body will move as if acted upon by a constant force~$\mathbf{F}$ at the end of the unit vector~$\rho'$ or---since this vector is fixed in the body---as if acted upon by a constant force acting through a point fixed in the body. From this kinetic analogue of Kirchhoff's the mathematical problem of the shape of such a wire as we are now considering, under the given circumstances, is shewn to be identical with the general problem of the pendulum of which the top is a variety. \Paragraph{33} We will now give the general equations for any wire under any external actions. The comparison of the Quaternion treatment of this with the Cartesian as given in Thomson and Tait's \emph{Nat.\ Phil}.\ \S~614 seems to me to be all in favour of the former. The equations of equilibrium of an element~$ds$ are with the notation explained in \Secref{29} above \begin{gather*} d\mathbf{F} + \mathbf{X}ds = 0, \\ Vd\rho \mathbf{F} + d\mathbf{M} + \mathbf{L}\, ds = 0, \end{gather*} or dividing by~$ds$ \begin{gather*} \EqnTag{III}{52} \mathbf{F}' + \mathbf{X} = 0, \\ % \EqnTag{III}{53} V\rho' \mathbf{F} + \mathbf{M}' + \mathbf{L} = 0. \end{gather*} Operating on the last equation by $V\rho'(\;)$ noting that $\rho'^2 = -1$ and %%% putting $S\rho' \mathbf{F} = -T$ we get \[ \EqnTag{III}{54} \mathbf{F} = \rho' T+ V\rho' (\mathbf{M}' + \mathbf{L}), \] whence by equations \Eref{III}{52} and~\Eref{III}{53} respectively \begin{gather*} \EqnTag{III}{55} \mathbf{X} + d\{\rho' T+ V \rho' (\mathbf{M}' + \mathbf{L})\}/ds = 0, \\ % \EqnTag{III}{56} S\rho' (\mathbf{M}'+ \mathbf{L}) = 0. \end{gather*} %% -----File: 071.png---Folio 53------- Now by equation~(48) above % [** PP: No hyperlink; equation repeated here w/manual \tag] \[ \tag{48}%%* [Sic; it's a copy of 48, not a misnumbered 56.] q^{-1}\mathbf{M}q = \phi(\omega-\omega_0). \] Also by the definition of~$q$ \[ \EqnTag{III}{57} \rho' = q\lambda q^{-1}, \] where $\lambda$ is some given constant unit vector. Finally as we are about to prove \[ \EqnTag{III}{58} \omega = 2V q^{-1}q'. \] It is usual in the Cartesian treatment to leave the problem in the form of equations equivalent to the above \Eref{III}{55} to~\Eref{III}{58}, 13~scalar equations for the 13~unknown scalars of $\rho$, $T$, $\mathbf{M}$, $q$ and~$\omega$. We can however as we shall directly in the Quaternion treatment quite easily reduce the general problem to one vector and one scalar equation involving the four unknown scalars of~$T$ and~$q$ in terms of which all the other unknowns are explicitly given. To prove \equationref{III}{58}\footnote{% This could be deduced from Tait's \textit{Quaternions}, \S~356, equation~(2). His $\epsilon$ is our~$q\omega q^{-1}$ and his dots our dashes.} observe that \[ (q + dq)\sigma(q + dq)^{-1} = q(\sigma + V\omega \sigma\, ds) q^{-1}, \] where $\sigma$ is any vector. The truth of this is seen by noticing that $(q + dq)(\;)(q + dq)^{-1}$ is the operator that rotates any vector of the element $s + ds$ from its geometrically normal position to its strained position. But we can also get to this final position by first in the geometrically normal wire making the small strain $\omega\, ds$ at the given element and then performing the strain of the wire up to the point~$s$. The first process is represented on the left of the last equation and the second on the right. Thus we get \begin{flalign*} && q^{-1}\,dq\, \sigma &+ \sigma d(q^{-1})\centerdot q = V\omega\sigma\, ds, && \\ &\text{or } \because & d(q^{-1}) &= -q^{-1}\, dq\, q^{-1}, &&\\ && q^{-1}q'\sigma &- \sigma q^{-1}q' = V \omega \sigma, && \\ &\text{whence}& \omega &= 2V\, q^{-1}q'. && \end{flalign*} Returning to equations \Eref{III}{55} to~\Eref{III}{58} observe that equations \Eref{III}{57} and~\Eref{III}{58} give $\rho'$ and~$\omega$ as explicit functions of~$q$. Hence by \equationref{III}{48} \[ \EqnTag{III}{59} \mathbf{M} = q \phi (2V\, q^{-1} q' - \omega_0) q^{-1}, \] %% -----File: 072.png---Folio 54------- which gives $\mathbf{M}$ also explicitly. Substituting for $\mathbf{M}$ and~$\rho'$ in equations \Eref{III}{55} and \Eref{III}{56} we have \begin{gather*} \EqnTag{III}{60} \mathbf{X} + \frac{d}{ds}\left\{ q\lambda q^{-1}\, T + V\, q\lambda q^{-1} \Bigl(\mathbf{L} + \frac{d}{ds}\Bigl[ q\, \phi(2V\, q^{-1}q' - \omega_0)\, q^{-1}\Bigr] \Bigr) \right\} = 0, \\ % \EqnTag{III}{61} S\, q\lambda q^{-1} \Bigl(\mathbf{L} + \frac{d}{ds}\Bigl[ q\, \phi(2V\, q^{-1}q' - \omega_0)\, q^{-1}\Bigr] \Bigr) = 0, \end{gather*} which are sufficient equations to determine $T$ and~$q$, whereupon $\mathbf{M}$ is given by \equationref{III}{59}, $\omega$~by \equationref{III}{58} and $\rho'$~and therefore also~$\rho$ by~\equationref{III}{57}. Spiral springs can be treated very simply by means of the above equations, but we have already devoted sufficient space to this subject. %% -----File: 073.png---Folio 55------- \Chapter{Section IV}{Electricity and Magnetism} \TCSection{34}{% \texorpdfstring{{\scshape Electrostatics}---general problem}{Electrostatics} }{{\scshape Electrostatics}\\[2ex] General Problem} \Paragraph{34} Merely observing that all the theorems in integration given in the Preliminary and~\textsc{iv}th chapters of Maxwell's treatise on \textit{Electricity and Magnetism}, Part~\textsc{i}., are easy particular cases of equations \Eref{II}{8} and~\Eref{II}{9} \Secref{6} above, we pass on to the one application of Quaternions that we propose to make in Electrostatics. This is to find the most general mechanical results arising from Maxwell's theory of Electrostatics, and to see if they can be explained by stress in the dielectric. This problem as far as I am aware has not hitherto in all its generality been attacked though the most important practical cases have been, as we shall see, considered by Maxwell, Helmholtz, Korteweg, Lorberg and Kirchhoff. It is necessary first of all to indicate as clearly as possible what I take to be Maxwell's theory of Electricity. He assumes\footnote{% Prof.\ J.~J.\ Thomson in his paper on \textit{Electrical Theories}, B.A.\ Reports, 1885, p.~125, does not credit Maxwell with such a definite and circumscribed theory as that described in the text, and he is thereby led to find fault with Maxwell's term ``Displacement'' and points out that there is an assumption made with reference to the connection between the true current and this polarisation (displacement). He says moreover, ``It is rather difficult to see what is meant in Maxwell's Theory by the phrase `Quantity of Electricity.'\,'' None of these remarks are called for if the view I take of Maxwell's theory be correct, and these grounds alone I consider sufficient for taking that view. The paper of Thomson's here mentioned I shall frequently have to refer to. [Note added, 1892. In the text I have given much too rigid a form to Maxwell's theory. What I have called his theory I ought rather to have called his analogy. Still I think the present foot-note is in the main just. In my opinion it is no more and no less difficult to see what is meant in Maxwell's Theory by ``Quantity of Electricity'' than by ``displacement'' since the two are connected by perfectly definite equations. Of course it is wrong to define ``displacement'' as ``displacement of quantity of electricity'' and then to define ``quantity of electricity'' in terms of ``displacement'' but Maxwell does not seem to me even tacitly to do this. Rather he says---the dielectric is polarised; this polarisation can be represented by a vector~$\mathbf{D}$; electrical quantity can be expressed in terms of~$\mathbf{D}$; the mathematical connections between electrical quantity and~$\mathbf{D}$ are the same as those between quantity of matter in a space and the displacement out of that space made by other matter to make room for the given matter; we will impress this useful analogy firmly on our minds by calling~$\mathbf{D}$ the displacement. But I have expressed my present views on the meaning of Maxwell's theory much more fully in \textit{Phil.\ Trans.}\ 1892, p.~685.]} %** end of footnote all space to be uniformly filled with a certain substance called Electricity. Whatever electrical actions take place %% -----File: 074.png---Folio 56------- depend on the continued or past \textit{motion} of this substance as an incompressible fluid. If electricity is brought from a distance by any means and placed in a given space there must be a displacement of the original electricity outwards from that space and the quantity of foreign electricity is conveniently measured by the surface integral of that displacement. \textit{Dielectrics} are substances in which this displacement tends to undo itself, so to speak, i.e.\ the original electricity tends to go back to its primitive position. In \textit{conductors}, on the other hand, there is no property distinguishing any imported electricity from the original electricity. The rate of variation of displacement, whether in dielectrics or conductors, of course constitutes an electric current as it is conveniently called. We have next to consider a vector at each point of space called the electro-motive force, which depends in some way at present undefined on the distribution of the displacement in the dielectrics, the distribution of currents whether in dielectrics or conductors, and on extra-electrical or semi-electrical action, e.g.\ chemical or mechanical. If at any point the electro-motive force be multiplied by a scalar the medium at the point remaining (except electrically) unchanged, the current in the case of conductors and the displacement in the case of dielectrics is altered in the same ratio. In other words the current or the displacement, as the case may be, is a linear vector function of the electro-motive force, and the coordinates of the linear vector function\footnote{% This frequently recurring cumbrous mode of description must be tolerated unless a single word can be invented for ``a linear vector function of a vector.'' Might I suggest the term ``Hamiltonian?'' Thus we should say that the displacement is a Hamiltonian of the electro-motive force, the Hamiltonian at any point being a function of the state of the medium.} at any point depend %% -----File: 075.png---Folio 57------- solely on the state of the medium (whether fluid, solid,~\&c.,\ or again strained or not) at that point. To complete the theory we have to explain how the part of the electro-motive force which is a function of the distribution of displacement and current depends on this distribution. This explanation is obtained by making the assumption that the electro-motive force bears to electricity defined as above exactly the same energy relation as ordinary force does to matter, i.e.--- \begin{align*} &\shoveleft{(\text{wk.\ dn.\ on electricity moved during any displacement})} \\ &\qquad\qquad= (\text{total displacement of elect.}) \\ &\qquad\qquad\qquad\times (\text{resolved part of \textsc{e.m.f.}\ in the direction of displacement}). \end{align*} In the \textsc{iv}th part of Maxwell's treatise he gives complete investigations of the mechanical results flowing from this theory so far as it refers to currents, but he has not given the general results in the case of Electrostatics. Nor has he shewn satisfactorily, it seems to me, that the ordinary laws of Electrostatics flow from his theory. It is these investigations we now propose to make. \Paragraph{35} Our notation will be as far as possible the same as Maxwell's. Thus for the displacement at any point we use~$\mathbf{D}$, and for the \textsc{e.m.f.}~$\mathbf{E}$. From the connection explained in last section between $\mathbf{D}$ and~$\mathbf{E}$ we have \[ \EqnTag{IV}{1} \mathbf{D} = K\mathbf{E}/4\pi, \] where $K$ at any point is some linear vector function depending on the state of the medium at the point. If the medium change in any manner not electrical, e.g.\ by means of ordinary strain~$K$ will in general also suffer change. Let $w$ be the potential energy per unit volume due to the electrical configuration. Thus if a small increment~$\delta\mathbf{D}$ be given to~$\mathbf{D}$ at all points, the increment $\smalliiint\delta w\, \ds$ in $\smalliiint w\, \ds$, the potential energy of the electrical configuration in any space, = work done on the electricity in producing the change, \[ \therefore\qquad \smalliiint\delta w\, \ds = -\smalliiint S\mathbf{E}\, \delta\mathbf{D}\, \ds, \] by the relation stated in \Secref{34} existing between $\mathbf{E}$ and~$\mathbf{D}$. Thus limiting the space to the element~$\ds$ \[ \EqnTag{IV}{2} \delta w = -S\mathbf{E}\, \delta\mathbf{D}. \] %% -----File: 076.png---Folio 58------- Now suppose $\mathbf{D} = n\mathbf{D}'$ so that by \equationref{IV}{1} $\mathbf{E} = n\mathbf{E}'$ where~$\mathbf{E}'$, $\mathbf{D}'$ are corresponding \textsc{e.m.f.} and displacement respectively. Thus \[ \delta w = -n\, \delta n\, S\mathbf{E}'\mathbf{D}'. \] Integrating from $n = 0$ to $n = 1$, and finally changing $\mathbf{D}'$, $\mathbf{E}'$ into $\mathbf{D}$, $\mathbf{E}$ we get \[ \EqnTag{IV}{3} w = -S\mathbf{D}\mathbf{E}/2. \] From this we get \[ \delta w = - S\mathbf{E}\, \delta\mathbf{D}/2 - S\mathbf{D}\, \delta\mathbf{E}/2, \] so that by \equationref{IV}{2} $ S\mathbf{E}\, \delta\mathbf{D} = S\mathbf{D}\, \delta\mathbf{E}$ or by \equationref{IV}{1} \[ S\mathbf{E} K\, \delta\mathbf{E} = S\,\delta\mathbf{E}\, K\mathbf{E}. \] Hence (because $\mathbf{E}$ and~$\delta\mathbf{E}$ are quite arbitrary) $K$ is self-conjugate and therefore involves only six instead of nine coordinates\footnote{% We see from this that $\mathbf{D} = \Nabla{\mathbf{E}} w$ or $\mathbf{E} = \Nabla{\mathbf{D}} w$ according as $w$ is looked upon as a function of $\mathbf{E}$ or~$\mathbf{D}$.}. In electrostatics the line integral of~$\mathbf{E}$ round any closed curve must be zero, for otherwise making a small conductor coincide with the curve we shall be able to maintain a current by \Secref{34}, and so (by the same section) constantly do work on it (i.e.\ as a matter of fact create heat) without altering the statical configuration. Hence~$\mathbf{E}$ must have a potential, say~$v$. Thus \[ \EqnTag{IV}{4} \mathbf{E} = -\nabla v. \] Since in an electrostatic field there is no current in a conductor, $\mathbf{E} = 0$ throughout any such conductor and therefore $v=\text{const.}$ \Paragraph{36} The charge in any portion of space is defined as the amount of foreign electricity within that space. Thus the charge in any space is the surface integral of the displacement outwards. Thus if there be a charge on the element~$d\Sigma$ of a surface in the dielectric this charge $= S\, d\Sigma_a\, \mathbf{D}_a + S\, d\Sigma_b\, \mathbf{D}_b$ where $a$, $b$ denote the two faces of the element (so that $d\Sigma_a = -d\Sigma_b$) and in accordance with \Secref{1} above $d\Sigma_a$ points away from the region in which the displacement is~$\mathbf{D}_a$. Thus $\sigma$ being the surface density \[ \EqnTag{IV}{5} \sigma = [S\mathbf{D} U\, d\Sigma]_{a+b}, \] where the notation $[\;]_{a+b}$ is used for $[\;]_a + [\;]_b$. Similarly if there be finite volume density of foreign electricity, i.e.\ finite volume density of charge in any space, the charge $= -\smalliint S\mathbf{D}d\Sigma = -\smalliiint S\nabla\mathbf{D}\, \ds$, so that if~$D$ be the volume density \[ \EqnTag{IV}{6} D = -S\nabla\mathbf{D}. \] %% -----File: 077.png---Folio 59------- [The reason for having $+S\,d\Sigma_a\,\mathbf{D}_a$ before and $-S\,d\Sigma\,\mathbf{D}$ here is that in the former case we were considering a charge \emph{outside} the region where~$\mathbf{D}_a$ is considered---\emph{between} the regions $a$ and~$b$ in fact---whereas in the latter case we are considering the charge \emph{inside} the region where~$\mathbf{D}$ is considered. The same explanation applies to the sign of~$-S\mathbf{D} U\,d\Sigma$ for the surface of a conductor given below.] In conductors, as we saw in \Secref{34}, the displacement has virtually no meaning (except when it is changing and so the phenomenon of a current takes place) for the foreign electricity and the original electricity are not to be distinguished. Not so however with the surface of the dielectric in contact with the conductor. We may therefore regard the electricity within the body of the conductor as the original electricity so that the charge is entirely at the surface. Thus the surface density will be~$-S\mathbf{D} U\,d\Sigma$ where~$d\Sigma$ points away from the conductor and~$\mathbf{D}$ is the displacement in the dielectric. This may be regarded as a particular case of \equationref{IV}{5} $\mathbf{D}$~being in accordance with what we have just said considered as zero in the conductor. \Paragraph{37} All the volume integrals with which we now have to deal may be considered either to refer to the whole of space or only to the dielectrics, as the conductors (except at their surfaces) in all cases contribute nothing. The boundary of space will be considered as a surface at infinity and all surfaces where either $\mathbf{D}$ or~$\mathbf{E}$ is discontinuous. Putting $W = \smalliiint w\, \ds$ we have already found one expression for~$W$, viz.\ \[ 2W = -\smalliiint S\mathbf{D}\mathbf{E}\, \ds. \] We now give another. By \equationref{IV}{4} \begin{align*} 2W &= \smalliiint S\mathbf{D} \nabla v\, \ds \\ &= \smalliint vS\, d\Sigma\,\mathbf{D} - \smalliiint vS\nabla\mathbf{D}\, \ds, \end{align*} by \equationref{II}{9} \Secref{6} above. Thus by equations \Eref{IV}{5} and~\Eref{IV}{6} \Secref{36}, \[ \EqnTag{IV}{7} 2W = \smalliint v\sigma\, ds + \smalliiint vD\, \ds, \] where $ds$ is put, as it frequently will be, for an element of surface, i.e.\ $T\,d\Sigma$. The value of~$W$ which we shall use\footnote{% This is for the general case following the example of Helmholtz in the particular case when~$K$ reduces to a single scalar. See \textit{Wiss.\ Abh.}\ vol.~\textsc{i}.\ equation~(2$d$), p.~805. The \emph{method} adopted in the following investigation is also similar to his.} is obtained by combining these two, viz.\ \[ \EqnTag{IV}{8} W = \smalliint v\sigma\, ds + \smalliiint vD\, \ds + \tfrac{1}{2} \smalliiint S\mathbf{D}\mathbf{E}\, \ds. \] %% -----File: 078.png---Folio 60------- So far we have merely been shewing that all the above results of Maxwell's flow from what in \Secref{34} has been described as his theory. We now proceed to the actual problem in hand which is proved from these results however they may be obtained. I may remark that \emph{some} such investigation as the above seems to me necessary to make the logic of Maxwell's treatise complete. \Paragraph{38} Suppose now that~$W$ is the potential energy of some dynamical system extending throughout space. Let us give to every point of space a small displacement~$\delta \eta$ vanishing at infinity and find the consequent increment~$\delta W$ in~$W$. If this can be put in the form \[ \EqnTag{IV}{9} \delta W = - \smalliiint S\, \delta \eta_1\, \phi \nabla_1\, \ds, \] we shall have the following expression for~$\mathbf{F}$ the force per unit volume due to the system \[ \EqnTag{IV}{10} \mathbf{F} = \phi \Delta, \] and the following expression for~$\mathbf{F}_S$ the force per unit surface at any surface of discontinuity in~$\phi$ \[ \EqnTag{IV}{11} \mathbf{F}_s = -[\phi U\, d\Sigma]_{a+b}, \] the notation being the same as in \equationref{IV}{5} \Secref{36}. Moreover if~$\phi$ be self-conjugate the forces both throughout the volume and at surfaces of discontinuity are producible by the \emph{stress}~$\phi$ as can be seen by \Secref{13} above. [Compare all these statements with \Secref{14} above.] For proof, we have by \equationref{II}{9} \Secref{6} \[ \delta W = -\smalliiint S\, \delta \eta_1\, \phi \nabla_1\, \ds = -\smalliint S\, \delta \eta\, \phi\, d\Sigma + \smalliiint S\, \delta \eta\, \phi_1 \nabla_1\, \ds, \] where of course the element~$d\Sigma$ is taken \emph{twice}, i.e.