£0.

THE

SCIENTIFIC PAPEES

OF

JAMES CLEEK MAXWELL.

HonDon: C. J. CLAY & SONS,

CAMBRIDGE UNIVERSITY PRESS WAREHOUSE,

AVE MARIA LANE.

CambtiBgt: DEIGHTON, BELL AND CO.
F. A. BHOCKHAU8.

\

<2^

THE

SCIENTIFIC PAPEE8

OF

JAMES CLEEK MAXWELL,

M.A., LL.D. EDIN., D.C.L., F.R.SS. LONDON AND EDINBURGH,
HONORARY FELLOW OF TRINITY COLLEGE,

/

CAVENDISH PROFESSOR OF EXPERIMENTAL PHYSICS IN THE UNIVERSITY

OF CAMBRIDGE.

EDITED BY

W. D. NIVEN, M.A., F.R.S.,

DIRECTOR OF STUDIES AT THE ROYAL NAVAL COLLEGE, GREENWICH ;
FORMERLY FELLOW OF TRINITY COLLEGE.

VOL. I.

CAMBRIDGE :

AT THE UNIVERSITY PRESS.
1890

[All Rights reserved']

CAMBRIDGE :

PRINTED BY C. J. CLAY, M.A. AND SONS,
AT THE UNIVERSITY PRESS.

n •;*'•
i p

TO HIS GRACE

THE DUKE OF DEVONSHIRE KG.

CHANCELLOR OF THE UNIVERSITY OF CAMBRIDGE

FOUNDER OF THE CAVENDISH LABORATORY

THIS MEMORIAL EDITION

OF

THE SCIENTIFIC PAPERS
OF

THE FIRST CAVENDISH PROFESSOR OF EXPERIMENTAL PHYSICS

is
BY HIS GRACE'S PERMISSION

RESPECTFULLY AND GRATEFULLY DEDICATED

^HORTLY after the death of Professor James Clerk Maxwell a Committee was
formed, consisting of graduate members of the University of Cambridge and
of other friends and admirers, for the purpose of securing a fitting memorial of
him.

The Committee had in view two objects : to obtain a likeness of Professor
Clerk Maxwell, which should be placed in some public building of the Uni-
versity ; and to collect and publish his scattered scientific writings, copies of
which, so far as the funds at the disposal of the Committee would allow,
should be presented to learned Societies and Libraries at home and abroad.

It was decided that the likeness should take the form of a marble bust.
This was executed by Sir J. E. Boehm, R.A., and , is now placed in the
apparatus room of the Cavendish Laboratory.

In carrying out the second part of their programme the Committee
obtained the cordial assistance of the Syndics of the University Press, who
willingly consented to publish the present work. At the request of the Syndics,
Mr W. D. Niven, M.A., Fellow and Assistant Tutor of Trinity College and
now Director of Studies at the Royal Naval College, Greenwich, undertook the
duties of Editor.

The Committee and the Syndics desire to take this opportunity of
acknowledging their obligation to Messrs Adam and Charles Black, Publishers
of the ninth Edition of the Encyclopedia Britannica, to Messrs Taylor and
Francis, Publishers of the London, Edinburgh, and Dublin Philosophical Maga-
zine and Journal of Science, to Messrs Macmillan and Co., Publishers of
Nature and of the Cambridge and Dublin Mathematical Journal, to Messrs
Metcalfe and Co., Publishers of the Quarterly Journal of Pure and Applied
Mathematics, and to the Lords of the Committee of Council on Education,
Proprietors of the Handbooks of the South Kensington Museum, for their
courteous consent to allow the articles which Clerk Maxwell had contributed to
these publications to be included in the present work ; to Mr Norman Lockyer
for the assistance which he rendered in the selection of the articles re-printed
from Nature ; and their further obligation to Messrs Macmillan and Co. for
permission to use in this work the steel engravings of Faraday, Clerk Maxwell,
and Helmholtz from the Nature Series of Portraits.

via

Numerous and important Papers, contributed by Clerk Maxwell to the
TVoMMOfMMU or Proceedings of the Royal Societies of London and of Edinburgh,
of the Cambridge Philosophical Society, of the Royal Scottish Society of Arts,
and of the London Mathematical Society ; Lectures delivered by Clerk Maxwell
at the Royal Institution of Great Britain published in its Proceedings ; as well
as Communications and Addresses to the British Association published in its
/.'• t>orts, are also included in the present work with the sanction of the above
mentioned learned bodies.