\ once for each face. But \begin{align*} \delta W &= - \text{(work done by the system $\mathbf{F}$, $\mathbf{F}_s$ of forces)} \\ &= \smalliint S\, \delta \eta\, \mathbf{F}_s ds + \smalliiint S\, \delta \eta\, \mathbf{F}\, \ds, \end{align*} where the element~$ds$ is taken only \emph{once}. Equating the coefficients of the arbitrary vector~$\delta\eta$ for each point of space we get the required equations \Eref{IV}{10} and~\Eref{IV}{11}. \Paragraph{39} We must then put $\delta W$ where~$W$ is given by \equationref{IV}{8} in the form given in \equationref{IV}{9}. We must first define~$\delta$ when applied to a function of the position of a point. Suppose by means of the small displacement~$\delta\eta$ %% -----File: 079.png---Folio 61------- any point~$P$ moves to~$P'$. Then~$Q$ being the value at~$P$, before the displacement~$\delta \eta$, of a function of the position of a point, $Q + \delta Q$ is defined as the value of the function at~$P'$ after the displacement. Thus even in the neighbourhood of a surface of discontinuity~$\delta Q$ is a small quantity of the same order as~$\delta\eta$. Now the charge within any space, that is the quantity of foreign electricity within that space will not be altered by the strain. \[ \EqnTag{IV}{12} \therefore\qquad \delta(D\, \ds)=0, \qquad \delta(\sigma\, ds) = 0. \] To find $\delta\nabla$ we have \[ S (d\rho + \delta d\rho) (\nabla + \delta \nabla)\centerdot = S\, d\rho\, \nabla\centerdot \] or, since $\delta d\rho = -S\,d\rho\,\nabla\centerdot \delta \eta$, \begin{flalign*} && S\,d\rho\, \delta\nabla &= S\, d\rho\, \nabla_1 S\, \delta\eta_1\,\nabla, && \\ % \EqnTag{IV}{13} &\text{whence}& \delta\nabla &= \nabla_1 S\, \delta\eta_1\, \nabla. && \end{flalign*} The part of~$\delta W$ depending on the first two terms of \equationref{IV}{8} is by \equationref{IV}{12} \begin{align*} &\smalliiint \delta v\, D\, \ds + \smalliint \delta v\, \sigma\, ds && \\ &\quad = - \smalliiint \delta v\, S \nabla \mathbf{D}\, \ds + \smalliint \delta v\, S\, d \Sigma\, \mathbf{D} & &\quad \text{[by equations \Eref{IV}{5} and~\Eref{IV}{6} \Secref{36}]} \\ &\quad = \phantom{-}\smalliiint S \mathbf{D} \nabla\, \delta v\, \ds & &\quad \text{[by \equationref{II}{9} \Secref{6}]}. \end{align*} Noticing that $\delta\,\ds = -\ds\, S\nabla\,\delta\eta$ and that $4\pi\delta \mathbf{D} = K \delta \mathbf{E} + \delta K \mathbf{E}$ we see that the last term in \equationref{IV}{8} contributes \[ \smalliiint S \mathbf{D}\, \delta\mathbf{E}\, \ds - \tfrac{1}{2}\smalliiint S\mathbf{D}\mathbf{E}S\nabla\, \delta\eta\, \ds + (8\pi)^{-1} \smalliiint S\mathbf{E}\, \delta K\,\mathbf{E}\, \ds. \] Combining the last result with the first term of this we get \begin{align*} \smalliiint S\mathbf{D}(\nabla\, \delta v &- \delta(\nabla v))\, \ds = -\smalliiint S\mathbf{D}\, \delta \nabla\centerdot v\, \ds \\ &= -\smalliiint S\mathbf{D}\nabla_1S\,\delta\eta_1\,\nabla v\, \ds\\ &= \phantom{-}\smalliiint S\mathbf{D}\nabla_1 S\mathbf{E}\,\delta\eta_1\, \ds. \end{align*} Thus we have \[ \EqnTag{IV}{14} \delta W = - \smalliiint S\,\delta\eta_1 (\tfrac{1}{2}\nabla_1 S\mathbf{D}\mathbf{E} - \mathbf{E}S\nabla_1 \mathbf{D})\, \ds + (8\pi)^{-1}\smalliiint S\mathbf{E}\,\delta K\,\mathbf{E}\, \ds. \] \Paragraph{40} Now the increment~$\delta K$ in~$K$ is caused by two things viz.\ the mere rotation of the body and the change of \emph{shape} of the body. Let us call these parts $\delta K_r$ and~$\delta K_s$ respectively. First consider~$\delta K_r$. Suppose the rotation is~$\epsilon$ so that any vector which was~$\omega$ becomes thereby $\omega + V\epsilon \omega$. Thus the result %% -----File: 080.png---Folio 62------- of operating on $\omega + V\epsilon\omega$ by $K + \delta K_r$ is the same as first operating on~$\omega$ by~$K$ and then rotating. In symbols \[ (K + \delta K_r) (\omega + V\epsilon\omega) = K\omega + V\epsilon K\omega, \] \begin{flalign*} &\text{whence}& \delta K_r\, \omega &= V\epsilon K\omega - K(V\epsilon\omega). && \\ &\text{Thus}& S\mathbf{E} \delta K_r\,\mathbf{E} &= S\mathbf{E}\epsilon K\mathbf{E} - S\mathbf{E}K(V\epsilon\mathbf{E}) && \\ && &= 2S\epsilon K\mathbf{E}\mathbf{E} = 8\pi S\epsilon \mathbf{D}\mathbf{E}, && \end{flalign*} whence giving $\epsilon$~its value $V\nabla\delta\eta/2$, \[ (8\pi)^{-1} S\mathbf{E}\, \delta K_r\,\mathbf{E} = S\nabla\, \delta\eta\, V\mathbf{D}\mathbf{E}/2. \] Substituting in \equationref{IV}{14} \[ \delta W = -\smalliiint S\,\delta\eta_1 \{ V\nabla_1\mathbf{D}\mathbf{E}/2 - \mathbf{E}S\nabla_1\mathbf{D} \}\, \ds + (8\pi)^{-1}\smalliiint S\mathbf{E}\,\delta K_s\, \mathbf{E}\, \ds, \] or \[ \EqnTag{IV}{15} \delta W = \tfrac{1}{2} \smalliiint S\,\delta\eta_1\, V\mathbf{D}\, \nabla_1\mathbf{E}\, \ds + (8\pi)^{-1} \smalliiint S\mathbf{E}\, \delta K_s\, \mathbf{E}\, \ds. \] It only remains to consider~$\delta K_s$. $K$ is a function of the pure strain of the medium and~$\delta K_s$ is the increment in~$K$ due to the increment in pure strain owing to~$\delta\eta$. Calling this increment of pure strain~$\delta\psi$ so that by \equationref{III}{3} \Secref{12} above \[ \delta\psi\omega = -\delta\eta_1\, S\omega \nabla_1/2 - \nabla_1 S\omega\, \delta\eta_1/2, \] we have \begin{align*} \EqnTag{IV}{16} \delta K_s &= -S\,\delta\psi\,\zeta \Pre{\psi}{\Dop}_2 \zeta \centerdot K_2, \\ % \intertext{by \equationref{II}{7} \Secref{5} above. This gives by \equationref{II}{5} \Secref{3}} \EqnTag{IV}{17} \delta K_s &= -S\,\delta\eta_1 \Pre{\psi}{\Dop}_2 \nabla_1\centerdot K_2, \\ % \EqnTag{IV}{18} \therefore\qquad S\mathbf{E}\, \delta K_s\,\mathbf{E} &= -S\,\delta\eta_1 \Pre{\psi}{\Dop}_2 \nabla_1\, S\mathbf{E} K_2\mathbf{E}. \end{align*} Now by equations \Eref{IV}{1} and~\Eref{IV}{3} \Secref{35}, \[ 8\pi w = -S\mathbf{E} K\mathbf{E}, \] so that~$w$ is a function of the independent variables $\mathbf{E}$, $\psi$ (because $K$ is a function of~$\psi$). Therefore \Erratum{err62a}% [** PP: In right-hand member, change D to \Dop] \[ \EqnTag{IV}{19} (8\pi)^{-1} S\mathbf{E}\, \delta K_s\,\mathbf{E} = S\delta\eta_1 \Pre{\psi}{\Dop} w\,\nabla_1. \] This equation might have been deduced at once thus \[ (8\pi)^{-1} S\mathbf{E}\, \delta K_s\,\mathbf{E} = S\,\delta\psi\,\zeta \Pre{\psi}{\Dop} w\,\zeta = S\,\delta\eta_1 \Pre{\psi}{\Dop} w\,\nabla_1,% <- \Dop \] but \equationref{IV}{17} is itself of importance so the above proof is preferable. Thus finally from \equationref{IV}{15} we get \[ \EqnTag{IV}{20} \delta W = \smalliiint S\,\delta\eta_1 (V\mathbf{D} \nabla_1\mathbf{E}/2 + \Pre{\psi}{\Dop} w\, \nabla_1)\, \ds. \] %% -----File: 081.png---Folio 63------- We therefore have for~$\phi$ in equations \Eref{IV}{10} and~\Eref{IV}{11} \Secref{38}, \[ \EqnTag{IV}{21} \phi\omega = -V\mathbf{D}\omega\mathbf{E}/2 - \Pre{\psi}{\Dop} w\, \omega. \] This is a self-conjugate function so that as we saw in \Secref{38} it is a stress which serves to explain forces both throughout the volume of the dielectric and over any surfaces of discontinuity in~$\mathbf{D}$ or~$\mathbf{E}$\footnotemark. \footnotetext{As far as I am aware nobody has hitherto attempted to find the electrical forces much less the stress except in the case when~$\mathbf{D}$ is parallel to $\mathbf{E}$ i.e.\ the dielectric is electrically isotropic when unstrained. The particular results contained in \Secref{45} below have been obtained by Korteweg, Lorberg and Kirchhoff as is stated in Prof.~J.~J.~Thomson's paper (p.~155) referred to in \Secref{34}.} % [** End of footnote] \Section{41}{The force in particular cases} \Paragraph{41} Let us first consider that part $-V\mathbf{D}\Delta\mathbf{E}/2$ of the force $\bigl($equations~\Eref{IV}{10} \Secref{38} and~\Eref{IV}{21} \Secref{40}$\bigr)$ which does not depend on the variation of~$K$ with the shape of the body. Suppose our dielectric is homogeneous and electrically isotropic so that $K$~is a simple constant scalar. In this case \[ \EqnTag{IV}{22} 4\pi\mathbf{D} = -K\nabla v, \] by equations \Eref{IV}{1} and~\Eref{IV}{4} \Secref{35}. Therefore by equations \Eref{IV}{5} and~\Eref{IV}{6} \Secref{36}, \begin{align*} \EqnTag{IV}{23} 4\pi \mathbf{D} &= K\nabla^2v, \\ % \EqnTag{IV}{24} 4\pi\sigma &= - K[SU\,d\Sigma\,\nabla v]_{a+b}. \end{align*} From these we at once get by the theory of potential that \[ \EqnTag{IV}{25} Kv = \smalliiint u\mathbf{D}\, \ds + \smalliint u\sigma\, ds. \] From this we know by the theory of potential that at the surface where the charge~$\sigma$ resides $\nabla v$ is discontinuous only with regard to its normal component and at all other points is continuous. Thus \[ \nabla v_a = \nabla v_b + xU\,d\Sigma_a \] and by \equationref{IV}{24} $x= 4\pi\sigma/K$ so that \begin{flalign*} && &(4\pi)^{-1} K(\nabla v_a - \nabla v_b) = \sigma U\,d\Sigma_a, && \\ % \EqnTag{IV}{26} &\text{whence}& &(4\pi)^{-1} KU\,d\Sigma_a(\nabla v_a - \nabla v_b) = -\sigma. && \end{flalign*} Now the force $\mathbf{F}$ per unit volume is \begin{flalign*} && -V\mathbf{D}\Delta\mathbf{E}/2 &= -KV\nabla v\Delta\nabla v/8\pi && \\ % && &= -K\nabla^2 v\nabla v/4\pi, && \\ % \EqnTag{IV}{27} &\text{or}& \mathbf{F} &= -\mathbf{D}\nabla v, && \end{flalign*} %% -----File: 082.png---Folio 64------- and the force per unit surface~$\mathbf{F}_s$ is by \equationref{IV}{11} \Secref{38}, \begin{align*} \tfrac{1}{2}[V\mathbf{D}U\,d\Sigma \mathbf{E}]_{a+b} &= (8 \pi)^{-1} K [V \nabla v U\, d\Sigma\, \nabla v]_{a+b} \\ &= (8 \pi)^{-1} K V (\nabla v_a U\, d \Sigma_a\, \nabla v_a - \nabla v_b U\, d \Sigma_a\, \nabla v_b) \\ &= (8 \pi)^{-1} K V (\nabla v_a + \nabla v_b) U\, d \Sigma_a\, (\nabla v_a - \nabla v_b), \end{align*} whence by \equationref{IV}{26} \[ \EqnTag{IV}{28} \mathbf{F}_s = -\tfrac{1}{2} \sigma [\nabla v]_{a+b}. \] Thus we see that Maxwell's theory as given in \Secref{34} above reduces to the ordinary theory when $K$~is a single scalar. In fact two particles containing charges $e\;e'$ apparently repel one another with a force $ee' /Kr^2$ where $r$~is the distance between them, for by equations \Eref{IV}{25}, \Eref{IV}{27} and~\Eref{IV}{28} the force in any charged body is that due to a field of potential~$v$ given by \[ \EqnTag{IV}{29} Kv = \Sigma ue. \] \Paragraph{42} If the medium when strained remain electrically isotropic $\Pre{\psi}{\Dop} K$ as well as~$K$ must be a simple scalar. Thus with Thomson and Tait's notation for strain, which makes the coordinates of $\psi$, $e$, $f$, $g$, $a/2$, $b/2$, $c/2$ we have \begin{gather*} \Pre{\psi}{\Dop} K = dK/de = dK/df = dK/dg \\ dK/da = dK/db = dK/dc = 0. \end{gather*} Therefore $K$ is a function of $e + f + g$ only, i.e.\ of the density ($m$) of the medium. Thus because \[ de + df + dg = -dm/m = -d \log m \] we get $\Pre{\psi}{\Dop} K = - dK/d \log m = -k$ suppose. Hence \[ \Pre{\psi}{\Dop} w = (8 \pi)^{-1} S \mathbf{E} (dK/d \log m) \mathbf{E} = k \mathbf{E}^2/8\pi. \] Thus the force $-\Pre{\psi}{\Dop} w\,\Delta$ [equations~\Eref{IV}{21} \Secref{40} and~\Eref{IV}{10} \Secref{38}] resulting from the change of~$K$ with pure strain is in the case we are now considering \[ \EqnTag{IV}{30} -k \nabla \mathbf{E}^2/8\pi, \] and is\footnote{This is the same result as Helmholtz's on the same assumption \textit{Wiss.\ Abh.}\ \textsc{i}.\ p.~798.} therefore, according as $k$~is positive or negative, in the direction of or that opposite to that of the most rapid increase of % [** PP: Added hyphenation to ``electro-motive'' (2 x) for consistency] the square of the electro-motive\Erratum{err64a} force. Thus even in the case of a fluid dielectric which has no internal charge but which forms part of a non-uniform field of (electro-motive) force the surfaces of equal pressure and therefore the free surface will if originally plane no longer remain so. %% -----File: 083.png---Folio 65------- \TCSection{43}{Nature of the stress}{Nature of the Stress} \Paragraph{43} We have seen that the stress which serves to explain the electrostatic forces is that given by equation~(21) \Secref{40}, viz.\ % [** PP: No hyperlink; equation repeated here w/manual \tag] \[ \tag{21} \phi \omega = -V \mathbf{D} \omega \mathbf{E}/2 - \Pre{\psi}{\Dop}w\, \omega. \] Let us first consider the part $-V \mathbf{D} \omega \mathbf{E}/2$ which does not depend on the variation of~$K$. Putting~$\omega$ first $= U\mathbf{D}$ and then $= U \mathbf{E}$ we get \begin{align*} \phi U \mathbf{D} &= T\mathbf{D} T\mathbf{E}\centerdot U\mathbf{E}/2, \\ \phi U \mathbf{E} &= T\mathbf{D} T\mathbf{E}\centerdot U\mathbf{D}/2. \end{align*} Therefore putting $\omega$ first $=$ any multiple of $U\mathbf{D} + U \mathbf{E}$ and then $=$ any multiple of $U \mathbf{D} - U \mathbf{E}$ we get \begin{align*} \phi \omega &= \phantom{-}T \mathbf{D} T \mathbf{E} \omega/2, \\ \phi \omega &= - T \mathbf{D} T \mathbf{E} \omega/2. \end{align*} Lastly, since \[ -V \mathbf{D} \omega \mathbf{E} = \omega S \mathbf{D} \mathbf{E} - \mathbf{D} S \omega \mathbf{E} - \mathbf{E} S \omega \mathbf{D}, \] we see that if we put $\omega =$ any multiple of $V \mathbf{D}\mathbf{E}$ \[ \phi \omega = \omega S \mathbf{D} \mathbf{E}/2 = - \omega w. \] Thus we see that the stress now considered is a tension along one of the bisectors of $\mathbf{D}$ and~$\mathbf{E}$ (the bisector of the positive directions or the negative directions of both) $= T\mathbf{D} T\mathbf{E}/2$, an equal pressure along the other bisector and a pressure $= w$ perpendicular to both these directions. When $\mathbf{D}$~is parallel to~$\mathbf{E}$ this at once reduces to Maxwell's case, viz.\ a tension in the direction of~$\mathbf{E}$ and a pressure in all directions at right angles each $= w$. \Paragraph{44} We have now to consider the other part of the stress, viz. \begin{flalign*} &&\phi \omega &= - \Pre{\psi}{\Dop} w\, \omega, && \\ % \EqnTag{IV}{31} &\text{or}& \phi \omega &= \Pre{\psi}{\Dop}_1\omega\, S \mathbf{E}K_1 \mathbf{E}/8\pi. && \end{flalign*} If we assume that $K$ is a function of the density ($m$) of the medium only we shall have \[ dK/de = dK/df= dK/dg = -dK/d \log m = -k, \] say, and \[ dK/da = dK/db = dK/dc = 0, \] as in \Secref{42}. Here however $k$ is not in general a mere scalar but a self-conjugate linear vector function. We have then in this case \[ \phi \omega = -\omega S \mathbf{E} k \mathbf{E}/8\pi, \] %% -----File: 084.png---Folio 66------- which is a hydrostatic pressure or an equal tension in all directions according as $S \mathbf{E} k \mathbf{E}$ is positive or negative. In this case the~36 coordinates of $\Pre{\psi}{\Dop}_1\omega\centerdot K_1$ reduce to the 6~of $k$ for each point of space. A more general assumption is that $\delta K_s$ (\Secref{40}) depends only on the elongations in the directions of the principal axes of~$K$. Taking $i$, $j$, $k$ as unit vectors in these directions we again have \begin{flalign*} && dK/da &= dK/db = dK/dc = 0, && \\ &\text{and thus}& \phi i &= iS\mathbf{E} (dK/de) \mathbf{E}/8\pi, && \end{flalign*} and similarly for $j$ and~$k$, so that the principal axes of the stress now considered are the principal axes of~$K$. \FNParagraph{45}{45\protect\footnotemark} \footnotetext{For references to former proofs of this see foot-note to \Secref{40} above.} The most natural simple assumption for solid dielectrics seems to me to be that the medium is electrically isotropic before strain, and also isotropic with regard to the strain in the sense that if the strain be, so to speak, merely rotated, $\delta K_s$ will suffer exactly the same rotation. We may treat this problem exactly as we did (\Secref{20}) that of stress in terms of strain for an isotropic solid. Thus splitting up~$\delta\psi$ into its principal elongations, i.e.\ putting \[ \delta\psi\, \omega = - \Sigma\, \delta ei\, S i \omega, \] we shall get, as in \Secref{20}, \begin{align*} \delta K_s\, \omega &= -(\alpha - \beta) \Sigma\, \delta ei\, Si \omega + \beta \omega \Sigma\, \delta e \\ &= (\alpha - \beta)\, \delta\psi\, \omega - \beta \omega S \zeta\, \delta\psi\, \zeta. \end{align*} But $\delta K_s \omega = -S\,\delta \psi\, \zeta \Pre{\psi}{\Dop}_1\zeta\centerdot K_1 \omega$ by \equationref{IV}{16} \Secref{40}, so that from \equationref{IV}{31} \[ -8 \pi S\,\delta \psi\, \zeta \phi \zeta = (\alpha - \beta) S \mathbf{E}\, \delta \psi\, \mathbf{E} - \beta \mathbf{E}^2 S \zeta\, \delta\psi\, \zeta, \] whence we see by \Secref{4} above that \[ \EqnTag{IV}{32} \phi \omega = \bigl\{(\alpha - \beta) \mathbf{E}S \mathbf{E}\omega + \beta \mathbf{E}^2 \omega\bigr\}/8\pi, \] which consists of a pressure in the directions of the lines of force $= -\alpha \mathbf{E}^2/8\pi$ and another pressure in all directions at right angles $= -\beta \mathbf{E}^2/8\pi$. %% -----File: 085.png---Folio 67------- \TCSection{46}{% \texorpdfstring{{\scshape Magnetism}---magnetic potential, force, induction}{Magnetism} }{{\scshape Magnetism}\\[2ex] Magnetic potential, force, induction} \Paragraph{46} We now go on to the ordinary theory of magnetism; and here we shall merely follow Maxwell in his General Theory, so as to give an opportunity of comparing Quaternion proofs with Cartesian, as we have already done in Elasticity. We shall not consider in detail the effect of one small magnet upon another, as this has already been done by Tait. In connection with this I am content to remark that I think the treatment of this problem can be made somewhat simpler than Tait's by means of \emph{potential}. Suppose we have a pole~$-m$ at~$O$ and a pole~$+m$ at~$O'$ where $OO'$ is small. Calling the vector from $O$ to~$O'\,~\overline{OO'}$, let us call the vector~$m\overline{OO'}~\mu$, so that~$\mu$ is the vector magnetic moment of the magnet. The potential of~$-m$ at any point~$P$ is~$-mu$, where~$u$ as usual ${}= PO^{-1}$. Similarly the potential of~$+m$ is~$mu'$, where $u' = PO'^{-1}$. Therefore the potential of the magnet \[ {} = m (u'-u) = mS\overline{OO'} \nabla u, \] where of course $P$~is the variable point implied by~$\nabla$. Thus the potential of a small magnet~$\mu$ at any point ${}= S\mu\nabla u$. Hence the potential of \emph{any} magnet whose magnetic moment per unit volume at any point is~$\mathbf{I}$ is \[ \EqnTag{IV}{33} \Omega = S\nabla\smalliiint u \mathbf{I}\, \ds = - \smalliiint S\mathbf{I}\nabla u\, \ds, \] according to the convention of \Secref{9} above. By \equationref{II}{9} \Secref{6} this may be put \[ \EqnTag{IV}{34} \Omega = -\smalliint uS \mathbf{I}\, d \Sigma + \smalliiint uS \nabla \mathbf{I}\, \ds, \] which shews that we may consider it due to a volume density $S\nabla\mathbf{I}$ and a surface density $-S\mathbf{I}U\,d\Sigma$\footnote{% [Note added, 1892. More generally and better $-[S\mathbf{I}U\,d\Sigma]_{a+b}$.]} of magnetic matter, the surface density occurring wherever there is discontinuity in~$\mathbf{I}$. By again considering the poles $m$ and~$-m$ of the small magnet~$\mu$ we see that its potential energy when placed in a field of magnetic potential~$\Omega$ is $-S\mu\nabla\Omega$, whence just as we obtained %% -----File: 086.png---Folio 68------- \equationref{IV}{33} we now see that the potential energy ($W$) of any magnet in such a field is \begin{flalign*} \EqnTag{IV}{35} && W &= -\smalliiint S \mathbf{I} \nabla \Omega\, \ds, && \\ % \EqnTag{IV}{36} &\text{or}& W &= -\smalliint \Omega S \mathbf{I}\, d\Sigma + \smalliiint \Omega S \nabla \mathbf{I}\, \ds, && \end{flalign*} by \equationref{II}{9} \Secref{6} above, so that the potential energy is just the same as it would be for the imaginary distribution of magnetic matter. \Paragraph{47} The force ($\mathbf{H}$) on a unit magnetic pole at any point external to the magnet is given by \begin{align*} \mathbf{H} &= -\nabla \Omega = -\nabla S \nabla \smalliiint u \mathbf{I}\, \ds = -\nabla^2 \smalliiint u \mathbf{I}\, \ds + \nabla V \nabla \smalliiint u \mathbf{I}\, \ds \\ &= -4\pi\mathbf{I} + \nabla\smalliiint V \mathbf{I} \nabla u\, \ds = -4\pi\mathbf{I} + \nabla \mathbf{A}, \end{align*} where \[ \EqnTag{IV}{37} \mathbf{A} = \smalliiint V \mathbf{I} \nabla u\, \ds. \] Thus we see that for all external points $\mathbf{H} = \nabla \mathbf{A}$, so that~$\mathbf{A}$ is called the vector magnetic potential. [It is to be observed that since $\nabla \mathbf{A} = \mathbf{H} + 4 \pi \mathbf{I} =$ a vector, $S \nabla \mathbf{A} = 0$.] $\nabla \mathbf{A}$ is called the magnetic induction and for it we use the single symbol~$\mathbf{B}$ so that \begin{flalign*} \EqnTag{IV}{38} && \mathbf{B} &= \nabla \mathbf{A}. && \\ % \EqnTag{IV}{39} &\text{Thus}& S\nabla\mathbf{B} &= S \nabla^2 \mathbf{A} = 0, && \end{flalign*} and also by the equation for $\mathbf{H}$ just given \[ \EqnTag{IV}{40} \mathbf{B} = \mathbf{H} + 4\pi \mathbf{I}. \] This is not the way in which Maxwell defines the magnetic force and induction, but he shews quite simply (\emph{Elect.\ and Mag.}\ \S\S~398--9) that his definition and the present one are identical. This can be shewn as easily without analysis at all. \Paragraph{48} Where $\mathbf{I}$ is discontinuous both $\mathbf{H}$ and~$\mathbf{B}$ are also discontinuous. From the surface density view we gave in \equationref{IV}{34} we see that, just as we have the expression, given in \Secref{41} for $\nabla v_a - \nabla v_b$, so now \[ \EqnTag{IV}{41} \mathbf{H}_b - \mathbf{H}_a = - 4 \pi U\, d \Sigma_a[S \mathbf{I} U\, d\Sigma]_{a+b}, \] so that the discontinuity in~$\mathbf{H}$ is entirely normal to the surface of discontinuity. Further from this equation we have \begin{flalign*} && & S(\mathbf{H}_b - \mathbf{H}_a)\, d\Sigma_a = 4\pi [S \mathbf{I}\, d\Sigma]_{a+b}, && \\ % &\text{i.e.}& &\quad [S(\mathbf{H} + 4\pi\mathbf{I})\, d\Sigma]_{a+b} = 0, && \\ % \EqnTag{IV}{42} &\text{or}& &\quad\qquad [S \mathbf{B}\, d\Sigma]_{a+b} = 0 && \end{flalign*} so that the discontinuity in~$\mathbf{B}$ is entirely tangential. %% -----File: 087.png---Folio 69------- From this equation we see that for any closed surface whatever whether it include surfaces of discontinuity in $\mathbf{I}$ or not \[ \smalliint S\, d\Sigma\,\mathbf{B} = 0. \] For adding these surfaces to the boundary of the inclosed space, in accordance with \Secref{7} above, we see by \equationref{IV}{42} that they contribute zero to the surface integral; but the total surface integral is by \equationref{II}{9} \Secref{6} $\smalliiint S\nabla \mathbf{B}\, \ds = 0$ by \equationref{IV}{39}. \TCSection{49}{Magnetic solenoids and shells}{Magnetic Solenoids and Shells} \Paragraph{49} A magnet is said to be solenoidal if the imaginary magnetic matter of \equationref{IV}{34} is entirely on the surface. Thus for a solenoidal distribution \[ \EqnTag{IV}{43} S\nabla\mathbf{I}=0. \] In this case the potential is by \equationref{IV}{34} \[ \EqnTag{IV}{44} \Omega = -\smalliint uS\mathbf{I}\, d\Sigma. \] \Paragraph{50} A simple magnetic shell is defined as a sheet magnetised everywhere normally to itself and such that, at any point, the magnetic moment per unit surface is a constant called the strength of the shell. Calling the strength~$\phi$ we have for the potential energy at any point by \equationref{IV}{33} \[ \EqnTag{IV}{45} \Omega = -\phi \smalliint S\, d\Sigma\,\nabla u. \] Now $-S\,d\Sigma\,\nabla u$ is the solid angle subtended by the element~$d\Sigma$ at the point considered, so that \[ \EqnTag{IV}{46} \Omega = \phi \times \text{(solid angle subtended by shell at point)}. \] Thus if $P$ be a point on the positive side of the shell and~$P'$ a point infinitely near~$P$ but on the negative side \[ \text{Potential at $P - $ Potential at $P' = 4\pi \phi$}, \] or what comes to the same thing \[ -\int_P^{P'} S\mathbf{H}\, d\rho = 4\pi \phi. \] This integral may be taken along any path, e.g.\ along a path which nowhere cuts the shell. The same integral is true if~$\mathbf{H}$ be the magnetic force due to a whole field of which the shell is only one of several causes, for the part contributed by the rest of the %% -----File: 088.png---Folio 70------- field is zero on account of the infinite proximity of $P$ and~$P'$. For future use in electro-magnetism observe that this statement cannot be made if for~$\mathbf{H}$ in the integral be substituted~$\mathbf{B}$. \Paragraph{51} The condition that any magnet can be divided up into such shells is at once seen to be that~$\mathbf{I}$ can be put in the form \[ \EqnTag{IV}{47} \mathbf{I} = \nabla \phi, \] where $\phi$ is some scalar. In this case the potential is by \equationref{IV}{33} \begin{flalign*} && \Omega &= -\smalliiint S\mathbf{I}\nabla u\, \ds = -\smalliiint S\nabla\phi\nabla u\, \ds, && \\ \intertext{or by \equationref{II}{9} \Secref{6}} && \Omega &= -\smalliint \phi S\, d\Sigma\, \nabla u + \smalliiint\phi\nabla^2 u\, \ds, && \\ % \EqnTag{IV}{48} &\text{i.e.}& \Omega &= -\smalliint \phi S\,d\Sigma\nabla u + 4\pi\phi. && \end{flalign*} Remarking that the solid angle again occurs here it is needless to interpret the equation further. By \equationref{IV}{37} we have for the vector potential \[ \mathbf{A} = \smalliiint V\mathbf{I}\nabla u\, \ds = \smalliiint V\nabla\phi\nabla u\, \ds, \] or by \equationref{II}{9} \Secref{6} \[ \EqnTag{IV}{49} A=\smalliint\phi V\, d\Sigma\, \nabla u. \] \Paragraph{52} The potential energy of a magnetic shell of strength~$\phi$ placed in a field of potential~$\Omega$ is of importance. We see by \equationref{IV}{35} that it is \[ W = -\phi\smalliint S\, d\Sigma\, \nabla\Omega. \] If then the magnets which cause $\Omega$ do not cut the shell anywhere, so that $-\nabla\Omega = \nabla \mathbf{A}$, we shall have \begin{flalign*} \EqnTag{IV}{50} && W &= \phi\smalliint S\, d\Sigma\, \nabla \mathbf{A} = \phi\smalliint S\, d\Sigma \mathbf{B}, && \\ % \EqnTag{IV}{51} &\text{or}& W &= \phi\int S\, d\rho\, \mathbf{A}, && \end{flalign*} by \equationref{II}{8} \Secref{6}. Suppose now that $\mathbf{A}$ is caused by another shell of strength~$\phi'$. Then by \equationref{IV}{49} \[ \mathbf{A} = \phi'\smalliint V\, d\Sigma'\,\nabla u = \phi'\int u\,d\rho', \] by \equationref{II}{8} \Secref{6}. Thus finally the potential energy~$M$ of these two shells is given by \[ \EqnTag{IV}{52} M = \phi\phi'\smalliint uS\, d\rho\, d\rho'. \] \Paragraph{53} The general theory of induced magnetism when once the proposition (given in \equationref{IV}{42} \Secref{48}) that $[S\,d\Sigma \mathbf{B}]_{a+b}$ is zero is %% -----File: 089.png---Folio 71------- established, is much the same whether treated by Quaternion or Cartesian notation. We shall therefore not enter into this part of the subject. \TCSection{54}{% \texorpdfstring{{\scshape Electro-magnetism}---general theory}{Electromagnetism} }{{\scshape Electro-magnetism}\\[2ex] General theory}\Erratum{err67a} \Paragraph{54} We now propose to prove the geometrical theorems connected with Max\-well's general theory of Electro-magnetism by means of Quaternions. We assume the dynamical results of Chaps.~V., VI. and~VII., and the first six paragraphs of Chap.~VIII. of the fourth part of his treatise. These assumptions amount to the following. Connected with any closed curve in an electro-magnetic field there is a function \[ \EqnTag{IV}{53} p = -\int S\mathbf{A}\, d\rho, \] where $\mathbf{A}$ is some vector function at every point of the field. The function~$p$ has the following properties. If any circuit be made to coincide with the curve the generalised force acting upon the \emph{electricity} in the circuit is \[ \EqnTag{IV}{54} E = -\dot{p}. \] Again, if there be a current of electricity~$\gamma$ flowing round this circuit, the generalised force~$X$, corresponding to any coordinate~$x$ of the position of the circuit due to the field acting upon the \emph{conductor}, is \[ \EqnTag{IV}{55} X = \gamma dp/dx. \] \Paragraph{55} The first thing to be noticed is that $p$ can be transformed into a surface integral by \equationref{II}{8} \Secref{6} above. \begin{flalign*} \EqnTag{IV}{56} &\text{Thus}& &p = -\smalliint S\mathbf{B}\, d\Sigma, && \\ % \EqnTag{IV}{57} &\text{where}& &\quad\mathbf{B} = V\nabla \mathbf{A}, && \\ &\text{so that}& &\quad S\nabla \mathbf{B} = 0. && \end{flalign*} Next we see by the fundamental connection (\Secref{34} above) between the \textsc{e.m.f.}\ $\mathbf{E}$ and electricity, that $E$ must equal the line integral of~$\mathbf{E}$ round the circuit, or \[ \EqnTag{IV}{58} E = -\smallint S\mathbf{E}\, d\rho. \] %% -----File: 090.png---Folio 72------- We are now in a position to find $\mathbf{E}$ in terms of $\mathbf{A}$ and~$\mathbf{B}$, i.e.\ of $\mathbf{A}$. \Paragraph{56} The rate of variation of $p$ is due to two causes, viz.\ the variation of the field ($\dot{\mathbf{A}}$) and the motion of the circuit ($\dot{\rho}$). In the time~$\delta t$ then there will be an increment~$\delta\mathbf{A}$ in~$\mathbf{A}$ and an increment~$\delta d\Sigma$ in~$d\Sigma$ to be considered. Thus \[ \dot{p} \delta t = - \smallint S \delta \mathbf{A}\, d \rho - \smalliint S\mathbf{B}\, \delta d\Sigma. \] [This amounts to assuming that $\int S \delta \mathbf{A}\, d\rho = \smalliint S \delta \mathbf{B}\, d\Sigma$, which of course is true by \equationref{II}{8} \Secref{6}.] Now when the circuit changes slightly we may suppose the surface over which the new integral extends to coincide with the original surface and a small strip at the boundary traced out by the motion ($\dot{\rho} \delta t$) of the boundary. Thus $\delta d \Sigma$ is zero everywhere except at the boundary and there it \[ = V (\dot{\rho} \delta t)\, d\rho, \] \begin{flalign*} &\text{so that}& \dot{p}\, \delta t &= - \smallint S\, \delta\mathbf{A}\, d\rho + \delta t \smallint S \dot{\rho} \mathbf{B}\, d\rho, && \\ % \intertext{whence dividing by $\delta t$} \EqnTag{IV}{59} && \dot{p} &= \smallint S(-\dot{\mathbf{A}} + V\dot{\rho} \mathbf{B})\, d\rho, && \\ % \intertext{but by equations \Eref{IV}{54} and~\Eref{IV}{58} $\dot{p} = \int S\mathbf{E}\, d\rho$. Thus} \EqnTag{IV}{60} && \mathbf{E} &= - \dot{\mathbf{A}} + V\dot{\rho}\mathbf{B} - \nabla\psi, \end{flalign*} where $\psi$ is a scalar and $-\nabla\psi$ is put as the most general vector whose line integral round \emph{any} closed curve is zero. \Paragraph{57} We now come to the mechanical forces exerted on an element through which a current~$\mathbf{C}$ per unit volume flows. We see by \equationref{IV}{55} that the work done by the mechanical forces on any circuit through which a current of magnitude~$\gamma$ flows in any small displacement of the circuit equals $\gamma \times$ the increment in~$p$ caused by the displacement. Give then to each element~$d\rho$ of the circuit an arbitrary small displacement~$\delta\rho$ and let $\mathbf{F}'$ be the mechanical force exerted by the field upon the element. Thus as in last section \[ - \smallint S \mathbf{F}'\, \delta \rho = \gamma\, \delta p = -\gamma \smalliint S \mathbf{B}\, \delta d\Sigma = -\gamma \smallint S \,\delta\rho\, d\rho\, \mathbf{B}. \] Thus the force $\mathbf{F}'$ on the element~$d\rho$ is $\gamma V\,d\rho\, \mathbf{B}$. But we may suppose this element to be an element~$\ds$ of volume through which the current~$\mathbf{C}$ flows. Thus if for $\gamma\, d\rho$ we write $\mathbf{C}\, \ds$, and for $\mathbf{F}'$, $\mathbf{F}\, \ds$, where $\mathbf{F}$ is the force per unit volume exerted by the field, we get \[ \EqnTag{IV}{61} \mathbf{F} = V \mathbf{C} \mathbf{B}. \] %% -----File: 091.png---Folio 73------- \Paragraph{58} So far we have been able to go by considering the electric field as a mechanical system, but to go further (as Maxwell points out) and find how $\mathbf{B}$ or~$\mathbf{A}$ depends on the distribution of current and displacement in the field we must appeal to experiment. It has been shewn by experiment that a small circuit produces exactly the same mechanical effects on magnets as would a small magnet, at the same point as the circuit, placed with its positive pole pointing in the direction of the positive normal to the plane of the circuit when the positive direction round the circuit is taken as that of the current\footnote{% See \Secref{1} above for the convention with respect to the relation between the positive side of a surface and the positive direction round its boundary. Hitherto we have had no reason for choosing either the right-handed or the left-handed screw as the type of positive and negative rotation. But to make the statement in the text correct we must take the former.}. Moreover the magnetic moment of the magnet which must be placed there is proportional to the strength of the current~$\times$ the area of the circuit. Further, the effect of this circuit upon other such small circuits is the same as the mutual effects of corresponding magnets. We have now only to consider a finite circuit as split up in the usual way into a number of elementary circuits to see that a finite circuit will act upon magnets or upon other circuits exactly like a magnetic shell of strength proportional to the strength of the current and boundary coinciding with that of the circuit. The unit current in the electro-magnetic system is so taken as to make this proportionality an equality. The one difference between the circuit and the magnetic shell is that there is no discontinuity in the magnetic potential in going round the circuit, so that by \Secref{50} above the line integral of~$\mathbf{H}$ round the circuit will be $4\pi\times$ the strength of the current. In symbols \[ \smallint S\mathbf{H}\, d\rho = 4\pi \smalliint S\mathbf{C}\, d\Sigma \] for any curve, so that by \equationref{II}{8} \Secref{6} \begin{flalign*} \EqnTag{IV}{62} && 4\pi\mathbf{C} = V\nabla\mathbf{H}, && \\ % \EqnTag{IV}{63} &\text{whence}& S\nabla\mathbf{C} = 0, && \end{flalign*} which of course is a direct result of our original assumption that electricity moves like an incompressible fluid. Maxwell tacitly assumes this by making the assumption that only one coordinate is required to express the motion of electricity in a circuit. %% -----File: 092.png---Folio 74------- \Paragraph{59} We are now in a position to identify the $\mathbf{B}$ we are now using with the magnetic induction for which we have already used the same symbol. We see by \equationref{IV}{50} \Secref{52} that the mechanical force on the shell corresponding to any coordinate~$x$ is \[ -\phi d \smalliint S\, d\Sigma\, \mathbf{B}'/dx, \] where $\mathbf{B}'$ is the magnetic induction; and by equations \Eref{IV}{53} and~\Eref{IV}{55} that the force on the corresponding electric circuit is \[ - \phi d \smalliint S\, d\Sigma\, \mathbf{B}/dx, \] therefore $\mathbf{B} = \mathbf{B}'$ wherever there is no magnetism. And where there is magnetism $\mathbf{B}$ is not $= \mathbf{H}$ for $S \nabla \mathbf{B} = 0$, as we have seen. Thus $\mathbf{B} = \mathbf{B}' $ at all points. In other words the two vectors are identical and we are justified in using the same symbol for the two. This practically ends the general theory of electro-magnetism. We content ourselves with one more application of Quaternions in this subject. We give it because it exhibits in a striking manner the advantages of Quaternion methods. \TCSection{60}{Electro-magnetic stress}{Electro-magnetic phenomena explained by Stress} \Paragraph{60} In \Secref{46} \equationref{IV}{35} we have seen that the potential energy of a magnetic element $ = \mathbf{I}\, \ds$ in a magnetic field is $S\mathbf{IH}\, \ds$ when $\mathbf{H}$ has a potential. Maxwell assumes that the same expression is true whether $\mathbf{H}$ have a potential or not. Assuming this point\footnote{% I do not defend the legitimacy of this assumption. It seems to me bold to assume that a magnet possesses any such thing as potential energy in a field which has no potential. If we assume $\mathbf{H}$ \emph{and its derivatives} to be continuous throughout our typical element~$\ds$ of volume containing a great number of molecules (both material and magnetic) the force on a magnetic molecule~$\mu$ consisting of two poles is $-S\mu\nabla \centerdot \mathbf{H}$ and the force per unit volume $-S \mathbf{I} \nabla \centerdot \mathbf{H}$, which is only identical with the expression $-\nabla_1 S \mathbf{I} \mathbf{H}_1$ obtained below when $\mathbf{H}$ has a potential. With this expression a stress cannot be found that produces the force. If, however, $\mathbf{H}$ and its derivatives be not assumed continuous in this manner the force on the magnet $\mu$ is quite indeterminate whether the magnetic pole or the molecular current view of magnetism be taken, unless it be specified in what way the poles and currents are distributed in the element of volume.} with him we can find the force and couple acting on the medium and a stress which will produce that force and couple. The force and couple due to the magnetism of an element %% -----File: 093.png---Folio 75------- is obtained by giving the element an arbitrary translation and rotation and assuming that the work done by this force and couple = the decrement in the potential energy of the element. Thus the force per unit volume is $-\nabla_1 S\mathbf{IH}_1$ for the decrement in the potential energy due to a small translation $\delta\rho$ is $S\,\delta\rho\,\nabla_1 S\mathbf{IH}_1$. Similarly the couple $\mathbf{M}$ is given by \[ \EqnTag{IV}{64} \mathbf{M} = V\mathbf{IH} \] for the decrement $-S\mathbf{M}\,\delta\omega$ in the potential energy due to a small rotation~$\delta\omega$ is $-SV\,\delta\omega\, \mathbf{I}\centerdot \mathbf{H} = -S\,\delta\omega\, V\mathbf{I}\mathbf{H}$. The total force $\mathbf{F}$ per unit volume is the sum of that just given and that given by \equationref{IV}{61}, so that \[ \mathbf{F} = V\mathbf{CB} - \nabla_1 S\mathbf{IH}_1. \] Therefore by \equationref{IV}{62} \begin{align*} 4\pi\mathbf{F} &= VV\nabla\mathbf{H}\centerdot\mathbf{B} - 4\pi\nabla_1 S\mathbf{IH}_1 \\ &= -\mathbf{H}_1 S\nabla_1\mathbf{B} + \nabla_1 S\mathbf{BH}_1 - 4\pi\nabla_1 S\mathbf{IH}_1 \\ &= -\mathbf{H}_1 S\nabla_1\mathbf{B} + \nabla_1 S\mathbf{HH}_1. \end{align*} Now $S\nabla\mathbf{B} = 0$ so that $\mathbf{H}_1 S\nabla_1\mathbf{B} = \mathbf{H}S\Delta\mathbf{B}$, and again \begin{flalign*} && \nabla_1 S\mathbf{H}\mathbf{H}_1 &= \nabla(\mathbf{H}^2)/2, && \\ % \EqnTag{IV}{65} &\text{so that}& \mathbf{F} &= \phi\Delta, && \\ % \EqnTag{IV}{66} &\text{where}& 8\pi\phi\omega &= -2\mathbf{H} S\omega\mathbf{B} + \omega\mathbf{H}^2. && \\ \intertext{From this we get} && 8\pi V\zeta\phi\zeta &= 2V\mathbf{B}\mathbf{H} = 8\pi V\mathbf{I}\mathbf{H}, && \\ % \EqnTag{IV}{67} &\text{so that}& \mathbf{M} &= V\zeta\phi\zeta. && \end{flalign*} From these two results $\bigl($equations \Eref{IV}{65} and~\Eref{IV}{67}$\bigr)$ we see by \Secref{13} above that the stress~$\phi$ will produce all the mechanical effects of the field. This stress, as can be seen by giving~$\omega$ the required values in \equationref{IV}{66}, is one of pressure $-\mathbf{H}^2/8\pi$ in all directions at right angles to~$\mathbf{B}$ and of tension $-S\mathbf{H}(2\mathbf{B}-\mathbf{H})/8\pi$ in the direction of~$\mathbf{H}$. When there is no magnetism $\mathbf{H} = \mathbf{B}$ so that this pressure and tension become equal and their directions at right angles to and along~$\mathbf{B}$ respectively. In fact we then have \[ \EqnTag{IV}{68} 8\pi\phi\omega = -\mathbf{B}\omega\mathbf{B}. \] %% -----File: 094.png---Folio 76------- \Chapter{Section V}{Hydrodynamics} \Tocline{61}{Preliminary} \Paragraph{61} In the applications I am about to make in this I have practically nothing new to shew except the utility of Quaternion methods in the general theory of Hydrodynamics in all its parts. I therefore take a treatise (Greenhill's article in the \emph{Encyc.\ Brit.}\ on this subject) and work out the general theory on the lines of the treatise. This is more necessary than at first sight it would seem, for I believe mathematicians who have studied Quaternions are under the impression that the method does not lend itself conveniently to the establishment and treatment of such equations as the Lagrangian and those of Cauchy's integrals. With our meaning of~$\nabla$, however, the Quaternion treatment of these equations is as much simpler than the Cartesian as in the case of the Eulerian equations. Down to \equationref{V}{13} below the subject has already been handled by Prof.\ Hicks\footnote{% [Note added, 1892. Prof.~Tait's name ought to be added to Prof.~Hicks's.]} in his \emph{Quaternion treatment of Strains and Fluid motion} (\emph{Quart.\ Journ.\ Math}.~\textsc{xiv}.\ [1877] p.~271). I do not hesitate to go over the ground again as my methods are different from his. \Section{62}{Notation} \Paragraph{62} For the vector velocity at any point we shall use~$\sigma$, for the density~$D$, for the force per unit mass~$\mathbf{F}$, for its potential, when it has one,~$v$, and for the pressure~$p$. For time-flux which follows a particle we shall use $d/dt$ or the Newtonian dot, and for that which refers to a fixed point of space $\partial/\partial t$\footnotemark. \footnotetext{[Note added, 1892. I am aware that this is contrary to the usual English custom, but that custom---of interchanging the meanings of $d/dt$ and~$\partial/\partial t$ as given in the text---seems to me out of harmony with the meaning attached to~$\partial$ in other branches of Mathematics. At any rate I have respectable fellow-sinners, e.g.\ Kirchhoff in his \textit{Mechanik}, zweite Vorlesung, et seq.]} %% -----File: 095.png---Folio 77------- Thus \[ \EqnTag{V}{1} d/dt = \partial/\partial t - S\sigma\nabla. \] This equation is given on p.~446 of Greenhill's article already mentioned. In future we shall refer to this article simply as ``Greenhill's article.'' \TCSection{63}{Euler's equations}{Euler's Equations} \Paragraph{63} To find the equation of continuity, with Greenhill, we merely express symbolically that the rate of increase of the mass of the fluid in any space equals the rate at which it is flowing through the boundary. Thus $M$~being the mass in any space, \[ \EqnTag{V}{2} \partial M/\partial t = \smalliint DS\sigma\, d\Sigma. \] This is Greenhill's equation~(1) p.~445. By it and \equationref{II}{9} \Secref{6} above we have \[ \partial M/\partial t = \smalliiint S\nabla(D\sigma)\, \ds, \] whence reducing the volume to the element~$\ds$, \[ \EqnTag{V}{3} \partial D/\partial t = S\nabla(D\sigma). \] This is Greenhill's equation~(2). Thus by \equationref{V}{1} of last section, \[ \EqnTag{V}{4} \dot{D}/D = S\nabla\sigma. \] \Paragraph{64} To obtain Euler's equations of motion we express that the vector sum of the impressed forces for any volume equals the vector sum of the bodily forces plus the vector sum of the pressures on the surface. Thus \begin{align*} \smalliiint D\dot{\sigma}\, \ds &= \smalliiint D\mathbf{F}\, \ds - \smalliint p\, d\Sigma \\ &= \smalliiint (D\mathbf{F} - \nabla p)\, \ds, \end{align*} by \equationref{II}{9} \Secref{6} above. Applying this to the element~$\ds$ and dividing by~$D\, \ds$ we have \[ \EqnTag{V}{5} \dot{\sigma} = \mathbf{F} - \nabla p/D, \] or by \equationref{V}{1} \Secref{62} \[ \EqnTag{V}{6} \partial\sigma/\partial t - S\sigma\nabla \centerdot \sigma = \mathbf{F} - \nabla p/D. \] This is Greenhill's equations~(4)~(5)~(6). %% -----File: 096.png---Folio 78------- If $F(\rho, t) = 0$ ($F$ a scalar) be the equation of a surface always containing the same particles $\dot{F}= 0$, or by \equationref{V}{1} \[ \EqnTag{V}{7} \partial F/ \partial t - S \sigma \nabla F = 0. \] This is Greenhill's equation~(7). \Paragraph{65} Let us now put \[ \EqnTag{V}{8} \smallint dp/D = P. \] This of course assumes that $D$ is a function of $p$~only, which is not always the case, for instance in a gas where diffusion of heat is taking place. If $\mathbf{F}$ have a potential~$v$, $\mathbf{F} = - \nabla v$. Thus \equationref{V}{5} becomes \[ \EqnTag{V}{9} \dot{\sigma} = -\nabla(v + P), \] and \equationref{V}{6} \[ \EqnTag{V}{10} \partial\sigma/\partial t - S\sigma\nabla\centerdot\sigma = -\nabla(v+P). \] Now \[ S\sigma\nabla_1\centerdot \sigma_1 = V \sigma V\nabla_1 \sigma_1 + \nabla_1 S \sigma \sigma_1 = V \sigma V\nabla \sigma + \nabla(\sigma^2)/2. \] Thus \equationref{V}{10} becomes \[ \EqnTag{V}{11} \partial\sigma/\partial t + 2V\epsilon\sigma + \nabla R = 0, \] where the scalar~$R$ is put for $P + v - \sigma^2 / 2$ and the vector~$\epsilon$ for $V\nabla\sigma/2$. This is Greenhill's equations~(8)~(9)~(10). If $\epsilon = 0$, i.e.\ $V\nabla\sigma = 0$, we have $\sigma = \nabla\phi$, whence our equation becomes \begin{flalign*} && \nabla(\partial\phi/\partial t + R) &= 0, && \\ % \EqnTag{V}{12} &\text{so that}& \partial\phi/\partial t + R &= H, && \end{flalign*} where $H$ is a function of $t$ only. In next section we shall obtain in the case of an infinite fluid a generalisation of this which I believe has not hitherto been obtained. Here we have made the assumption that if $V \nabla \sigma= 0$ at one epoch it will be so always. This we shall prove later. \Paragraph{66} Greenhill next considers the case of steady motion. In this case $\partial\sigma/\partial t = 0$, so that \equationref{V}{11} becomes \[ \EqnTag{V}{13} 2V \epsilon \sigma + \nabla R = 0, \] and therefore the surface $R = \text{const.}$\ contains both vortices and stream-lines and the relation $dR/dn = 2q\omega\,\sin\theta$ given on p.~446 of Greenhill's article is the natural interpretation of our equation. So far we have been going over much the same ground as %% -----File: 097.png---Folio 79------- Hicks, but now we enter upon applications of Quaternions that I think have not been made before. \Paragraph{67} Greenhill next considers rotating axes and finds the form of the equations of motion when referred to these. Let $\sigma$ be the velocity referred to them; so that if the axes have at any time made the rotation~$q(\;)q^{-1}$ the real velocity will be~$q \sigma q^{-1}$. Thus\footnote{% This is quoted as a known result because it occurs generally in the subject of \emph{Rigid Dynamics}. No Quaternion proof, however as far as I am aware, has been given. We therefore give one here. What is meant by rotating axes may be thus explained.---Instead of choosing as our coordinates the vectors $\alpha'$, $\beta' \dotsc$ which occur in any problem, we take others $\alpha$, $\beta\dotsc$ such that $\alpha' = q \alpha q^{-1}$, $\beta' = q \beta q^{-1} \dotsc$ $q(\;)q^{-1}$ may be called the integral rotation of the axes. Thus if we say that the vector angular velocity of these axes themselves is~$\omega$ we mean that the real angular velocity is~$q \omega q^{-1}$, so that, as can be seen by putting in \equationref{V}{$a$} below $\alpha = \text{const.}$, or as in \equationref{IV}{58} \Secref{33} above, $\omega = 2V q^{-1} \dot{q}$. (This maybe established also by Tait's \textit{Quaternions}, \S~356, equation~(2), from which $q\omega q^{-1} = 2V \dot{q} q^{-1}$ or $\omega = 2V q^{-1} \dot{q}$). Again, when we say that the rate of increase of~$\alpha$ in space is~$\tau$, we mean that $\dot{\alpha}' = q\tau q^{-1}$ or $\tau = q^{-1}\centerdot d(q\alpha q^{-1})/dt\centerdot q$, or \begin{flalign*} \EqnTag{V}{$a$} && \tau &= \dot{\alpha} + 2VVq^{-1} \dot{q}\centerdot\alpha, && \\ % \EqnTag{V}{$b$} &\text{or}& \tau &= \dot{\alpha} + V \omega \alpha. && \end{flalign*} This could have been proved with fewer symbols and more explanation, but the above seems to me the most characteristic Quaternion proof. We might have started with not quite so general an explanation of reference to rotating axes and so refrained from introducing the \emph{integral} rotation, and therefore also~$q$.}, % [** End of footnote] as always with rotating axes, if~$\alpha$ be any vector function of a particle the rate of increase of~$\alpha$ in space is \[ \dot{\alpha} + V \omega \alpha, \] where $\omega$ is the angular velocity referred to the same system of rotating axes. Thus we see that the acceleration of a particle $=\dot{\sigma} + V \omega \sigma$. The velocity $\sigma = \dot{\rho}' + V \omega \rho'$, where $\rho'$ is the vector coordinate referred to our present axes (i.e.\ in the notation of the footnote $\rho' = q\rho q^{-1}$). Thus our equation of motion is now \[ \dot{\sigma} + V \omega \sigma = \mathbf{F} - \Nabla{\rho'} p/D, \] but now \[ d/dt = \partial/\partial t - S{\dot{\rho}'} \Nabla{\rho'}\centerdot = \partial/\partial t - S(\sigma - \omega\rho')\Nabla{\rho'}, \] whence changing $\rho'$~into $\rho$ we have \[ \EqnTag{V}{14} \partial\sigma/\partial t - S(\sigma - \omega\rho) \nabla\centerdot \sigma + V\omega\sigma = \mathbf{F} - \nabla p/D, \] which is the equation Greenhill gives on p.~446. %% -----File: 098.png---Folio 80------- \TCSection{68}{The Lagrangian equations} {The Lagrangian Equations\protect\footnotemark} \footnotetext{% [Note added, 1892. At the time of writing the essay I did not notice that these equations are a particular case of the general equation for an elastic body already established (see equations~\Eref{III}{15$m$} \Secref{16} and~\Eref{III}{15$n$} \Secref{16$a$}) \[ \left. D \ddot{\rho}' = \mathbf{F} - 2\rho'_1 S\nabla_1 \Pre{\Psi}{\Dop} w\, \Delta.\right] \]} % [** End of footnote] \Paragraph{68} We now consider the history of a single particle and for this we require different notation. We consider the vector coordinate ($\rho$) of a particle as a function of some other vector~$\alpha$ (say the initial value of~$\rho$) and of~$t$. We first require the connection between $\Nabla{\alpha}$ and $\Nabla{\rho}$. We shall drop the affix~$\alpha$ and retain~$\rho$, so that now \emph{not} $Q (\nabla_1, \rho_1)$ but $Q (\nabla_1, \alpha_1)$ \[ = Q(\zeta,\zeta). \] \begin{flalign*} &\text{\indent Now}& S\, d\alpha\, \nabla &= S\, d\rho\, \Nabla{\rho}, && \\ % &\text{but}& d\rho &= -\rho_1 S\, d\alpha\, \nabla_1, && \\ % && \therefore\qquad S\, d\alpha\, \nabla &= -S\, d\alpha\, \nabla_1 S \rho_1 \Nabla{\rho}, && \end{flalign*} or since $d\alpha$ is perfectly arbitrary \[ \EqnTag{V}{15} \nabla = - \nabla_1 S \rho_1 \Nabla{\rho}. \] Thus operating upon \equationref{V}{9} \Secref{65} by $\nabla_1S\rho_1(\;)$ and remembering that in that equation~$\nabla$ must be changed to $\Nabla{\rho}$ we get \[ \EqnTag{V}{16} \nabla_1 S \rho_1 \dot{\sigma} = \nabla (v + P), \] and this is our new equation of motion (Greenhill's~(1)~(2)~(3) p.~448). In \equationref{III}{1} \Secref{12} above let us put for $\rho + \eta$, $\rho$; and for $\rho$, $\alpha$. \begin{flalign*} \EqnTag{V}{17} &\text{Thus}& \chi\omega &= -S\omega \nabla\centerdot \rho. && \end{flalign*} Hence by \equationref{III}{6$d$} \Secref{14} above, \[ \text{strained vol.\ of el./original vol.} = S \nabla_1 \nabla_2 \nabla_3 S \rho_1 \rho_2 \rho_3 / 6, \] whence we see that \[ \EqnTag{V}{18} D S \nabla_1 \nabla_2 \nabla_3\, S \rho_1 \rho_2 \rho_3 = 6 D_0, \] where $D_0$ is a constant which when~$\alpha$ is taken as the initial value of~$\rho$ is the original density. This is the equation of continuity. %% -----File: 099.png---Folio 81------- \Section{69}{Cauchy's integrals of these equations} \Paragraph{69} To obtain these integrals we require to express $\Nabla{\rho}$ in terms of~$\nabla$. Now \equationref{V}{15} of last section expresses~$\nabla$ as a linear function of~$\Nabla{\rho}$. The converse we have already seen how to get. In fact from \equationref{II}{6$h$} \Secref{3$a$} above \[ \EqnTag{V}{19} 2J \Nabla{\rho} = -V\rho_1\rho_2\, S\nabla_1\nabla_2\nabla, \] where, with Greenhill, for brevity $J$ is put for $S\nabla_1\nabla_2\nabla_3\, S\rho_1\rho_2\rho_3/6$. The only function we wish to apply this to is $V \Nabla{\rho}\sigma$. We have \begin{align*} 2JV \Nabla{\rho}\sigma &= V\sigma_3\, V\rho_1\rho_2\, S\nabla_1\nabla_2\nabla_3 \\ &= (\rho_2 S\rho_1\sigma_3 - \rho_1 S\rho_2\sigma_3)\,S\nabla_1\nabla_2\nabla_3, \end{align*} or interchanging the suffixes 1 and~2 in the last term \[ \EqnTag{V}{20} JV \Nabla{\rho}\sigma = \rho_2 S\rho_1\sigma_3\, S\nabla_1\nabla_2\nabla_3, \] which gives the spin at any instant in terms of our present independent variables $\alpha$ and~$t$. \Paragraph{70} In order to obtain Cauchy's integral of \equationref{V}{16} operate on it by $V\nabla(\;)$. Thus \[ 0 = V(\nabla_1 + \nabla_2)\nabla_1 S\rho_1 \dot{\sigma}_2 = S\rho_1\dot{\sigma}_2\centerdot V\nabla_2\nabla_1. \] Now \[ d(S\rho_1\sigma_2 \centerdot V\nabla_2\nabla_1)/dt = S\sigma_1\sigma_2 \centerdot V\nabla_2\nabla_1 + S\rho_1\dot{\sigma}_2 \centerdot V\nabla_2\nabla_1. \] The first term of this last expression is zero since the sign is changed by interchanging the suffixes. From the last expression then \begin{align*} S\rho_1\sigma_2 \centerdot V\nabla_2\nabla_1 &= \text{constant} \\ &= \text{initial value of } S\zeta\sigma_2 V\nabla_2\zeta = -V\nabla\sigma_0, \end{align*} where $\sigma_0$ is the initial value of~$\sigma$. Changing the suffixes and substituting in \equationref{V}{20} we have \[ JV \Nabla{\rho}\sigma = -\rho_1 S\nabla\sigma_0\nabla_1, \] or giving $J$ its value~$D_0/D$ from \equationref{V}{18} and putting $V\Nabla{\rho}\sigma = \epsilon$, $V\nabla\sigma_0 = \epsilon_0$, we have \[ \EqnTag{V}{21} \epsilon/D = -S(\epsilon_0/D_0)\nabla \centerdot \rho. \] This is Greenhill's equations~(4)~(5)~(6) p.~448. The physical interpretation of this equation is quite easy. Consider a small vector~$d\alpha$ drawn in the fluid initially. At the %% -----File: 100.png---Folio 82------- time~$t$ this will have become $d\rho = -S\,d\alpha\,\nabla\centerdot \rho$. Thus we see that if a small vector~$c\epsilon_0/D_0$ be drawn in the fluid initially it will at the time~$t$ be~$c\epsilon/D$, from which we infer that an element of a vortex filament will always remain an element of a vortex filament; or, a vortex filament or tube always remains a vortex filament or tube. Again we see that $T \epsilon /D$ at any time varies directly with the elongation in the direction of the vortex filament so that $T\epsilon$ varies as that elongation~$\times$ the density, i.e.\ inversely as the cross-section of a small vortex tube at the point. This is not % [** PP: Keeping ``show''] the easiest way of arriving at these results, but it is well to show in passing how easy of interpretation are our results. We see from \equationref{V}{21} that if $\epsilon_0 = 0$, $\epsilon = 0$. In other words, if the motion have a velocity potential at one instant it will have one always. \Section{71}{Flow, circulation, vortex-motion} \Paragraph{71} We are about to consider vortex-motion from another point of view, viz.\ that of circulation. In \Secref{12} we saw that a strain due to a small displacement~$\eta$ could be decomposed into a pure strain followed by a rotation, the vector rotation being~$V\nabla\eta/2$. If now for~$\eta$ we put the small vector~$\sigma\,\delta t$ we see the propriety of calling $V\nabla\sigma/2$ the \emph{spin}. This therefore is taken as a \emph{definition} of spin. Greenhill does not take this (usual) course but uses the property we shall immediately prove concerning circulation and spin to lead to his definition. It is not necessary here to define flow and circulation. Putting $\sigma = \nabla \phi$ we see that for irrotational motion the flow $= -\int S \sigma d \rho = \int d \phi$ from one point to another is the increment in the velocity potential. Thus for mutually reconcilable paths it is always the same. Taking the circulation round a closed curve in the general case, the curve not inclosing any singular region of the fluid, we may transform the line integral into a surface integral by \equationref{II}{8} \Secref{6} above. Thus \[ \EqnTag{V}{22} -\smallint S \sigma\, d\rho = - \smalliint S\, d\Sigma\, \nabla \sigma, \] so that the circulation round the curve equals twice the surface integral of the spin. Hence Greenhill's definition of the spin. %% -----File: 101.png---Folio 83------- \Paragraph{72} From \equationref{V}{9} \Secref{65} we see that \begin{flalign*} && -\dfrac{d}{dt}\left(\int_A^B S\sigma\, d\rho\right) &= -\int_A^B S\sigma\, d\sigma - \int_A^B S\dot{\sigma}\, d\rho && \\ && &= -\tfrac{1}{2}\int_A^B d(\sigma^2) - \int_A^B d(v+P), && \\ % \EqnTag{V}{23} &\text{or}& -\dfrac{d}{dt}\left(\int_A^B S\sigma\, d\rho\right) % [** PP: Hard-coded bracket size; contents aren't tall] &= -\Bigl[v + P + \sigma^2/2 \Bigr]_A^B. && \end{flalign*} Therefore for a closed curve \[ d(\smallint S\sigma\, d\rho)/dt = 0, \] so that the circulation round the curve, and therefore the corresponding surface integral of the spin remains always constant. Taking the curve round a small vortex tube we once more arrive at the propositions enunciated in \Secref{70} about vortices. Thus at any point of a vortex tube the strength which is defined as the product of the spin into the cross-section is constant throughout all time. Also it is the same for all points of the tube, for by \equationref{II}{9} \Secref{6} we have \[ \smalliint S\,d\Sigma\,\nabla\sigma = \smalliiint S\nabla^2 \sigma\, \ds = 0 \] for any portion of the tube. But the only part contributing to the surface integral is the ends of the part of the tube considered, so that the strength at these two ends is the same, and is therefore constant for the whole tube. \Paragraph{73} These propositions are proved in the paper by Hicks already referred to. There is yet a third method of proof which, like Hicks's, is derived directly from the \equationref{V}{9} of motion. We have from \equationref{V}{1} \Secref{62} \[ \label{eqn:V.24} % [** PP: Footnoted tag, handle manually] \tag*{(24)\protect\footnotemark} \frac{d}{dt}\nabla - \nabla\frac{d}{dt} = -S\sigma\nabla\centerdot \nabla + \nabla S\sigma\nabla\centerdot = \nabla_1 S\sigma_1\nabla\centerdot\,; \] therefore\qquad $dV\nabla\sigma/dt - V\nabla\dot{\sigma} = V(\nabla_1 S\sigma_1\nabla_2\centerdot \sigma_2)$. \footnotetext{% [Note added, 1892. When first giving this in the \textit{Mess.\ of Math.}\ 1884 and when again putting it in the present essay, I was unaware of Prof.\ Tait's paper in \textit{Proc.\ R.~S.~E.}\ 1869--70, p.~143, where in the present notation he has for an incompressible fluid, (1) $d/dt = \partial/\partial t - S\sigma\nabla$ [given in Cartesian form], (2) $\dot{\sigma} = -\nabla v - \nabla p/D$, (3) $S\nabla\sigma=0$, (4) $\nabla\dot{\sigma} - d(\nabla\sigma)/dt = -S\nabla\sigma\nabla\centerdot\sigma$, (5) $V\nabla\dot{\sigma}=0$, (6) $d\nabla\sigma/dt = -S\nabla\sigma\nabla\centerdot\sigma$. Perhaps I have not interpreted Prof.~Tait's notation which is but briefly described correctly, but (4)~should apparently be $V\nabla\dot{\sigma} - d(\nabla\sigma)/dt = S\nabla\sigma\nabla\centerdot \sigma$, and so agree with (5) and~(6). It will be seen that the whole of \S\S~512--13 of Tait's \textit{Quaternions}, 3rd ed., is contained in this essay. I cannot at present recall whether this is owing wholly or in part to my being indebted to other old papers of Prof.~Tait, or whether in writing his 3rd ed.\ he has arrived independently at the same treatment, but am inclined to the latter belief.]} %[** End of footnote] %% -----File: 102.png---Folio 84------- Now by \equationref{V}{9} \Secref{65}, $V \nabla \dot{\sigma} = 0$, so that \begin{gather*} \EqnTag{V}{25} d V \nabla \sigma/dt = V \nabla_1 \sigma_2 S \nabla_2 \sigma_1, \\ \therefore\qquad d(\rho + x V \nabla \sigma)/dt = \sigma + \dot{x} V \nabla \sigma + x V \nabla_1 \sigma_2\, S \nabla_2 \sigma_1. \end{gather*} Now $\sigma_1 S\, \nabla_2\sigma_2 \nabla_1 = V \sigma_2 \nabla_1 S\, \nabla_2\sigma_1 + V \nabla_1 \nabla_2 S\, \sigma_1\sigma_2 + V \nabla_2 \sigma_2 S\, \nabla_1\sigma_1$, but $V \nabla_1\nabla_2 S\, \sigma_1\sigma_2 = 0$, for an interchange of suffixes leads to a change of sign. Therefore \[ V \nabla_1 \sigma_2 S \nabla_2 \sigma_1 = V \nabla \sigma S \nabla \sigma - S \nabla \sigma \nabla\centerdot \sigma. \] Put now $x = c/D$ where $c$ is some small constant. Thus \begin{flalign*} && \dot{x}/c = -\dot{D}/D^2 &= -S\nabla\sigma /D, && \\ % \EqnTag{V}{26} &\text{and we get}& d(\rho + cV\nabla \sigma\centerdot /D)\,dt &= \sigma - cS\,\nabla\sigma\, \nabla\centerdot \sigma\centerdot/D. && \end{flalign*} This shews that $\rho' \equiv \rho + cV\nabla\sigma/D$ is the variable vector of a particle, for the equation asserts that the time-flux of~$\rho' =$ the velocity at~$\rho'$. Thus again we get the laws of vortices given in \Secref{70}. \Section{74}{Irrotational Motion} \Paragraph{74} We use this heading merely to connect what follows with what Greenhill has under the same on p.~449. It is not very appropriate here. For irrotational motion we have seen that we may put \[ \EqnTag{V}{27} \sigma = \nabla \phi. \] If the fluid be incompressible we have further \[ \EqnTag{V}{28} 0 = S \nabla \sigma = \nabla^2 \phi. \] Let $T$ be the kinetic energy of this liquid. Thus \[ -2T = D \smalliiint (\nabla \phi)^2\, \ds = D \smalliint \phi S\, d\Sigma\, \nabla \phi - D \smalliiint \phi \nabla^2 \phi\, \ds \] by \equationref{II}{9} \Secref{6} above. Thus by \equationref{V}{28} \[ \EqnTag{V}{29} T = -\tfrac{1}{2} D \smalliint \phi S\, d\Sigma\, \nabla\phi. \] Hence we see that if we have a vector~$\sigma$ which satisfies the equations $V \nabla \sigma = 0$ and~$S \nabla \sigma = 0$ throughout any singly-connected space, or more generally satisfies the second of these equations %% -----File: 103.png---Folio 85------- throughout \emph{any} space, and also has zero circulation round any closed curve in that space\footnote{% To insure that~$\phi$ is single valued and so, that the part of the surface integral of \equationref{V}{29} due to ``barriers'' is zero.}; and further satisfies the equation $S\, d\Sigma\, \sigma = 0$ at the boundary; then the function \[ \smalliiint \sigma^2\, \ds = 0, \] or $ \sigma = 0$ at all points of the space, for each element of the integral being essentially negative must be zero. \Paragraph{75} The following important theorem now follows: If there are given; at every point of any region the convergence $S \nabla \sigma$ and the spin $V \nabla \sigma$, at every point of the boundary the normal velocity (and therefore $S\, d\Sigma\, \sigma$), and the circulation round every cavity which increases the cyclomatic number of the region; then the motion (as given by~$\sigma$ at every point) if possible, is unique. For let~$\sigma$ be one possible velocity, and if possible $\sigma + \tau$ another. Thus~$\tau$ satisfies all the conditions that~$\sigma$ does at the end of last section, and therefore $\tau = 0$ at every point or~$\sigma$ is unique. \Section{76}{Motion of a solid through a liquid} \Paragraph{76} We assume that there is no circulation of the liquid for any cycle; in other words, that if all the solids be brought to rest, so will be also the liquid. We shall take axes of reference fixed in the (one) moving solid. In the foot-note to \Secref{67} is explained how the \emph{rotation} of these axes is taken account of. The effect of the \emph{translation} of the origin will be easily found. Let $\sigma$, $\omega$ be the linear and angular velocities respectively of the moving solid. Let us put \[ \EqnTag{V}{30} \phi = - S \sigma \psi - S \omega \chi, \] where $\phi$ is the required velocity potential of the liquid, and $\psi$ and~$\chi$ are two vector functions of the position of a point independent of $\sigma$ and~$\omega$. Let us see whether $\psi$ and~$\chi$ \emph{can} be found so as to satisfy these conditions. $\sigma$ and~$\omega$ are of course assumed as quite arbitrary. % [** PP: No hyperlink; equation repeated here] The conditions are first the equation~(28) of continuity \[ \tag{28} \nabla^2 \phi = 0, \] %% -----File: 104.png---Folio 86------- which gives \[ \EqnTag{V}{31} \nabla^2 \psi = 0, \nabla^2\chi = 0, \] and second, the equality of the normal velocity $-SU\, d\Sigma\, \nabla\phi$ of the liquid at any point of the boundary with that of the boundary at the same point. Thus for a fixed boundary \[ S \sigma (S\, d\Sigma\, \nabla \centerdot \psi) + S \omega (S\, d\Sigma\, \nabla \centerdot \chi) = 0, \] whence on account of the arbitrariness of $\sigma$ and~$\omega$ \[ \EqnTag{V}{32} S\, d\Sigma\, \nabla\centerdot \psi = S\, d\Sigma\, \nabla\centerdot \chi = 0. \] For a moving boundary again we have \begin{flalign*} && S\, d\Sigma\, \nabla\phi &= S\, d\Sigma\, (\sigma + V \omega \rho), && \\ &\text{or}& S \sigma (d \Sigma + S\, d\Sigma\, \nabla\centerdot \psi) &+ S \omega (V \rho\, d\Sigma + S\, d\Sigma\, \nabla\centerdot \chi) = 0, && \\ % \intertext{which gives} \EqnTag{V}{33} && d\Sigma &+ S\, d\Sigma\, \nabla\centerdot\psi = 0, && \\ % \EqnTag{V}{34} && V\rho\, d\Sigma &+ S\, d\Sigma\, \nabla\centerdot\chi = 0. && \end{flalign*} Now it is well known that $\psi$ and~$\chi$ can be determined to satisfy all these conditions. In fact $x$ being a coordinate of either $\psi$ or~$\chi$ these conditions amount to:--- $\nabla^2 x = 0$ throughout the space, and $S\, d\Sigma\, \nabla x = {}$ given value at the boundary. \Paragraph{77} We do not propose to find $\psi$ and~$\chi$ in any particular case. We leave p.~454 of Greenhill's article and go on to p.~455, i.e.\ we proceed to find the equations of motion of the solid. For this purpose we require the kinetic energy of the system. Calling this~$T$ we shall have \[ \EqnTag{V}{35} T = -S \sigma\, \Sigma\sigma/2 - S \omega\, \Phi\sigma - S \omega\, \Omega\omega/2, \] where $\Sigma$, $\Omega$ and~$\Phi$ are linear vector functions and the first two are self-conjugate. This is the most general\footnote{% For let $A(\sigma, \sigma) + B(\omega, \sigma) + C(\omega, \omega)$ be this general function $A$, $B$, and~$C$ being scalar functions each linear in each of its constituents, and let \[ \Sigma\sigma = \zeta \{A(\zeta,\sigma) + A(\sigma,\zeta)\}, \quad \Phi\sigma = \zeta B(\zeta,\sigma), \quad \Omega\omega = \zeta \{C(\zeta,\omega) + C(\omega,\zeta)\}. \]} % [** End of footnote] quadratic function of $\omega$ and~$\sigma$. It involves 21~independent constants, six in~$\Sigma$, six in~$\Omega$ and nine in~$\Phi$. When $\psi$ and~$\chi$ are known $\Sigma$, $\Phi$ and~$\Omega$ are all known. We will obtain the expressions for them, and for simplicity will assume that the origin is at the centre of gravity of the moving solid. Let~$M$ be the mass of this last and~$\mu\omega$ (where, as is well-known, $\mu$~is a self-conjugate linear vector function) the moment of momentum. Thus by \equationref{V}{29} \[ 2T = -D \smalliint \phi S\, d\Sigma\, \nabla\phi - M \sigma^2 - S \omega \mu \omega. \] %% -----File: 105.png---Folio 87------- Putting $S\,d\Sigma\, \nabla\phi = S\,d\Sigma (\sigma + V\omega\rho)$ at the moving boundary and zero at the fixed, \[ 2T = D \smalliint (S \sigma \psi + S \omega \chi) (S\sigma\, d\Sigma + S\omega\rho\, d\Sigma) - M \sigma^2 - S\omega\,\mu\omega \] where the surface integral must be taken only over the moving boundary. Thus noting that $\Sigma$ and~$\Omega$ are self-conjugate \begin{align*} \EqnTag{V}{36} \Sigma \lambda &= M\lambda - \tfrac{1}{2} D \smalliint (\psi S\lambda\, d\Sigma + d\Sigma\, S\lambda\psi), \\ % \EqnTag{V}{37} \Omega\lambda &= \mu\lambda - \tfrac{1}{2} D \smalliint (\chi S\lambda\rho\, d\Sigma + V \rho\, d\Sigma\, S\lambda\chi), \\ % \EqnTag{V}{38} -S \lambda \Phi \lambda' &= \tfrac{1}{2} D \smalliint (S\lambda\chi\, S\lambda'\, d\Sigma + S\lambda\rho\, d\Sigma\, S\lambda'\psi). \end{align*} These surface integrals can be simplified, for in each of these equations the first surface integral equals the second. In \equationref{V}{36} in order to prove this we have merely to put for $d\Sigma$, $-S\,d\Sigma\,\nabla \centerdot \psi$ $\bigl($\equationref{V}{33}$\bigr)$ when we shall find by \equationref{II}{9} \Secref{6} above (since we may now suppose the integrations to extend over the \emph{whole} boundary), that \[ \smalliint \psi S \lambda\, d\Sigma = - \smalliiint \psi_1 S \nabla_1 \nabla_2 S \lambda \psi_2\, \ds = \smalliint d\Sigma\, S \lambda \psi. \] Similarly for \equationref{V}{37}. Again in \equationref{V}{38}, \[ \smalliint S\lambda\chi\, S\lambda'\, d\Sigma = - \smalliiint S \nabla_1 \nabla_2\, S\lambda\chi_1\, S\lambda'\psi_2\, \ds = \smalliint S \lambda\rho\, d\Sigma\, S\lambda'\psi. \] Thus finally for $\Sigma$, $\Omega$ and~$\Phi$ \begin{align*} \EqnTag{V}{39} \Sigma\lambda &= M \lambda - D \smalliint \psi S \lambda\, d\Sigma = M \lambda - D \smalliint d\Sigma\, S\lambda\psi, \\ % \EqnTag{V}{40} \Omega\lambda &= \mu\lambda - D \smalliint \chi S \lambda \rho\, d\Sigma = \mu\lambda - D \smalliint V \rho\, d\Sigma\, S\lambda\chi, \\ % \EqnTag{V}{41} -S\lambda \Phi\lambda' &= D \smalliint S \lambda\chi\, S \lambda'\, d \Sigma = D \smalliint S \lambda\rho\, d\Sigma\, S \lambda' \psi. \end{align*} This last may be put in the following four forms \begin{align*} \EqnTag{V}{42} \Phi' \lambda &= -D \smalliint d\Sigma\, S\lambda\chi = -D \smalliint \psi S \lambda\rho\, d\Sigma, \\ % \EqnTag{V}{43} \Phi \lambda &= -D \smalliint \chi S \lambda\, d\Sigma = -D \smalliint V \rho\, d \Sigma\, S \lambda \psi. \end{align*} Thus assuming that $\psi$ and~$\chi$ are determined we have found $T$ as a quadratic function of $\sigma$ and~$\omega$. \Paragraph{78} If $\mathbf{P}, \mathbf{G}$ be the linear and angular impulses of the system respectively our equations of motion are \begin{align*} \mathbf{\dot{P}} + V \omega \mathbf{P} &= \mathbf{F}, \\ \mathbf{\dot{G}} + V \omega \mathbf{G} &= \mathbf{M}, \end{align*} by the footnote to \Secref{67} above. Here $\mathbf{F}$ and~$\mathbf{M}$ are the external force and couple applied to the body. Now at the instant under consideration $\mathbf{P} = \Nabla{\sigma} T$, $\mathbf{G} = \Nabla{\omega} T$. But if the origin had been at the point~$-\rho$, $\mathbf{P}$ would still be $\Nabla{\sigma} T$ whereas $\mathbf{G}$ would be $\Nabla{\omega} T + V \rho \mathbf{P}$. %% -----File: 106.png---Folio 88------- Thus differentiation with regard to~$t$ does not affect the form of~$\mathbf{P}$ but does that of~$\mathbf{G}$. In fact using the last stated values of $\mathbf{P}$ and~$\mathbf{G}$ along with the last two equations and eventually putting $\dot{\rho} = \sigma$, $\rho=0$ we get \begin{align*} \EqnTag{V}{44} &d \Nabla{\sigma} T/dt + V\omega \Nabla{\sigma} T = \mathbf{F}, \\ % \EqnTag{V}{45} &d \Nabla{\omega} T/dt + V\omega \Nabla{\omega} T + V\sigma \Nabla{\sigma} T = \mathbf{M}. \end{align*} Now from \equationref{V}{35} we see that \begin{align*} \EqnTag{V}{46} \Nabla{\sigma} T &= \Sigma \sigma + \Phi' \omega, \\ % \EqnTag{V}{47} \Nabla{\omega} T &= \Phi \sigma + \Omega \omega, \end{align*} whence from equations \Eref{V}{44} and~\Eref{V}{45} \begin{align*} \EqnTag{V}{48} &\Sigma \dot{\sigma} + \Phi' \dot{\omega} + V\omega\, \Sigma\sigma + V\omega\, \Phi'\omega = \mathbf{F}, \\ % \EqnTag{V}{49} &\Phi \dot{\sigma} + \Omega \dot{\omega} + V\sigma\, \Sigma\sigma + (V\sigma\, \Phi' \omega + V\omega\, \Phi\sigma) + V\omega\, \Omega\omega = \mathbf{M}. \end{align*} Thus we see that with Quaternion notation even the general equations of motion are not too complicated to write down conveniently. We now leave Greenhill's article and proceed to certain theorems not contained therein. The Cartesian treatment of these subjects will be found in Lamb's treatise on \textit{The Motion of Fluids}, Chapters \textsc{vi.} and~\textsc{ix.}, which are headed \textit{Vortex Motion} and \textit{Viscosity} respectively. \Section{79}{The velocity in terms of the convergences and spins} \Paragraph{79} We have seen in \Secref{75} that if the spin and convergence are given for each point of a bounded fluid and the normal velocity at each point of the boundary, there is if any, but one possible motion. We shall see directly that this unique value always exists. At present we observe that we cannot find a motion giving an assigned spin and convergence for each point and \emph{any} assigned velocity for each point of the boundary. If however with such data a motion be possible we can find the velocity at any point explicitly. By \equationref{II}{19} \Secref{10} above we have \[ 4\pi\sigma = -\nabla \smalliiint \nabla u \sigma\, \ds, \] but by \equationref{II}{9} \Secref{6} \begin{gather*} \smalliiint \nabla u \sigma\, \ds = \smalliint u\, d\Sigma\, \sigma - \smalliiint u \nabla\sigma\, \ds \\ % \EqnTag{V}{50} \therefore\qquad 4\pi\sigma = \smalliint [V] \nabla u\, d\Sigma\, \sigma - \smalliiint [V] \nabla u\, \nabla\sigma\, \ds, \end{gather*} %% -----File: 107.png---Folio 89------- where the square brackets indicate that we may or may not retain the $V$ at our convenience. This equation may be put \[ \EqnTag{V}{51} 4\pi\sigma = \smalliint [V](\nabla u\, d\Sigma\,\sigma - u\,d\Sigma\,\nabla\sigma) + \smalliiint u\nabla^2\sigma \, \ds. \] Both of these equations solve the question now proposed. If the fluid be considered infinite the surface integral of \equationref{V}{50} vanishes if~$\sigma$ converges to zero at infinity and also that of \equationref{V}{51} if in addition the spin and convergence converge to a quantity infinitely small compared with the reciprocal of the distance of the surface. \Paragraph{80} We may now consider the case when $\nabla\sigma$ (spin and convergence) is given at all points and $S\,d\Sigma\,\sigma$ (normal velocity) at the boundary. It must be observed that the given value of the spin must be distributed in a solenoidal manner for \[ S\nabla V\, \nabla\sigma = 0. \] Whenever the quaternion \[ \EqnTag{V}{52} 4\pi q = -\smalliiint \nabla u\, \nabla\sigma\, \ds = \nabla \smalliiint u\, \nabla\sigma\, \ds, \] is a vector, we see that all the conditions except the boundary ones are satisfied by putting $\sigma = q$. Let us then see when $q$~reduces to a vector. By \equationref{II}{9} \Secref{6} above \[ 4\pi Sq = -\smalliint uS\, d\Sigma\, \nabla\sigma + \smalliiint uS\, \nabla^2\sigma\, \ds. \] Thus $q$ is a vector if this surface integral vanish. The surface integral vanishes if all the vortices form closed curves within the space. If they do not we must extend the space and make them form closed curves outside the original space. Extending the volume integral accordingly, we may put \[ \EqnTag{V}{53} 4\pi\sigma' = -\smalliiint \nabla u\, \nabla\sigma\, \ds, \] and now $\nabla\sigma' = \nabla\sigma$. Suppose then that \[ \EqnTag{V}{54} \sigma = \sigma' +\sigma''. \] We thus get $\nabla\sigma'' = 0$ and may therefore put \[ \EqnTag{V}{56} \sigma'' = \nabla\phi, \] where $\phi$ satisfies the equations \begin{flalign*} \EqnTag{V}{57} && &\nabla^2\phi = 0, && \\ % \EqnTag{V}{58} &\text{and}& S\, d\Sigma\, \nabla\phi = S\, &d\Sigma(\sigma - \sigma') = \text{known quantity.} && \end{flalign*} Now it is well known that $\phi$ can be determined so as to satisfy %% -----File: 108.png---Folio 90------- these two equations. Therefore the problem under discussion always admits of solution. The above is equivalent to Lamb's \S\S~128--131. His \S~130 is the natural interpretation of the equation \[ 4\pi\sigma = -\smalliiint V \nabla u\, \nabla\sigma\, \ds. \] His \S~132 is seen at once from \equationref{V}{50} above. For in the case he considers we, in accordance with \Secref{7} above, take each side of the surface of discontinuity as a part of the boundary. Now $[S\, d\Sigma\, \sigma]_{a+b} = 0$ so that for this part of the boundary we can leave out the part $S\, d\Sigma\, \sigma$ and we get \[ \EqnTag{V}{59} 4\pi\sigma = \smalliint V \nabla u\, V\, d\Sigma\, \sigma - \smalliiint V \nabla u\, \nabla\sigma\, \ds, \] so that if we regard $[V\, d\Sigma\, \sigma]_{a+b}$ as $-2\times$ an element of a vortex, we get the same law for these vortex \emph{sheets} as for the vortices in the rest of the fluid. \Paragraph{81} The velocity potential due to a single vortex filament of strength~$d\theta$ is at once obtained by putting in \equationref{V}{50} for $V \nabla \sigma\, \ds$, $2\, d\theta\, d\rho$. Thus calling $\sigma'$ the part of the velocity due to the filament \[ 2\pi\sigma' = -d\theta \smallint V \nabla u\, d\rho = -d\theta \smalliint u_1 V \nabla_1 V\, d\Sigma\, \nabla_1 \] by \equationref{II}{8} \Secref{6} above. Now \[ V \nabla_1 V\, d\Sigma\, \nabla_1 = -d\Sigma\, \nabla_1^2 + \nabla_1 S\, d\Sigma\, \nabla_1. \] Hence, since $\nabla^2 u = 0$ for all points not on the filament \[ \EqnTag{V}{60} 2\pi\sigma' = \nabla (d\theta \smalliint S\, d\Sigma\, \nabla u). \] Thus the velocity potential $= (2\pi)^{-l} \times$ the strength~$\times$ the solid angle subtended by the filament at the point considered. \SectionTitle{Kinetic Energy} \Paragraph{82} Let us put \begin{flalign*} \EqnTag{V}{61} && 4\pi q &= \smalliiint u \nabla \sigma\, \ds, && \\ % \EqnTag{V}{62} &\text{so that}& \sigma &= \nabla q. && \end{flalign*} We may now find expressions for $T$, the kinetic energy, in one or two interesting forms. We have \begin{flalign*} && 2T &= - \smalliiint D \sigma^2\, \ds, && \\ &\text{whence}& 2T = - \smalliiint D S\sigma\, \nabla q\, \ds &= - \smalliint D S\sigma\, d\Sigma\, q + \smalliiint D_1 S\sigma_1\, \nabla_1 q\, \ds. && \end{flalign*} %% -----File: 109.png---Folio 91------- We assume that the fluid extends to infinity and that the vortices and convergences are all at a finite distance, so that at infinity $\sigma$ is of order~$1/R^2$ and $q$ of order~$1/R$. Thus the surface integral vanishes and we have \[ \EqnTag{V}{63} 2T = \smalliiint D_1 S \sigma_1\, \nabla_1 q\, \ds, \] or, putting in the value of~$q$ from \equationref{V}{61}, \[ \EqnTag{V}{64} T = (8 \pi)^{-1} \smalliiint\smalliiint u D_1 S\sigma_1\, \nabla_1 \nabla_2 \sigma_2\, \ds_1\, \ds_2. \] These are Lamb's equations (28) and~(29), p.~160, generalised. Similarly his equation~30 generalised is \begin{flalign*} \EqnTag{V}{65} && 2T &= \smalliiint D_1 S \rho\sigma_1\, \nabla_1 \sigma_1\, \ds, && \\ &\text{for}& \smalliiint D_1 S \rho \sigma_1\, \nabla_1\sigma_1\, \ds &= \smalliint D S \rho \sigma\, d\Sigma\, \sigma - \smalliiint D S \zeta\sigma\zeta\sigma\, \ds, && \end{flalign*} by \equationref{II}{9} \Secref{6} above. But \[ \zeta\sigma\zeta = 2\zeta S\sigma\zeta - \zeta^2\sigma = \sigma, \] and the surface integral vanishes as before. \Section{83}{Viscosity} \Paragraph{83} To consider viscosity the assumption is made that the shearing stress which causes it is $\mu \times{}$ the rate of shear of the moving fluid. $\mu$~is assumed independent of the velocity and experiment seems to shew that it is also independent of the density. This last we assume though the variation with density can be easily treated. Consider a general strain $\chi\omega$. Since \begin{gather*} mS\lambda\mu\nu = S \chi\lambda\, \chi\mu\, \chi\nu = S \lambda \chi' V\, \chi\mu\, \chi\nu\footnotemark, \\ V \chi\mu\, \chi\nu = m \chi'^{-1} V \mu \nu. \end{gather*} \footnotetext{% Kelland and Tait's \textit{Quaternions}, chap.~x.\ equation~($n$). We have already used a particular case of this in \Secref{15}, above.}% % [** PP: End of footnote, no paragraph] Putting $\mu$ and~$\nu$ for any two vectors perpendicular to~$\omega$ we see that the normal~$U \omega$ to any interface becomes~$U \chi'^{-1} \omega$ by the strain. Hence the interface~$\omega$ experiences a shear (strain) which equals the resolved part of $\chi U \omega$ perpendicular to the vector $\chi'^{-1} \omega$ equals resolved part of $(\chi - \chi'^{-1})U \omega$ perpendicular to~$\chi'^{-1} \omega$. Now when $\chi$ is the strain function due to a small displacement~$\sigma\,dt$, by \Secref{12} above \begin{flalign*} \EqnTag{V}{66} && \chi\omega &= \omega - S \omega\, \nabla \centerdot \sigma\, dt, && \\ %% -----File: 110.png---Folio 92------- \EqnTag{V}{67} &\text{whence}& \chi'^{-1} \omega &= \omega + \nabla_1 S \omega \sigma_1\, dt, && \end{flalign*} and we see that the shear is the resolved part of \[ -(SU\omega\nabla\centerdot \sigma + \nabla_1 SU\omega\sigma_1)\, dt \] perpendicular to $\omega$, i.e.\ parallel to the interface~$\omega$. From this we see by our assumption concerning viscosity that any element of the fluid is subject to a stress~$\phi$ given by \[ \phi \omega = -R\omega - \mu(S\omega\, \nabla\centerdot \sigma + \nabla_1 S \omega \sigma_1) \] $R$ being a scalar. Now we define the pressure~$p$ by putting \begin{gather*} 3p = S \zeta \phi \zeta = 3R + 2 \mu S\, \nabla\sigma \\ % \EqnTag{V}{68} \therefore\qquad \phi\omega = -p\omega + \tfrac{2}{3} \mu \omega S\, \nabla\sigma - \mu(S\omega\nabla\centerdot \sigma + \nabla_1 S\omega\sigma_1). \end{gather*} The equation of motion is \[ D \dot{\sigma} = D\mathbf{F} + \phi\Delta, \] or \[ \EqnTag{V}{69} D(\partial\sigma/\partial t - S\sigma\, \nabla\centerdot \sigma) = D \dot{\sigma} = D \mathbf{F} - \nabla p - \mu\nabla S\, \nabla\sigma/3 - \mu\nabla^2 \sigma. \] If $\mu$ be not as stated independent of~$D$ this equation must be made to contain certain space fluxes of~$\mu$ which are quite easy to insert. \Paragraph{84} In \S~179 of Lamb's treatise he considers the dissipation function due to the viscosity of a fluid in a way that seems to me misleading if not wrong. It appears as if he should add to the first expression of that section \[ u\,dp_{xx}/dx + v\,dp_{xy}/dx + w\,dp_{xz}/dx. \] He refers to Stokes, \textit{Camb.\ Trans.}\ vol.~\textsc{ix.}\ p.~58. Let us give the quaternion treatment of Stokes's method. If $T$ is the kinetic energy of any portion of the fluid \begin{flalign*} && 2T &= -\smalliiint D\sigma^2\, \ds. && \\ % &\text{Thus}& \because\quad &d(D\, \ds)/dt = 0, && \\ % && \dot{T} &= -\smalliiint DS\sigma \dot{\sigma}\, \ds, && \\ % &\text{but}& D\dot{\sigma} &= D\mathbf{F} + \phi_1 \nabla_1, && \\ % &\text{so that}& \dot{T} &= -\smalliiint DS \sigma \mathbf{F}\, \ds -\smalliiint S \sigma \phi_1 \nabla_1\, \ds && \\ % \EqnTag{V}{70} &\text{or}& \dot{T} = -\smalliiint DS \sigma \mathbf{F}\, \ds &- \smalliint S \sigma \phi\, d\Sigma + \smalliiint S \sigma_1 \phi \nabla_1\, \ds. && \end{flalign*} If now $\phi$ be expressed in terms of $p$ and~$\mu$ we have $\dot{T}$~depending on $\mathbf{F}$, $p$ and~$\mu$. The part of~$\dot{T}$ depending on $\mathbf{F}$ and~$p$ represents energy stored up as potential energy of position and potential %% -----File: 111.png---Folio 93------- energy of strain respectively, but the part depending on~$\mu$ represents a loss of energy to the system we are considering the energy being converted into heat. Thus putting $\phi = p + \varpi$ we have by \equationref{V}{68} \[ \EqnTag{V}{71} \left. \begin{aligned} \varpi \omega &= \tfrac{2}{3} \mu\omega S \nabla\sigma - \mu(S\omega\nabla\centerdot \sigma + \nabla_1 S\omega\sigma_1) \\ &= \tfrac{2}{3} \mu\omega S \zeta\psi\zeta + 2\mu\psi\omega, \end{aligned}\right\} \] where $\psi$ is given by \equationref{V}{75} below. Thus we see that the rate of loss of energy is \[ \EqnTag{V}{72} \smalliint S \sigma \varpi\, d\Sigma - \smalliiint S \sigma_1 \varpi\nabla_1\, \ds. \] The surface integral is the work done by viscosity against the moving fluid at the boundary and the volume integral is considered due to the work done against the straining of the fluid. Thus we put \[ \EqnTag{V}{73} F = -S\sigma_1 \varpi \nabla_1, \] and call $F$ the ``dissipation function.'' By \equationref{II}{6} \Secref{3} above \[ \EqnTag{V}{74} F = - S \psi \zeta \varpi \zeta, \] where $\psi$ is the rate of pure strain of the fluid, i.e.\ \[ \EqnTag{V}{75} 2\psi\omega = -S\omega\,\nabla\centerdot \sigma - \nabla_1 S \omega \sigma_1. \] Again by \equationref{III}{18} \Secref{18} above because $F$ is quadratic in~$\psi$ \[ \EqnTag{V}{76} F = -S\psi\zeta \Pre{\psi}{\Dop} F\, \zeta/2, \] so that from \equationref{V}{74} we have, by \Secref{4} above, \[ \EqnTag{V}{77} \varpi = \Pre{\psi}{\Dop} F/2. \] Substituting for $\varpi$ from \equationref{V}{71} in \equationref{V}{74} \[ \EqnTag{V}{78} F = - \tfrac{2}{3} \mu(S \zeta \psi \zeta )^2 - 2\mu\psi \zeta\psi\zeta, \] which gives $F$ in terms of $\psi$. %% -----File: 112.png---Folio 94------- \Chapter{Section VI}{The Vortex-Atom Theory} \Tocline{85}{Preliminary} \Paragraph{85} If Quaternions can give valuable hints or indicate a promising method of dealing with the highly interesting mathematical theory of Vortex-Atoms, I think this alone ought to be sufficient defence of its claims to be within the range of practical methods of investigation. In what follows I think I may be said to have indicated a hopeful path to follow in order to test to some extent the soundness of this theory. \Section{86}{Statement of Sir Wm.~Thomson's and Prof.~Hicks's theories} \Paragraph{86} Sir Wm.~Thomson's theory is so well-known that it is not necessary to state it in detail. Matter is some differentiation of space which can vary its position carrying with it so to speak certain phenomena, some of which admit of definite quantitative measurement. Perhaps the most important of these phenomena is what is called mass. Now, says in effect Sir Wm. Thomson, if we suppose all space filled with an incompressible perfect fluid the vortices in it are just such differentiations of space. They also carry about with them definite quantitative phenomena. Can we prove that these hypothetical vortices would act upon one another as do the atoms of matter, the laws of whose action are contained in the various Sciences, e.g.\ Physics, Chemistry and Physiology? The problem in its first stages at any rate is a mathematical one, but during the many years it has been before the mathematical world very little progress has been made with it. Hicks's extension of this theory is perhaps not so well known, but it seems to me quite as interesting and more likely to tally with the known phenomena of matter. He enunciates his theory %% -----File: 113.png---Folio 95------- in the \emph{Proceedings of the Cambridge Philosophical Society}, vol.~\textsc{iii}. p.~276. It differs from Thomson's simply in assuming that the fluid does not quite fill space---that there are in it bubbles\footnote{% By ``bubbles'' of course I do not mean spaces occupied by another kind of fluid of smaller density but actual vacua. Thus a bubble may start into existence where none previously existed, or again a bubble may completely disappear.}. These bubbles will find their way to where the pressure is least, i.e.\ speaking generally to the centre of some at least of the vortices. Thus we have another source of differentiation of space and general considerations seem to point to \emph{these} differentiations being the true atoms, though of course ``atom'' is no longer a descriptive term. \Section{87}{General considerations concerning these theories} \Paragraph{87} In Maxwell's article \emph{Atom} in the \emph{Encyc.\ Brit.}\ p.~45, he says---``One of the first if not the very first desideratum in a complete theory of matter is to explain first mass and second gravitation\ldots\ldots\ In Thomson's theory the mass of bodies requires explanation. We have to explain the inertia of what is only a mode of motion and inertia is a property of matter, not of modes of motion. It is true that a vortex ring at any given instant has a definite momentum and a definite energy, but to shew that bodies built up of vortex rings would have such momentum and energy as we know them to have is in the present state of the theory a very difficult task. ``It may seem hard to say of an infant theory that it is bound to explain gravitation.'' Now as Hicks tells us what induced him to give his theory was the promise it gave of explaining gravitation. But I believe nowhere does he point out the still more important result that probably on his theory we can explain inertia. The statement of the principal property of inertia put scientifically is that the motion of the centre of gravity of any two bodies approximates more and more nearly to uniform velocity in a straight line, the more nearly they are isolated from external influence. But this property is probably true of Hicks's bubbles. The centre of gravity of any portion of the fluid containing certain bubbles $(1)$, $(2) \dotsc (n)$ will if approximately isolated from all the %% -----File: 114.png---Folio 96------- rest of the fluid move approximately in a straight line, but this amounts to saying that the centre of volume of the $n$~bubbles will also move uniformly in a straight line \emph{if} the centre of the whole volume (bubbles \emph{and} fluid) considered, move similarly. These conditions are not evidently true, but I think they are probably so when we consider groups of large numbers of bubbles. Whether this theory explains gravitation is one of the principal questions to be considered in the following first trial. \Section{88}{Description of the method here adopted} \Paragraph{88} Finding it practically impossible to consider the real problem of a number of bubbles in the liquid, I consider the fluid continuous and not incompressible. Let us assume that \[ \EqnTag{VI}{1} D = \tanh (p/c), \] where $c$ is some very small constant which in the limit $= 0$. For ordinary values of~$p$, $D \text{ very nearly} = 1$. It is only when $p$~becomes comparable with~$c$ that~$D$ varies. When $p=0$, $D=0$. Also \[ \EqnTag{VI}{2} P = \smallint(dp/D) = c \log \sinh (p/c). \] For ordinary values of $p$ then, $P = p$ but when $p$ becomes comparable with~$c$, $P$~varies and when $p = 0$, $P = -\infty$. If we assume then for the greater part of the fluid that $p$~is large compared with the small quantity~$c$ we see that the fluid will have almost exactly the same properties as a liquid containing bubbles. We may now apply the equations of motion of \Secref{64}, \Secref{65}. Now we know the velocity at any point of an infinite fluid in terms of the spins and convergences at all points. From this we may deduce the acceleration in terms of the spins, convergences and time-fluxes. It so happens that by the equations in \Secref{73} we can get rid of the time-fluxes of the spins. This greatly simplifies the further discussion of the problem. The equation that we thus deduce forms the starting point of our investigation, for the phenomena of gravitation, electro-magnetism, stress,~\&c.,\ are exhibited and measured by the \emph{acceleration} in bodies due to their relative positions. The equation we obtain at once gives a generalisation of the integral, \equationref{V}{12} \Secref{65} above. At the end of this section I give an equation which is rather %% -----File: 115.png---Folio 97------- complicated but which promises to enable us to deal more rigorously with our problem than I profess to have done below. We proceed to the investigation just indicated. \Section{89}{Acceleration in terms of the convergences, their time-fluxes, and the spins} \Paragraph{89} Let us put for the convergence and twice the spin $m$ and~$\tau$ respectively, so that \begin{align*} \EqnTag{VI}{3} S \nabla \sigma &= m, \\ % \EqnTag{VI}{4} V \nabla \sigma &= \tau. \end{align*} We have seen that in \equationref{V}{50} \Secref{79} we may neglect the surface integral. Thus \begin{flalign*} \EqnTag{VI}{5} && 4\pi\sigma &= \nabla \smalliiint u (m + \tau)\, \ds, && \\ && \therefore\qquad 4\pi\partial\sigma/\partial t &= \nabla \smalliiint u \partial(m+\tau)/\partial t\centerdot\, \ds. && \\ &\text{\indent By \Secref{73}}& \dot{\tau} &= \tau S \nabla \sigma - S\tau\nabla\centerdot \sigma, && \end{flalign*} but by \equationref{V}{1} \Secref{62} \begin{align*} \partial\tau/\partial t &= \dot{\tau} + S \sigma \nabla \centerdot \tau, \\ \therefore\qquad \partial\tau/\partial t &= S \sigma \nabla\centerdot \tau + \tau S\nabla\sigma - S \tau\nabla\centerdot \sigma, \\ % \intertext{or because $S \nabla \tau = 0$} \EqnTag{VI}{6} \partial\tau/\partial t &= \tau S \Delta\sigma - \sigma S\Delta\tau = V \nabla V \sigma \tau \\ % \EqnTag{VI}{7} \therefore\qquad 4\pi\partial\sigma/\partial t &= \nabla \smalliiint u(V\nabla V\sigma\tau + \partial m/\partial t)\, \ds. \end{align*} This equation is not in a convenient form for our purpose, for $\tau$ and~$m$ may be and most probably will be discontinuous, so that it is advisable to get rid of their derivatives. Moreover it is advisable to allow the time-flux of~$m$ to appear only under the form $d(m\, \ds)/dt$ because in case of pulsations the time integral of this expression will be zero, whereas we can predicate no such thing of $(\partial m/\partial t)\, \ds$, or indeed of $\dot{m}\, \ds$. Let us first then consider the first term in \equationref{VI}{7}, viz.\ $\nabla \smalliiint u V \nabla V \sigma \tau\, \ds$. Using \equationref{II}{9} \Secref{6}, and neglecting the surface integral at infinity, this becomes \begin{align*} &-\nabla \smalliiint V \nabla u V \sigma \tau\, \ds = \nabla V \nabla \smalliiint u V \sigma \tau\, \ds\qquad \tag*{[\Secref{9}]} \\ % &= \nabla^2 \smalliiint u V \sigma \tau\, \ds - \nabla S \nabla \smalliiint u V \sigma \tau\, \ds \\ % &= 4\pi V \sigma\tau + \nabla \smalliiint S\sigma\tau\, \nabla u\, \ds, \end{align*} by \equationref{II}{19} \Secref{10} above. %% -----File: 116.png---Folio 98------- Remembering now (\Secref{65} above), that \[ \dot{\sigma} = \partial\sigma/\partial t - V\sigma\tau - \nabla(\sigma^2)/2, \] we see by \equationref{VI}{7} that \[ \EqnTag{VI}{8} 4\pi\dot{\sigma} = -2\pi\nabla\centerdot \sigma^2 + \nabla \smalliiint S\sigma\tau\, \nabla u\, \ds + \nabla \smalliiint u (\partial m/\partial t)\, \ds, \] whence by \equationref{V}{9} \Secref{65} above \[ \EqnTag{VI}{9} P + v - \sigma^2/2 + (4 \pi)^{-1} \smalliiint (S\sigma\tau\, \nabla u + u\,\partial m/\partial t)\, \ds = H, \] where $H$ is a function of the time only. This is a generalisation of \equationref{V}{12} \Secref{65}, for assuming that $\tau = 0$ we know by \Secref{79} that $\partial\phi/\partial t = (4 \pi)^{-1}\! \smalliiint u(\partial m/\partial t)\, \ds$. \Paragraph{90} This integral equation may be put into several different forms by means of \equationref{II}{9} \Secref{6} above. The form that is useful to us is the one in which instead of $\partial m/\partial t$ we have $d(m\, \ds)/dt$ as we have already seen. Now $d(\ds)/dt = -m\, \ds$. Therefore \begin{align*} d(m\, \ds)/dt &= (\dot{m} - m^2) \, \ds \\ &= (\partial m/\partial t - S\sigma\,\nabla m - m^2)\, \ds. \end{align*} Substituting from this for $\partial m/\partial t$, and then transforming by \equationref{II}{9} \Secref{6} so as to get rid of the space variations of~$m$ involved in $S\sigma\,\nabla m$ we get \begin{align*} \smalliiint u\, (\partial m/\partial t)\, \ds &= \smalliiint u\,d(m\, \ds)/dt + \smalliiint (um^2 - mS\sigma\, \nabla u - u m^2)\, \ds \\ &= \smalliiint u\, d(m\, \ds)/dt - \smalliiint m S\sigma\, \nabla u\, \ds. \end{align*} Thus from equations \Eref{VI}{8} and~\Eref{VI}{9} \[ \EqnTag{VI}{10} 4\pi\dot{\sigma} = -2\pi\nabla \centerdot \sigma^2 + \nabla \smalliiint S \nabla u(V \sigma\tau - m\sigma)\, \ds + \nabla \smalliiint u\,d(m\, \ds)/dt, \] and \[ \EqnTag{VI}{11} P + v - \sigma^2/2 + (4\pi)^{-1} \smalliiint \{\ds\, S \nabla u(V\sigma\tau - m\sigma) + u\,d(m\, \ds)/dt\} = H. \] \TCSection{91}{Sir Wm.~Thomson's theory}{Sir Wm.~Thomson's Theory} \Paragraph{91} We are now in a position to examine the two theories. We first take Thomson's, which is considerably the simpler, and which therefore serves as an introduction to the other. We have then $m = 0$. Thus equations \Eref{VI}{10} and~\Eref{VI}{11} become \begin{align*} \EqnTag{VI}{12} &4\pi\dot{\sigma} = -2\pi\nabla\centerdot \sigma^2 + \nabla \smalliiint S\sigma\tau\, \nabla u\, \ds, \\ % \EqnTag{VI}{13} &p/D - \sigma^2/2 + (4\pi)^{-1} \smalliiint S\sigma\tau\, \nabla u\, \ds = H, \end{align*} for in the present theories we may put $v = 0$. We shall consider the two terms on the right of \equationref{VI}{12} separately. The second term gives then an apparent force per unit mass due to a potential \[ \EqnTag{VI}{14} -(4\pi)^{-1} \smalliiint S\sigma\tau\, \nabla u\, \ds. \] %% -----File: 117.png---Folio 99------- Comparing this with \equationref{IV}{33} \Secref{46} above we see that this potential is the same as that of a magnetic system given by \[ \EqnTag{VI}{15} 4 \pi \mathbf{I} = V \sigma \tau. \] Now this magnetism is zero where $\tau$ is zero. In other words it is only present where the vortex-atoms are, and therefore it cannot be so distributed as to give an apparent force of gravitation, for taking the view of magnetic matter expressed in \equationref{IV}{34} \Secref{46} we see that there must, in every complete vortex-atom, be a sum of magnetic matter exactly $= 0$. Nor again is it likely to explain the phenomena of permanent magnets, because, assuming that for any given small space including many vortex-atoms $\smalliiint V\sigma\tau\, \ds$ is not zero, the apparent force produced will affect all other parts of space independently of whether this same integral for them is not or is zero. But to explain the phenomena of permanent magnets we must assume that the effect takes place only on portions of space where there is positive magnetic matter. This term then gives us no phenomena analogous to physical phenomena. As a matter of fact it probably has no visible effect on large groups of vortices, for there is no reason to suppose that the vector~$V\sigma\tau$ is distributed otherwise than at random. \Paragraph{92} Let us now consider the other term in \equationref{VI}{12}, and neglecting the term already considered, put \[ \EqnTag{VI}{16} \dot{\sigma} = -\nabla\centerdot \sigma^2/2. \] The phenomena resulting from this are the same as would follow from a stress in a medium, the stress being an equal tension in all directions $= -\sigma^2/2$. Now comparing \equationref{IV}{62} \Secref{58} with \equationref{VI}{4} \Secref{89}, we see that $\sigma$ depends on $\tau/4\pi$ in exactly the same way as does~$\mathbf{H}$ on~$\mathbf{C}$. The question then arises---is the stress we are now considering equivalent in its mechanical effects upon the vortex-atoms to the stress given by \equationref{IV}{68} \Secref{60} above, which explains the mechanical effect of one current on another? We saw in \Secref{60} that this stress is a tension $-{\mathbf{H}}^2/8\pi$ along the vector~$\mathbf{H}$ and an equal pressure in all directions at right angles. The effects then would be the same only if $\sigma$~be at right angles to the surface of our atom. But this is obviously not in general the case. From this analogy we can see however what approximately will happen to our atom. For instead of $\sigma$~being at right angles to the surface it is in all probability very nearly tangential. Assuming that it is actually tangential we see that at the surface %% -----File: 118.png---Folio 100------- of the atom we have a tension exactly corresponding to the pressure which in the electric analogue will be exerted on this surface. In other words, the atoms will act on each other very approximately in what may be called a converse way to the small circuits in the electric analogue; i.e.\ where, in the electric analogue there is an attraction, in the hydrodynamic case there will be an apparent repulsion, and vice~versâ. Now each vortex-atom forms a small circuit and therefore acts in the converse way to a small magnet. In other words, each atom acts upon each other atom as if it were charged with \emph{attracting} magnetic matter. Thus we see that if we could suppose certain extra atomic vortices to exist and to be disposed throughout space in what at present must be considered quite an artificial manner with reference to the \emph{atomic} vortices we could rear up a fabric which would explain gravitation. This conception however is of very little use for our present purpose. From these considerations I think we have every reason to believe that Thomson's theory in its native simplicity does not promise to lead us to the physical phenomena of matter. We pass on therefore to Hicks's. \TCSection{93}{Prof.~Hicks's theory}{Prof.~Hicks's Theory} \Paragraph{93} Hicks in his theory, as I understand him, assumes that the bubbles always remain associated with the same particles of the fluid. This of course is probably not the case. By reason of the variation of the pressure with the time it is probable that evanescent bubbles start into existence and disappear at various parts of the fluid. This requires some few preliminary remarks. The particles of the fluid with which bubbles are permanently (i.e.\ throughout the greater part of each small but not infinitely small interval of time) associated are those where the intensity of spin is greatest. If the intensity of spin is quite various at different points we shall thus have vortices where there is generally no bubble, extra material vortices in fact. We must suppose these distributed quite at random till the more exact mathematical treatment of our problem leads us to suppose otherwise. Now evanescent bubbles will occur rather in these vortices than in parts of the fluid where is no vortex at all (if we may %% -----File: 119.png---Folio 101------- suppose such parts to exist), and of course they will occur more readily in stronger than in weaker vortices. At the present stage of the theory then we may suppose evanescent bubbles to occur in all parts of the fluid. As a first approximation to the consideration of the effect of these bubbles we may assume a part~$m'$ of~$m$ to be continuously distributed through space. Putting \[ \EqnTag{VI}{17} m = m' + M, \] we must suppose $M$ to be present only at the material bubbles where it is probably discontinuous, whereas $m'$~is continuous throughout both the material vortices and the rest of the fluid. Now when on account of variation of pressure $m'$ and~$M$ are affected---is it probable that in the neighbourhood of a permanent bubble $m'$ and~$M$ are of the same or opposite signs? To answer this question observe that what we call~$m'$ continuously distributed is really a series of discontinuous values of~$m$ scattered at random through space so that~$m'$ is probably very small. A decrement of pressure will cause an increment of evanescent bubbles, i.e.\ a decrement in~$m'$. An increment of the permanent bubbles will also take place the magnitude of which by what has been said concerning~$m'$ will not be by any means accounted for by the decrement in~$m'$. There will therefore also be a decrement in~$M$. Similarly for an increment in the pressure. $M$ and~$m'$ may therefore be assumed to be of the same sign. After noticing that $m'$~is continuous and therefore that there is no objection to introducing its space variations we are furnished with all the materials necessary for discussing our problem. The equation we shall use is~\Eref{VI}{10} of \Secref{90} above. We divide its discussion into two parts as follows. % [** PP: To prevent an overfull hbox, \Paragraph does not indent 94.] \Section{94}{Consideration of all the terms except \texorpdfstring{$-\nabla\centerdot(\sigma^2)/2$}{the last}}\Erratum{err101a}% [** PP: Added \centerdot] \Paragraph{94} The reasons that we have already seen in \Secref{91} for neglecting $\nabla \smalliiint S\sigma\tau\, \nabla u\, \ds$ still hold good so we put this aside. This is not the case with $-\nabla \smalliiint m S\sigma\, \nabla u\, \ds$ but we can neglect $\nabla \smalliiint u\, d(m\, \ds)/dt$ for the average value of $d(m\, \ds)/dt$ for any particle is zero. The only term to consider then is $-\nabla\smalliiint mS\sigma\, \nabla u\, \ds$. Putting as in \equationref{VI}{17} $m = m' + M$ we have \[ -\nabla \smalliiint M S\sigma\, \nabla u\, \ds -\nabla \smalliiint m' S\sigma\, \nabla u\, \ds. \] %% -----File: 120.png---Folio 102------- The first term of this can be neglected for the same reasons as for neglecting that containing~$V\sigma\tau$. Applying \equationref{II}{9} \Secref{6} to the last term and neglecting the surface integral as usual we get \[ \EqnTag{VI}{18} \nabla \smalliiint u S \nabla(m' \sigma)\, \ds = \nabla \smalliiint u \{m'(m' +M) + S\sigma\, \nabla m'\}\, \ds. \] The last term can probably be neglected though we cannot give such good reasons as for the other terms we have neglected. At any rate in places not near permanent bubbles $S\sigma\nabla m'$ is as likely to be positive as negative and vice~versâ, so that such portions of space will on the whole produce no effect on the permanent bubbles. If $S\sigma\,\nabla m'$ contributes anything for parts of space in the neighbourhood of permanent bubbles we must be content at present with the assumption either that the contribution is in general positive or that if it be negative it is not sufficient to cancel the effect of the positive term~$m'M$. Remembering that $m'$~is small compared with~$M$ we are left with the positive term~$m' M$. This as can be easily seen from the form of \equationref{VI}{18} leads to an apparent law of gravitation for our permanent bubbles. \Paragraph{95} The gravitational mass which we must on this supposition assign to each permanent bubble varies as the average value of~$m' M$ for that bubble\footnote{% There is one important difference to be noticed between this and Hicks's explanation of gravitation. His depends on the \emph{synchronous} pulsations of distant vortices. I do not wish to imply that I do not believe in the existence of such synchronous pulsations, but by the above we see that gravitation can probably be explained independently of them. As a matter of fact such synchronous pulsations probably actually occur on account of the variation of~$H$ with the time.}. Now we saw in \Secref{87} that the probable measure of mass of a permanent bubble was proportional to its average size. Do these two results agree? I cannot say, but even if they do not these considerations would still explain the motions of the solar system, but if the sun and Jupiter (say) were to collide their subsequent motion would not be that due to the collision of two bodies the ratio of whose masses is that which is accepted as the ratio of the sun's and Jupiter's. As a matter of fact however I should imagine that the average value of~$m'M$ for a permanent bubble is proportional to its average volume and this simply as a consequence of the reasoning in \Secref{87} above. A conclusion at any rate to be drawn from the above is that there is a \emph{presumption} in favour of Hicks's theory explaining gravitation. %% -----File: 121.png---Folio 103------- \Section{96}{Consideration of the \texorpdfstring{term $-\nabla\centerdot(\sigma^2)/2$}{final term}} \Paragraph{96} In considering this term we adopt the method of \Secref{92} and consider an electric analogue. The analogue is an electro-magnetic field for which in the notation of \Secref{46} to~\Secref{60} above at every point,\Erratum{err103a} % [** PP: Changed period to comma] \[ \EqnTag{VI}{20} \mathbf{H} = \sigma. \] For this field we have at once \[ \EqnTag{VI}{21} 4\pi \mathbf{C} = V\nabla \mathbf{H} = \tau. \] The distribution of the magnetism in the field is somewhat arbitrary, but in the notation of equations \Eref{IV}{61} and~\Eref{V}{62} \Secref{82} it will be found that everything is satisfied by putting \[ \EqnTag{VI}{22} 4\pi \mathbf{I} = -\nabla S q. \] This gives as it should $S\nabla (\mathbf{H} + 4\pi \mathbf{I}) = 0$, which is in fact the only equation it is necessary to satisfy. We have further \begin{flalign*} && \mathbf{B} = \mathbf{H} + 4\pi\mathbf{I} &= \sigma - \nabla Sq = \nabla Vq, && \\ % \EqnTag{VI}{23} &\text{so that}& \mathbf{A} &= Vq. && \end{flalign*} Thus all the important vectors in the analogue are determined. It remains to compare the mechanical effects of the analogue with the term $-\nabla\centerdot \sigma^2/2$. {\stretchyspace $\mathbf{I}$, it must be observed, is not confined to the bubbles, but is distributed throughout space.} \Paragraph{97} If we now assume that bubbles have not always existed in the positions which we call permanent, there cannot at the surface of the bubbles be any circulation round them. This makes the velocity at the surface almost normal to it, so that the stress given in \equationref{IV}{66} \Secref{60} reduces to a tension $-S\mathbf{H} (2\mathbf{B}-\mathbf{H})/8\pi$ over the surface, i.e.\ we have a pressure \[ -\sigma^2/8\pi + S\sigma\,\nabla V q/4\pi. \] Now on account of the absence of circulation $\nabla V q$ is very small and may therefore be neglected. Thus we get a \emph{pressure} $-\sigma^2/8\pi$, and are thus led once more to a ``converse'' of the analogue. This at once\footnote{% [Note added, 1892. Because the density of attracting magnetic matter of the analogue $= -S\nabla \mathbf{H}/4\pi = -m/4\pi$.]} leads to another reason for the law of gravitation if %% -----File: 122.png---Folio 104------- the pulsations are synchronous. This we have already seen to be probable. The present consideration of the subject is merely to point to a \emph{method} of investigating the theory of vortex-atoms. I therefore leave the subject here, not attempting to force the phenomena we have been considering to tally with the known phenomena of electricity and magnetism. Nevertheless I may say that the prospect of discussing these things by means of the present subject can scarcely be considered as distant after what has gone before. To sum up, this first application of the method leads to a presumption in favour of Hicks's theory leading to an explanation of both the important properties of matter---inertia and the law of gravitation---and there is also reason from it to hope that the phenomena of electro-magnetism are not unlikely to receive an explanation. Thomson's theory on the other hand would seem to fail in the first two at any rate of those endeavours. \Paragraph{98} We close the essay with the fulfilment of the promise made towards the end of \Secref{88}. In that section it will be remembered we considered a hypothetical fluid for which $D = \tanh (p/c)$, and made this do duty for a liquid containing bubbles. Strictly speaking our liquid is \emph{bounded} at the bubbles and therefore as a bounded liquid should it be treated. For such a liquid we require an equation corresponding to \equationref{VI}{10} \Secref{90}, and if possible \equationref{VI}{11} also. This last I have been unable to obtain, and I am not sure that to solve the problem explicitly is possible. Our problem only deals with an incompressible fluid, but as the removal of this restriction does not greatly complicate the work we will consider the general case of a bounded compressible fluid. We have \begin{align*} 4\pi\sigma &= \nabla^2\smalliiint u\sigma\, \ds = -\nabla \smalliiint \nabla u\sigma\, \ds \\ &= \nabla \smalliiint u(\tau + m)\, \ds - \nabla \smalliint u\, d\Sigma\,\sigma, \end{align*} whence \[ 4\pi\partial\sigma/\partial t = \nabla \smalliiint d \{u(\tau + m)\, \ds\}/dt -\nabla \smalliint d (u\, d\Sigma\, \sigma)/dt. \] The justification of using $\partial/\partial t$ on the left and~$d/dt$ under the integral sign will appear if the increment $(\partial\sigma/\partial t)\, dt$ in~$\sigma$ at a given point in the time~$dt$ be considered. It will be observed that the meaning here to be attached to~$\dot{u}$ will be $-S\sigma\, \nabla u$, as in the differentiation~$\partial/\partial t$ with regard to the time the origin of~$u$ is assumed to be fixed. %% -----File: 123.png---Folio 105------- I shall now merely indicate the method of procedure. By the method exhibited in \Secref{89} we can prove that \begin{align*} d(u\tau\, \ds)/dt &= -V\nabla u\,V\sigma\tau\, \ds - u\sigma S\tau\Delta\, \ds \\ \intertext{and that} d(um\, \ds)/dt &= -mS\sigma\, \nabla u\, \ds + u\,d(m\, \ds)/dt. \end{align*} From this we can deduce that \begin{align*} 4\pi\,\partial\sigma/\partial t &= \nabla V\, \nabla\smalliiint uV\sigma\tau\, \ds + \nabla\smalliiint \bigl\{-mS\sigma\, \nabla u\, \ds + u\,d(m\, \ds)/dt\bigr\} \\ &- \nabla\smalliint u\sigma S\tau\, d\Sigma - \nabla\smalliint u\, d(d\Sigma\,\sigma)/dt + \nabla\smalliint d\Sigma\, \sigma S\sigma\, \nabla u, \end{align*} and from this again we get \[ \EqnTag{VI}{24} \dot{\sigma} = \nabla w + \nabla w' + \nabla v, \] where $w$ and~$w'$ are scalars and~$v$ a vector given by \[ \EqnTag{VI}{25} 4\pi w = -2\pi\sigma^2 + \smalliiint\{\ds\,S\, \nabla u(V\sigma\tau - m\sigma) + u\,d(m\, \ds)/dt\}, \] so that $w$ is in fact the $H-v-P$ of \equationref{VI}{11} \Secref{90} \begin{align*} \EqnTag{VI}{26} 4\pi w' &= \smalliint(S\, d\Sigma\, \sigma S\sigma\, \nabla u - u\, dS\, d\Sigma\, \sigma/dt), \\ % \EqnTag{VI}{27} 4\pi v &= \smalliint(-u\sigma S\tau\, d\Sigma + V\,d\Sigma\,\sigma S\sigma\, \nabla u - u\, dV\, d\Sigma\, \sigma/dt). \end{align*} This last equation may be put in what for our purposes is the more convenient form \[ \EqnTag{VI}{28} 4\pi v = -\smalliint\{u\sigma S\tau\, d\Sigma + d(uV\, d\Sigma\, \sigma)/dt\}. \] Again it may be put in a form free from~$d/dt$; for $V\,d\Sigma\,\dot{\sigma} = 0$ because the surface is a free surface and $d(d\Sigma)/dt = \nabla_1 S\,d\Sigma\,\sigma_1$\footnote{% [Note added, 1892. This should be $d(d\Sigma)/dt = -m\,d\Sigma + \nabla_1 S\,d\Sigma\,\sigma_1$, and therefore \equationref{VI}{29} should be \[ 4\pi v = \smalliint\{V\, d\Sigma\, \sigma S\sigma\, \nabla u - u(\sigma S\tau\, d\Sigma - mV\, d\Sigma\, \sigma + V\nabla_1\sigma S\, d\Sigma\sigma_1)\}. \] This does not affect our present problem because $m=0$ in our case.]}, as can easily be proved by considerations similar to those in \Secref{83}. Thus \[ \EqnTag{VI}{29} 4\pi v = \smalliint(- u\sigma S\tau\, d\Sigma + V\,d \Sigma\, \sigma S\sigma\, \nabla u - uV\nabla_1\sigma S\, d\Sigma\, \sigma_1). \] In the problem we have to discuss $m = 0$, so that $w$ gives only terms which we discussed in considering Sir Wm.~Thomson's theory. Observing that if in~$w$ we change~$m\, \ds$ into $-S\,d\Sigma\,\sigma$ we get for the part of~$w$ containing~$m$, $w'$; we see that $w'$ only gives terms that we have virtually discussed under Hicks's theory. We have however entirely neglected~$v$. Are we justified in this? In the first place we have seen that if the bubbles have not always been associated with those parts of the fluid with which they now are there is round every bubble absolutely no circulation. This %% -----File: 124.png---Folio 106------- % [** PP: Keeping ``shows''] shows that for any one bubble $\smalliint V\, d\Sigma\, \sigma = 0$\footnote{% One way out of many of proving this is as follows. Divide the bubble up into a number of infinitely near sections by planes perpendicular to the unit vector~$\alpha$. For any one section $\int S\,d\rho\,\sigma=0$. Consider~$d\Sigma$ to be the element of the surface cut off by the following four planes, (1)~the plane of section considered, (2)~the consecutive plane of section, (3)~the two planes perpendicular to~$d\rho$, and through the extremities of the element~$d\rho$. Thus if $x$~be the distance between the two sections, $d\rho=x^{-1}V\alpha\, d\Sigma$, whence \[ \smalliint S\alpha\, d\Sigma\,\sigma=0, \] for the surface of the bubble between the two sections. But adding, we may suppose this integral to extend over the whole bubble. Thus $S\alpha\smalliint V\,d\Sigma\,\sigma=0$ for the whole bubble; therefore $\alpha$ being a quite arbitrary unit vector we have for the whole bubble $\smalliint V\, d\Sigma\, \sigma=0$.}, %*end of footnote text and therefore we are justified in neglecting the last term in \equationref{VI}{28}. We are probably also justified in neglecting the first term, for probably $\tau$ is very nearly tangential to the surface and therefore $S\tau\, d\Sigma = 0$. [Note added, 1892. The whole of this last section is in rather a nebulous stage, and since writing it I have not had sufficient leisure to return to the matter. I hesitated whether to include it in the present issue. But since, notwithstanding the absence of any reliable results, it serves very well to illustrate how investigations are conducted by Quaternions, I have thought it worth publishing. Should anybody feel inclined to attempt to apply the method or an analogous one it is well to note that in the \textit{Phil.\ Mag.}\ June, 1892, p.~490, I have given the more general result sought in this last section. As the result is not there proved I give one proof which seems instructive. It exhibits the great variety of suitable quaternion methods of dealing with physical questions. It furnishes incidentally a fourth quaternion proof of the properties of vortices. It also illustrates how special quaternion methods developed for use in one branch of Physics at once prove themselves useful in other branches. Adopting the notation and terminology of \textit{Phil.\ Trans.}\ 1892, p.~686, \S\S~5--7, let $\sigma$ and~$\tau$ be taken as an intensity and flux respectively, $\tau$~still being ${} = V\, \nabla\sigma$ and therefore $\tau' = V\, \nabla'\sigma'$. Thus $\sigma'$ and~$\tau'$ are the actual velocity and double spin respectively and the equation of motion is \[ \dot{\sigma}' = -\nabla'(v+P). \] Putting $\sigma' = \chi'^{-1}\sigma$ and operating on the equation by~$\chi'$ \[ \dot{\sigma} +\chi' d\left(\chi'^{-1}\right)/dt\centerdot \sigma = -\nabla(v+P). \] %% -----File: 125.png---Folio 107------- Now \begin{gather*} \chi'd(\chi'^{-1})/dt\centerdot \sigma = -\dot{\chi}'\chi'^{-1}\sigma = -\dot{\chi}'\sigma' = \nabla_1 S\sigma_1'\sigma' = \nabla\centerdot \sigma'^2/2. \\ % \EqnTag{VI}{$A$} \therefore\qquad \dot{\sigma} + \nabla\centerdot \sigma'^2/2 = -\nabla(v+P), \end{gather*} which can easily be deduced from or utilised to prove Lord Kelvin's theorem concerning ``flow,'' \equationref{V}{23} \Secref{72} above. As $d/dt$ is commutative with~$\nabla$, $\dot{\tau}=0$ or $\tau$ is an absolute constant for each element of matter. This being interpreted at once gives the well-known properties of vortices in their usual form. If in the equation $4\pi(v+P)=S\centerdot \nabla^2\smalliiint u(v+P)\, \ds$ we carry one~$\nabla$ across the integral sign, get rid of its differentiations which affect~$u$ by \equationref{II}{9} \Secref{6} above, and then do the same with the other~$\nabla$ we get \[ 4\pi(v+P) = \smalliint\{(v+P)S\, d\Sigma\, \nabla u - uS\,d\Sigma\,\nabla(v+P)\} +\smalliiint u\nabla^2(v+P)\, \ds. \] At surfaces of discontinuity in $\sigma$, $v$ and~$P$ will both be continuous, so that instead of $\smalliint(v+P)S\, d\Sigma\,\nabla u$ we may write $\smalliint_b(v+P)S\,d\Sigma\,\nabla u$. In the last equation substitute throughout for $\nabla(v + P)$ from \equationref{VI}{$A$}. Thus \Erratum{err107a}% [** PP: Added \centerdot below] \begin{gather*} \begin{aligned} 4\pi(v+P) = \smalliint_b (v+P)S\, d\Sigma\, \nabla u &+ \smalliint uS\, d\Sigma\, \dot{\sigma} -\smalliiint uS\, \nabla\dot{\sigma}\, \ds \\ &+ \left(\smalliint uS\, d\Sigma\, \nabla\centerdot \sigma'^2/2 -\smalliiint u\nabla^2\sigma'^2\, \ds/2\right) \end{aligned} \\ {} = \smalliint_b (v+P)S\, d\Sigma\, \nabla u +\smalliint uS\, d\Sigma\, \dot{\sigma} -\smalliiint uS\nabla\dot{\sigma}\, \ds +\smalliiint S\, \nabla u\, \nabla\centerdot(\sigma'^2)\, \ds/2,% [** <- \centerdot] \end{gather*} which is equation~(36) of the \textit{Phil.\ Mag.}\ paper. If the standard position and present position of matter coincide it is quite easy to prove that \begin{align*} dS\, d\Sigma'\, \sigma'/dt &= S\, d\Sigma(\dot{\sigma} + V\nabla_1\sigma\sigma_1) \\ d(S\nabla'\sigma'\, \ds')/dt &= S\nabla(\dot{\sigma} + V\nabla_1\sigma\sigma_1)\, \ds. \end{align*} Substituting for $S\,d\Sigma\,\dot{\sigma}$, $S\nabla\dot{\sigma}$ from these in the last equation we get equation~(32) of the \textit{Phil.\ Mag.}\ paper; but this equation can also be proved directly. It should be noticed that equations (34) and~(35) of that paper have been wrongly written down from equation~(32). In each read ${} + \nabla\centerdot(\sigma^2/2)$ for ${} - \nabla\centerdot(\sigma^2/2)$.]\Erratum{err107b}% [** PP: Added two \centerdots] \cleardoublepage \enlargethispage{50pt} \SectionTitle{Transcriber's Notes} \pdfbookmark[0]{Transcriber's Notes}{Transcriber's Notes} \fancyhead{} \fancyhead[C]{{\scshape\small transcriber's notes.}} \ifthenelse{\boolean{ForPrinting}} {\fancyhead[RO,LE]{\thepage}} {\fancyhead[R]{\thepage}} Minor spelling inconsistencies (\textit{foot-note}/\textit{footnote}, \textit{neutralise}/\textit{neutralize}, and \textit{shew}/\textit{show}) have been retained. Changes listed below are marked with double asterisks in the \LaTeX\ file. \smallskip %% Changes of spelling: % caligraphy -> calligraphy (p. ix) % paralleled -> parallelled (p. vii) The punctuation on displayed equations has been placed at the end of the equation instead of after the tag at the right margin. Aside from this, equation punctuation has been retained unless it conflicts obtrusively with grammar. \bigskip Page numbers refer to the folio numbers in the scanned images. \begin{itemize} \item p.~7: \hyperref[err7a]{Set long verbal equation as a display.} \\ (``$w\times\text{the element of volume}\ldots$'') \item p.~33: \hyperref[err33a]{Just after equation~($15f$), add missing word.} (``We saw in \emph{the} last section\dots'') \item p.~49: \hyperref[err49a]{Replace opening equals sign by ``is equal to'' just above \S\,29.} (Two times.) \item p.~62: \hyperref[err62a]{Replace ``$\mathrm{D}$'' by~``$\Dop$'' in right-hand member of display following equation~(19).} \item p.~101: \hyperref[err101a]{Add dot in title of \S\,94 (``$-\nabla\centerdot(\sigma^2)/2$'').} \item p.~107: Add missing dot after~$\nabla$ three times: in last member of \hyperref[err107a]{second-to-last display}, and twice in \hyperref[err107b]{last line.} \end{itemize} %% List of minor changes: \iffalse \item p.~6: \hyperref[err6a]{Re-break displayed equation at second equals sign.} \\ (``$A=a(f+g)+bc/2=\cdots$'') \item p.~16: \hyperref[err16a]{Add period to first line of equation~(5).} \item p.~17: \hyperref[err6f]{Re-break equation~(6$f$).} \item p.~18: \hyperref[err18a]{Add comma to equation just before equation~($6g$).} \item p.~27: \hyperref[err27a]{Remove comma in middle of display in first paragraph of \S\,14.} \\ $\bigl($``$(\ldots\text{on surface of portion considered}) - (\text{work done}\ldots)$''$\bigr)$ \item p.~39: \hyperref[err39a]{Add missing closing square bracket at end of footnote.} \item p.~64: \hyperref[err64a]{Hyphenate ``electro-motive'' twice in the final paragraph of \S\,42.} \item p.~67: \hyperref[err67a]{Reformat title of \S\,54 for consistency with \S\S\,34 and~46.} \item p.~103: \hyperref[err103a]{Change period to comma just before equation~(20).} \fi \cleardoublepage \phantomsection \pdfbookmark[0]{Project Gutenberg License}{Project Gutenberg License} \fancyhead[C]{{\scshape\small licensing.}} \begin{PGtext} End of the Project Gutenberg EBook of Utility of Quaternions in Physics, by Alexander McAulay *** END OF THIS PROJECT GUTENBERG EBOOK UTILITY OF QUATERNIONS IN PHYSICS *** ***** This file should be named 26262-pdf.pdf or 26262-pdf.zip ***** This and all associated files of various formats will be found in: http://www.gutenberg.org/2/6/2/6/26262/ Produced by Joshua Hutchinson, Andrew D. Hwang, Carolyn Bottomley and the Online Distributed Proofreading Team at http://www.pgdp.net (This ebook was produced from images from the Cornell University Library: Historical Mathematics Monographs collection.) 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, is 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{PGtext} % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % % % % End of the Project Gutenberg EBook of Utility of Quaternions in Physics, by % Alexander McAulay % % % % *** END OF THIS PROJECT GUTENBERG EBOOK UTILITY OF QUATERNIONS IN PHYSICS *** % % % ***** This file should be named 26262-t.tex or 26262-t.zip ***** % % This and all associated files of various formats will be found in: % % http://www.gutenberg.org/2/6/2/6/26262/ % % % % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % \end{document} ### lprep configuration @ControlwordReplace = ( ['\\Preface',"\n\nPreface.\n"], ['\\McAulay','McAULAY'], ['\\TransNote',"Transcriber's Note"], ['\\TransNoteText',"\nThis book may be easily recompiled for printing. Please see the\nsource file preamble for instructions.\n"] ); @ControlwordArguments = ( ['\\HeadChapter',1, 1, "\n\n","\n",1,1,'',"\n\n",1,0,'',''], ['\\FNChapter',1, 1, "\n\n","\n",1,0,'','',1,1,'',"\n\n"], ['\\Chapter',1, 1, "\n\n","\n",1,1,'',"\n\n"], ['\\Tocline',1,0,'','',1,0,'',''], ['\\TCSection',1,0,'','',1,0,'',"\n"], ['\\Section',1,0,'',"\n"], ['\\Paragraph',1,1,'','.'], ['\\FNParagraph',1,0,'','',1,1,'','.'], ['\\Eref',1,0,'','',1,1,'(',')'], ['\\equationref',1,0,'','',1,1,'equation~(',')'], ['\\Equationref',1,0,'','',1,1,'Equation~(',')'], ['\\Secref',1,1,'\\S~',''], ['\\Erratum',1,0,'',''], ['\\hyperref',0,0,'',''] ); ### This is pdfeTeX, Version 3.141592-1.30.5-2.2 (Web2C 7.5.5) (format=pdflatex 2008.5.6) 11 AUG 2008 02:47 entering extended mode **26262-t.tex (./26262-t.tex LaTeX2e <2003/12/01> Babel and hyphenation patterns for english, usenglishmax, dumylang, noh yphenation, greek, monogreek, ancientgreek, ibycus, loaded. (/usr/share/texmf-texlive/tex/latex/base/book.cls Document Class: book 2004/02/16 v1.4f Standard LaTeX document class (/usr/share/texmf-texlive/tex/latex/base/bk12.clo File: bk12.clo 2004/02/16 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 ) LaTeX Warning: You have requested, on input line 149, version `2005/09/16' of document class book, but only version `2004/02/16 v1.4f Standard LaTeX document class' is available. (/usr/share/texmf-texlive/tex/latex/amslatex/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/amslatex/amstext.sty Package: amstext 2000/06/29 v2.01 (/usr/share/texmf-texlive/tex/latex/amslatex/amsgen.sty File: amsgen.sty 1999/11/30 v2.0 \@emptytoks=\toks14 \ex@=\dimen103 )) (/usr/share/texmf-texlive/tex/latex/amslatex/amsbsy.sty Package: amsbsy 1999/11/29 v1.2d \pmbraise@=\dimen104 ) (/usr/share/texmf-texlive/tex/latex/amslatex/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/base/ifthen.sty Package: ifthen 2001/05/26 v1.1c Standard LaTeX ifthen package (DPC) ) (/usr/share/texmf-texlive/tex/latex/base/inputenc.sty Package: inputenc 2004/02/05 v1.0d Input encoding file (/usr/share/texmf-texlive/tex/latex/base/latin1.def File: latin1.def 2004/02/05 v1.0d Input encoding file )) LaTeX Warning: You have requested, on input line 155, version `2006/05/05' of package inputenc, but only version `2004/02/05 v1.0d Input encoding file' is available. (/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@=\toks19 ) (/usr/share/texmf-texlive/tex/latex/graphics/graphics.sty Package: graphics 2001/07/07 v1.0n 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) ) (/usr/share/texmf-texlive/tex/latex/config/graphics.cfg File: graphics.cfg 2001/08/31 v1.1 graphics configuration of teTeX/TeXLive ) Package graphics Info: Driver file: pdftex.def on input line 80. (/usr/share/texmf-texlive/tex/latex/graphics/pdftex.def File: pdftex.def 2005/06/20 v0.03m graphics/color for pdftex \Gread@gobject=\count100 )) \Gin@req@height=\dimen112 \Gin@req@width=\dimen113 ) (/usr/share/texmf-texlive/tex/latex/base/alltt.sty Package: alltt 1997/06/16 v2.0g defines alltt environment ) (/usr/share/texmf-texlive/tex/latex/tools/indentfirst.sty Package: indentfirst 1995/11/23 v1.03 Indent first paragraph (DPC) ) (/usr/share/texmf-texlive/tex/latex/footmisc/footmisc.sty Package: footmisc 2005/03/17 v5.3d a miscellany of footnote facilities \FN@temptoken=\toks20 \footnotemargin=\dimen114 \c@pp@next@reset=\count101 Package footmisc Info: Declaring symbol style bringhurst on input line 817. Package footmisc Info: Declaring symbol style chicago on input line 818. Package footmisc Info: Declaring symbol style wiley on input line 819. Package footmisc Info: Declaring symbol style lamport-robust on input line 823. Package footmisc Info: Declaring symbol style lamport* on input line 831. Package footmisc Info: Declaring symbol style lamport*-robust on input line 840 . ) (/usr/share/texmf-texlive/tex/latex/ltxmisc/perpage.sty Package: perpage 2002/12/20 v1.0 Reset counters per page ) LaTeX Warning: You have requested, on input line 169, version `2006/07/15' of package perpage, but only version `2002/12/20 v1.0 Reset counters per page' is available. (/usr/share/texmf-texlive/tex/latex/txfonts/txfonts.sty Package: txfonts 2005/01/03 v3.2 LaTeX Font Info: Redeclaring symbol font `operators' on input line 20. LaTeX Font Info: Overwriting symbol font `operators' in version `normal' (Font) OT1/cmr/m/n --> OT1/txr/m/n on input line 20. LaTeX Font Info: Overwriting symbol font `operators' in version `bold' (Font) OT1/cmr/bx/n --> OT1/txr/m/n on input line 20. LaTeX Font Info: Overwriting symbol font `operators' in version `bold' (Font) OT1/txr/m/n --> OT1/txr/bx/n on input line 21. \symitalic=\mathgroup6 LaTeX Font Info: Overwriting symbol font `italic' in version `bold' (Font) OT1/txr/m/it --> OT1/txr/bx/it on input line 25. LaTeX Font Info: Redeclaring math alphabet \mathrm on input line 27. LaTeX Font Info: Redeclaring math alphabet \mathbf on input line 28. LaTeX Font Info: Overwriting math alphabet `\mathbf' in version `normal' (Font) OT1/cmr/bx/n --> OT1/txr/bx/n on input line 28. LaTeX Font Info: Overwriting math alphabet `\mathbf' in version `bold' (Font) OT1/cmr/bx/n --> OT1/txr/bx/n on input line 28. LaTeX Font Info: Redeclaring math alphabet \mathit on input line 29. LaTeX Font Info: Overwriting math alphabet `\mathit' in version `normal' (Font) OT1/cmr/m/it --> OT1/txr/m/it on input line 29. LaTeX Font Info: Overwriting math alphabet `\mathit' in version `bold' (Font) OT1/cmr/bx/it --> OT1/txr/m/it on input line 29. LaTeX Font Info: Overwriting math alphabet `\mathit' in version `bold' (Font) OT1/txr/m/it --> OT1/txr/bx/it on input line 30. LaTeX Font Info: Redeclaring math alphabet \mathsf on input line 39. LaTeX Font Info: Overwriting math alphabet `\mathsf' in version `normal' (Font) OT1/cmss/m/n --> OT1/txss/m/n on input line 39. LaTeX Font Info: Overwriting math alphabet `\mathsf' in version `bold' (Font) OT1/cmss/bx/n --> OT1/txss/m/n on input line 39. LaTeX Font Info: Overwriting math alphabet `\mathsf' in version `bold' (Font) OT1/txss/m/n --> OT1/txss/b/n on input line 40. LaTeX Font Info: Redeclaring math alphabet \mathtt on input line 49. LaTeX Font Info: Overwriting math alphabet `\mathtt' in version `normal' (Font) OT1/cmtt/m/n --> OT1/txtt/m/n on input line 49. LaTeX Font Info: Overwriting math alphabet `\mathtt' in version `bold' (Font) OT1/cmtt/m/n --> OT1/txtt/m/n on input line 49. LaTeX Font Info: Overwriting math alphabet `\mathtt' in version `bold' (Font) OT1/txtt/m/n --> OT1/txtt/b/n on input line 50. LaTeX Font Info: Redeclaring symbol font `letters' on input line 57. LaTeX Font Info: Overwriting symbol font `letters' in version `normal' (Font) OML/cmm/m/it --> OML/txmi/m/it on input line 57. LaTeX Font Info: Overwriting symbol font `letters' in version `bold' (Font) OML/cmm/b/it --> OML/txmi/m/it on input line 57. LaTeX Font Info: Overwriting symbol font `letters' in version `bold' (Font) OML/txmi/m/it --> OML/txmi/bx/it on input line 58. \symlettersA=\mathgroup7 LaTeX Font Info: Overwriting symbol font `lettersA' in version `bold' (Font) U/txmia/m/it --> U/txmia/bx/it on input line 66. LaTeX Font Info: Redeclaring math alphabet \mathfrak on input line 69. LaTeX Font Info: Redeclaring symbol font `symbols' on input line 76. LaTeX Font Info: Overwriting symbol font `symbols' in version `normal' (Font) OMS/cmsy/m/n --> OMS/txsy/m/n on input line 76. LaTeX Font Info: Overwriting symbol font `symbols' in version `bold' (Font) OMS/cmsy/b/n --> OMS/txsy/m/n on input line 76. LaTeX Font Info: Overwriting symbol font `symbols' in version `bold' (Font) OMS/txsy/m/n --> OMS/txsy/bx/n on input line 77. LaTeX Font Info: Redeclaring symbol font `AMSa' on input line 92. LaTeX Font Info: Overwriting symbol font `AMSa' in version `normal' (Font) U/msa/m/n --> U/txsya/m/n on input line 92. LaTeX Font Info: Overwriting symbol font `AMSa' in version `bold' (Font) U/msa/m/n --> U/txsya/m/n on input line 92. LaTeX Font Info: Overwriting symbol font `AMSa' in version `bold' (Font) U/txsya/m/n --> U/txsya/bx/n on input line 93. LaTeX Font Info: Redeclaring symbol font `AMSb' on input line 101. LaTeX Font Info: Overwriting symbol font `AMSb' in version `normal' (Font) U/msb/m/n --> U/txsyb/m/n on input line 101. LaTeX Font Info: Overwriting symbol font `AMSb' in version `bold' (Font) U/msb/m/n --> U/txsyb/m/n on input line 101. LaTeX Font Info: Overwriting symbol font `AMSb' in version `bold' (Font) U/txsyb/m/n --> U/txsyb/bx/n on input line 102. LaTeX Font Info: Redeclaring math alphabet \mathbb on input line 105. \symsymbolsC=\mathgroup8 LaTeX Font Info: Overwriting symbol font `symbolsC' in version `bold' (Font) U/txsyc/m/n --> U/txsyc/bx/n on input line 112. LaTeX Font Info: Redeclaring symbol font `largesymbols' on input line 119. LaTeX Font Info: Overwriting symbol font `largesymbols' in version `normal' (Font) OMX/cmex/m/n --> OMX/txex/m/n on input line 119. LaTeX Font Info: Overwriting symbol font `largesymbols' in version `bold' (Font) OMX/cmex/m/n --> OMX/txex/m/n on input line 119. LaTeX Font Info: Overwriting symbol font `largesymbols' in version `bold' (Font) OMX/txex/m/n --> OMX/txex/bx/n on input line 120. \symlargesymbolsA=\mathgroup9 LaTeX Font Info: Overwriting symbol font `largesymbolsA' in version `bold' (Font) U/txexa/m/n --> U/txexa/bx/n on input line 128. LaTeX Info: Redefining \not on input line 1042. ) (/usr/share/texmf-texlive/tex/latex/fancyhdr/fancyhdr.sty \fancy@headwidth=\skip46 \f@ncyO@elh=\skip47 \f@ncyO@erh=\skip48 \f@ncyO@olh=\skip49 \f@ncyO@orh=\skip50 \f@ncyO@elf=\skip51 \f@ncyO@erf=\skip52 \f@ncyO@olf=\skip53 \f@ncyO@orf=\skip54 ) (/usr/share/texmf-texlive/tex/latex/fancyhdr/extramarks.sty \@temptokenb=\toks21 ) (/usr/share/texmf-texlive/tex/latex/geometry/geometry.sty Package: geometry 2002/07/08 v3.2 Page Geometry \Gm@cnth=\count102 \Gm@cntv=\count103 \c@Gm@tempcnt=\count104 \Gm@bindingoffset=\dimen115 \Gm@wd@mp=\dimen116 \Gm@odd@mp=\dimen117 \Gm@even@mp=\dimen118 \Gm@dimlist=\toks22 (/usr/share/texmf-texlive/tex/latex/geometry/geometry.cfg)) (/usr/share/texmf-t exlive/tex/latex/hyperref/hyperref.sty Package: hyperref 2003/11/30 v6.74m Hypertext links for LaTeX \@linkdim=\dimen119 \Hy@linkcounter=\count105 \Hy@pagecounter=\count106 (/usr/share/texmf-texlive/tex/latex/hyperref/pd1enc.def File: pd1enc.def 2003/11/30 v6.74m Hyperref: PDFDocEncoding definition (HO) ) (/usr/share/texmf-texlive/tex/latex/config/hyperref.cfg File: hyperref.cfg 2002/06/06 v1.2 hyperref configuration of TeXLive ) Package hyperref Info: Option `hyperfootnotes' set `false' on input line 1830. Package hyperref Info: Option `bookmarks' set `true' on input line 1830. Package hyperref Info: Option `linktocpage' set `true' on input line 1830. Package hyperref Info: Option `pdfdisplaydoctitle' set `true' on input line 183 0. Package hyperref Info: Option `pdfpagelabels' set `true' on input line 1830. Package hyperref Info: Option `bookmarksopen' set `true' on input line 1830. Package hyperref Info: Option `colorlinks' set `true' on input line 1830. Package hyperref Info: Hyper figures OFF on input line 1880. Package hyperref Info: Link nesting OFF on input line 1885. Package hyperref Info: Hyper index ON on input line 1888. Package hyperref Info: Plain pages ON on input line 1893. Package hyperref Info: Backreferencing OFF on input line 1900. Implicit mode ON; LaTeX internals redefined Package hyperref Info: Bookmarks ON on input line 2004. (/usr/share/texmf-texlive/tex/latex/ltxmisc/url.sty \Urlmuskip=\muskip11 Package: url 2005/06/27 ver 3.2 Verb mode for urls, etc. ) LaTeX Info: Redefining \url on input line 2143. \Fld@menulength=\count107 \Field@Width=\dimen120 \Fld@charsize=\dimen121 \Choice@toks=\toks23 \Field@toks=\toks24 Package hyperref Info: Hyper figures OFF on input line 2618. Package hyperref Info: Link nesting OFF on input line 2623. Package hyperref Info: Hyper index ON on input line 2626. Package hyperref Info: backreferencing OFF on input line 2633. Package hyperref Info: Link coloring ON on input line 2636. \Hy@abspage=\count108 \c@Item=\count109 ) *hyperref using driver hpdftex* (/usr/share/texmf-texlive/tex/latex/hyperref/hpdftex.def File: hpdftex.def 2003/11/30 v6.74m Hyperref driver for pdfTeX (/usr/share/texmf-texlive/tex/latex/psnfss/pifont.sty Package: pifont 2004/09/15 PSNFSS-v9.2 Pi font support (SPQR) LaTeX Font Info: Try loading font information for U+pzd on input line 63. (/usr/share/texmf-texlive/tex/latex/psnfss/upzd.fd File: upzd.fd 2001/06/04 font definitions for U/pzd. ) LaTeX Font Info: Try loading font information for U+psy on input line 64. (/usr/share/texmf-texlive/tex/latex/psnfss/upsy.fd File: upsy.fd 2001/06/04 font definitions for U/psy. )) \Fld@listcount=\count110 \@outlinefile=\write3 ) LaTeX Warning: You have requested, on input line 257, version `2007/02/07' of package hyperref, but only version `2003/11/30 v6.74m Hypertext links for LaTeX' is available. \c@pcabs@footnote=\count111 \Tocskip=\skip55 (./26262-t.aux) \openout1 = `26262-t.aux'. LaTeX Font Info: Checking defaults for OML/txmi/m/it on input line 462. LaTeX Font Info: Try loading font information for OML+txmi on input line 462 . (/usr/share/texmf-texlive/tex/latex/txfonts/omltxmi.fd File: omltxmi.fd 2000/12/15 v3.1 ) LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for T1/cmr/m/n on input line 462. LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for OT1/cmr/m/n on input line 462. LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for OMS/txsy/m/n on input line 462. LaTeX Font Info: Try loading font information for OMS+txsy on input line 462 . (/usr/share/texmf-texlive/tex/latex/txfonts/omstxsy.fd File: omstxsy.fd 2000/12/15 v3.1 ) LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for OMX/txex/m/n on input line 462. LaTeX Font Info: Try loading font information for OMX+txex on input line 462 . (/usr/share/texmf-texlive/tex/latex/txfonts/omxtxex.fd File: omxtxex.fd 2000/12/15 v3.1 ) LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for U/txexa/m/n on input line 462. LaTeX Font Info: Try loading font information for U+txexa on input line 462. (/usr/share/texmf-texlive/tex/latex/txfonts/utxexa.fd File: utxexa.fd 2000/12/15 v3.1 ) LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Checking defaults for PD1/pdf/m/n on input line 462. LaTeX Font Info: ... okay on input line 462. LaTeX Font Info: Try loading font information for OT1+txr on input line 462. (/usr/share/texmf-texlive/tex/latex/txfonts/ot1txr.fd File: ot1txr.fd 2000/12/15 v3.1 ) (/usr/share/texmf-texlive/tex/context/base/supp-pdf.tex (/usr/share/texmf-tex live/tex/context/base/supp-mis.tex loading : Context Support Macros / Miscellaneous (2004.10.26) \protectiondepth=\count112 \scratchcounter=\count113 \scratchtoks=\toks25 \scratchdimen=\dimen122 \scratchskip=\skip56 \scratchmuskip=\muskip12 \scratchbox=\box28 \scratchread=\read1 \scratchwrite=\write4 \zeropoint=\dimen123 \onepoint=\dimen124 \onebasepoint=\dimen125 \minusone=\count114 \thousandpoint=\dimen126 \onerealpoint=\dimen127 \emptytoks=\toks26 \nextbox=\box29 \nextdepth=\dimen128 \everyline=\toks27 \!!counta=\count115 \!!countb=\count116 \recursecounter=\count117 ) loading : Context Support Macros / PDF (2004.03.26) \nofMPsegments=\count118 \nofMParguments=\count119 \MPscratchCnt=\count120 \MPscratchDim=\dimen129 \MPnumerator=\count121 \everyMPtoPDFconversion=\toks28 ) -------------------- Geometry parameters paper: letterpaper landscape: -- twocolumn: -- twoside: true asymmetric: -- h-parts: 90.91446pt, 0.63\paperwidth , 136.3717pt v-parts: 95.39737pt, 0.7\paperheight , 143.09605pt hmarginratio: 2:3 vmarginratio: 2:3 lines: -- heightrounded: -- bindingoffset: 0.0pt truedimen: -- includehead: -- includefoot: -- includemp: -- driver: pdftex -------------------- Page layout dimensions and switches \paperwidth 614.295pt \paperheight 794.96999pt \textwidth 387.00883pt \textheight 556.47656pt \oddsidemargin 18.64447pt \evensidemargin 64.10172pt \topmargin -8.74641pt \headheight 12.0pt \headsep 19.8738pt \footskip 30.0pt \marginparwidth 98.0pt \marginparsep 7.0pt \columnsep 10.0pt \skip\footins 10.8pt plus 4.0pt minus 2.0pt \hoffset 0.0pt \voffset 0.0pt \mag 1000 \@twosidetrue \@mparswitchtrue (1in=72.27pt, 1cm=28.45pt) ----------------------- (/usr/share/texmf-texlive/tex/latex/graphics/color.sty Package: color 1999/02/16 v1.0i Standard LaTeX Color (DPC) LaTeX Info: Redefining \color on input line 71. (/usr/share/texmf-texlive/tex/latex/config/color.cfg File: color.cfg 2001/08/31 v1.1 color configuration of teTeX/TeXLive ) Package color Info: Driver file: pdftex.def on input line 125. ) Package hyperref Info: Link coloring ON on input line 462. (/usr/share/texmf-texlive/tex/latex/hyperref/nameref.sty Package: nameref 2003/12/03 v2.21 Cross-referencing by name of section \c@section@level=\count122 ) LaTeX Info: Redefining \ref on input line 462. LaTeX Info: Redefining \pageref on input line 462. (./26262-t.out) (./26262-t.out) \openout3 = `26262-t.out'. LaTeX Font Info: Try loading font information for OT1+txtt on input line 471 . (/usr/share/texmf-texlive/tex/latex/txfonts/ot1txtt.fd File: ot1txtt.fd 2000/12/15 v3.1 ) LaTeX Font Info: Try loading font information for U+txsya on input line 492. (/usr/share/texmf-texlive/tex/latex/txfonts/utxsya.fd File: utxsya.fd 2000/12/15 v3.1 ) LaTeX Font Info: Try loading font information for U+txsyb on input line 492. (/usr/share/texmf-texlive/tex/latex/txfonts/utxsyb.fd File: utxsyb.fd 2000/12/15 v3.1 ) LaTeX Font Info: Try loading font information for U+txmia on input line 492. (/usr/share/texmf-texlive/tex/latex/txfonts/utxmia.fd File: utxmia.fd 2000/12/15 v3.1 ) LaTeX Font Info: Try loading font information for U+txsyc on input line 492. (/usr/share/texmf-texlive/tex/latex/txfonts/utxsyc.fd File: utxsyc.fd 2000/12/15 v3.1 ) [1 {/var/lib/texmf/fonts/map/pdftex/updmap/pdftex.map}] [2 ] ! pdfTeX warning (ext4): destination with the same identifier (name{page.1}) ha s been already used, duplicate ignored \penalty l.573 \clearpage [1] <./images/pubmark.pdf, id=244, 469.755pt x 134.5025pt> File: ./images/pubmark.pdf Graphic file (type pdf) ! pdfTeX warning (ext4): destination with the same identifier (name{page.2}) ha s been already used, duplicate ignored \penalty l.586 \clearpage [2 <./images/pubmark.pdf>] [3 ] [4 ] ! pdfTeX warning (ext4): destination with the same identifier (name{page.1}) ha s been already used, duplicate ignored \penalty l.738 [1 ] Underfull \vbox (badness 1874) has occurred while \output is active [] ! pdfTeX warning (ext4): destination with the same identifier (name{page.2}) ha s been already used, duplicate ignored \penalty l.808 [2] Underfull \vbox (badness 1874) has occurred while \output is active [] ! pdfTeX warning (ext4): destination with the same identifier (name{page.3}) ha s been already used, duplicate ignored \penalty l.861 [3] ! pdfTeX warning (ext4): destination with the same identifier (name{page.4}) ha s been already used, duplicate ignored \penalty l.907 [4] LaTeX Font Info: Try loading font information for OMS+txr on input line 909. (/usr/share/texmf-texlive/tex/latex/txfonts/omstxr.fd File: omstxr.fd 2000/12/15 v3.1 ) LaTeX Font Info: Font shape `OMS/txr/m/n' in size <12> not available (Font) Font shape `OMS/txsy/m/n' tried instead on input line 909. Underfull \vbox (badness 6332) has occurred while \output is active [] [5] [6] (./26262-t.toc [7 ]) \tf@toc=\write5 \openout5 = `26262-t.toc'. [8] LaTeX Font Info: Font shape `OMS/txr/m/n' in size <10> not available (Font) Font shape `OMS/txsy/m/n' tried instead on input line 1151. ! pdfTeX warning (ext4): destination with the same identifier (name{page.1}) ha s been already used, duplicate ignored \penalty l.1157 B ut, what has produced this notion, that the subject of [1 ] Underfull \vbox (badness 1874) has occurred while \output is active [] ! pdfTeX warning (ext4): destination with the same identifier (name{page.2}) ha s been already used, duplicate ignored \penalty l.1214 [2] ! pdfTeX warning (ext4): destination with the same identifier (name{page.3}) ha s been already used, duplicate ignored \penalty l.1295 [3] Underfull \vbox (badness 6332) has occurred while \output is active [] ! pdfTeX warning (ext4): destination with the same identifier (name{page.4}) ha s been already used, duplicate ignored \penalty l.1339 [4] ! pdfTeX warning (ext4): destination with the same identifier (name{page.5}) ha s been already used, duplicate ignored \penalty l.1365 \[ [5] ! pdfTeX warning (ext4): destination with the same identifier (name{page.6}) ha s been already used, duplicate ignored \penalty l.1435 T o express the part of the stress ($P$~\&c.)\ which depends on [6] ! pdfTeX warning (ext4): destination with the same identifier (name{page.7}) ha s been already used, duplicate ignored \penalty l.1498 T o put down the equations of motion let $X_x$, $Y_x$, $Z_x$ be the [7] ! pdfTeX warning (ext4): destination with the same identifier (name{page.8}) ha s been already used, duplicate ignored \penalty l.1557 \SectionTitle{Electricity} [8] [9] [10] [11 ] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24 ] [25] [26] [27] [28] [29] [30] [31] [32] LaTeX Font Info: Font shape `OMS/txr/m/it' in size <12> not available (Font) Font shape `OMS/txsy/m/n' tried instead on input line 3057. [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] <./i mages/cylinder.pdf, id=1007, 130.4875pt x 88.33pt> File: ./images/cylinder.pdf Graphic file (type pdf) [48 <./images/cylinder.pdf>] [49] [50] [51] [52] [5 3] [54] [55 ] Underfull \vbox (badness 10000) has occurred while \output is active [] [56] [57] [58] [59] [60] [61] [62] Overfull \hbox (2.23517pt too wide) in paragraph at lines 4862--4867 \OT1/txr/m/n/12 From this we know by the the-ory of po-ten-tial that at the sur -face where the charge $\OML/txmi/m/it/12 ^^[$ [] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75] [76] [77 ] [78] [79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [9 3] [94] [95] [96 ] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [1 10 ] [111] [112 ] Underfull \vbox (badness 10000) has occurred while \output is active [] [113] Underfull \vbox (badness 10000) has occurred while \output is active [] [114] Underfull \vbox (badness 10000) has occurred while \output is active [] [115] Underfull \vbox (badness 10000) has occurred while \output is active [] [116] Underfull \vbox (badness 10000) has occurred while \output is active [] [117] Underfull \vbox (badness 10000) has occurred while \output is active [] [118] Underfull \vbox (badness 10000) has occurred while \output is active [] [119] [120] (./26262-t.aux) *File List* book.cls 2004/02/16 v1.4f Standard LaTeX document class bk12.clo 2004/02/16 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 ifthen.sty 2001/05/26 v1.1c Standard LaTeX ifthen package (DPC) inputenc.sty 2004/02/05 v1.0d Input encoding file latin1.def 2004/02/05 v1.0d Input encoding file 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 2001/07/07 v1.0n Standard LaTeX Graphics (DPC,SPQR) trig.sty 1999/03/16 v1.09 sin cos tan (DPC) graphics.cfg 2001/08/31 v1.1 graphics configuration of teTeX/TeXLive pdftex.def 2005/06/20 v0.03m graphics/color for pdftex alltt.sty 1997/06/16 v2.0g defines alltt environment indentfirst.sty 1995/11/23 v1.03 Indent first paragraph (DPC) footmisc.sty 2005/03/17 v5.3d a miscellany of footnote facilities perpage.sty 2002/12/20 v1.0 Reset counters per page txfonts.sty 2005/01/03 v3.2 fancyhdr.sty extramarks.sty geometry.sty 2002/07/08 v3.2 Page Geometry geometry.cfg hyperref.sty 2003/11/30 v6.74m Hypertext links for LaTeX pd1enc.def 2003/11/30 v6.74m Hyperref: PDFDocEncoding definition (HO) hyperref.cfg 2002/06/06 v1.2 hyperref configuration of TeXLive url.sty 2005/06/27 ver 3.2 Verb mode for urls, etc. hpdftex.def 2003/11/30 v6.74m Hyperref driver for pdfTeX pifont.sty 2004/09/15 PSNFSS-v9.2 Pi font support (SPQR) upzd.fd 2001/06/04 font definitions for U/pzd. upsy.fd 2001/06/04 font definitions for U/psy. omltxmi.fd 2000/12/15 v3.1 omstxsy.fd 2000/12/15 v3.1 omxtxex.fd 2000/12/15 v3.1 utxexa.fd 2000/12/15 v3.1 ot1txr.fd 2000/12/15 v3.1 supp-pdf.tex color.sty 1999/02/16 v1.0i Standard LaTeX Color (DPC) color.cfg 2001/08/31 v1.1 color configuration of teTeX/TeXLive nameref.sty 2003/12/03 v2.21 Cross-referencing by name of section 26262-t.out 26262-t.out ot1txtt.fd 2000/12/15 v3.1 utxsya.fd 2000/12/15 v3.1 utxsyb.fd 2000/12/15 v3.1 utxmia.fd 2000/12/15 v3.1 utxsyc.fd 2000/12/15 v3.1 ./images/pubmark.pdf omstxr.fd 2000/12/15 v3.1 ./images/cylinder.pdf *********** ) Here is how much of TeX's memory you used: 6129 strings out of 95148 76537 string characters out of 1184452 134140 words of memory out of 1000000 8370 multiletter control sequences out of 10000+50000 69014 words of font info for 159 fonts, out of 500000 for 2000 73 hyphenation exceptions out of 8191 28i,19n,43p,260b,477s stack positions out of 1500i,500n,5000p,200000b,5000s PDF statistics: 2041 PDF objects out of 300000 758 named destinations out of 131072 459 words of extra memory for PDF output out of 10000 < /usr/share/texmf-texlive/fonts/type1/public/txfonts/txsy.pfb>{/usr/share/texmf-texlive/fonts/enc/dvips/base/8r.e nc} Output written on 26262-t.pdf (134 pages, 732335 bytes).