The Essay which gained the Adams Prize for the year 1856 in the
University of Cambridge, the introductory Lecture on the Study of Experimental
Physics delivered in the Cavendish Laboratory, and the Rede Lecture delivered
before the University in 1878, complete this collection of Clerk Maxwell's scientific
writings.

The diagrams in this work have been re-produced by a photographic
process from the original diagrams in Clerk Maxwell's Papers by the Cambridge
Scientific Instrument Company.

It only remains to add that the footnotes inserted by the Editor are
enclosed between square brackets.

CAMBRIDGE, August, 1890.

PEEFACE.

/^ILERK MAXWELL'S biography has been written by Professors Lewis Campbell and
Wm. Garnett with so much skill and appreciation of their subject that nothing further
remains to be told. It would therefore be presumption on the part of the editor of his
papers to attempt any lengthened narrative of a biographical character. At the same time
a memorial edition of an author's collected writings would hardly be complete without
some account however slight of his life and works. Accordingly the principal events of
Clerk Maxwell's career will be recounted in the following brief sketch, and the reader
who wishes to obtain further and more detailed information or to study his character in
its social relations may consult the interesting work to which reference has been made.

James Clerk Maxwell was descended from the Clerks of Penicuick in Midlothian,
a well-known Scottish family whose history can be traced back to the ICth century. The
first baronet served in the parliament of Scotland. His eldest son, a man of learning,
was a Baron of the Exchequer in Scotland. In later times John Clerk of Eldin a
member of the family claimed the credit of having invented a new method of breaking
the enemy's line in naval warfare, an invention said to have been adopted by Lord
Rodney in the battle which he gained over the French in 1782. Another John Clerk,
son of the naval tactitian, was a lawyer of much acumen and became a Lord of the
Court of Session. He was distinguished among his Edinburgh contemporaries by his ready
and sarcastic wit.

The father of the subject of this memoir was John, brother to Sir George Clerk of
Penicuick. He adopted the surname of Maxwell on succeeding to an estate in Kirkcud-
brightshire which came into the Clerk family through marriage with a Miss Maxwell. It
cannot be said that he was possessed of the energy and activity of mind which lead
to distinction. He was in truth a somewhat easy-going but shrewd and intelligent
man, whose most notable characteiistics were his perfect sincerity and extreme benevolence.
He took an enlightened interest in mechanical and scientific pursuits and was of an
essentially practical turn of mind. On leaving the University he had devoted himself
to law and was called to the Scottish Bar. It does not appear however that he met
with any great success in that profession. At all events, a quiet life in the country
VOL. I. b

X PREFACE.

presented so many attractions to his wife as well as to himself that he was easily induced
to relinquish his prospects at the bar. He had been married to Frances, daughter of
Robert Cay of N. Charlton, Northumberland, a lady of strong good sense and resolute
character.

The country house which was their home after they left Edinburgh was designed
by John Clerk Maxwell himself and was built on his estate. The house, which was named
Glenlair, was surrounded by fine scenery, of which the water of Urr with its rocky and
wooded banks formed the principal charm.

James was born at Edinburgh on the 13th of June, 1831, but it was at Glenlair
that the greater part of his childhood was passed. In that pleasant spot under healthful
influences of all kinds the child developed into a hardy and courageous boy. Not
precociously clever at books he was yet not without some signs of future intellectual
strength, being remarkable for a spirit of inquiry into the causes and connections of the
phenomena around him. It was remembered afterwards when he had become distinguished,
that the questions he put as a child shewed an amount of thoughtfulness which for his
years was very unusual.

At the age of ten, James, who had lost his mother, was placed under the charge of
relatives in Edinburgh that he might attend the Edinburgh Academy. A charming account
of his school days is given in the narrative of Professor Campbell who was Maxwell's
schoolfellow and in after life an intimate friend and constant correspondent. The child is
father to the man, and those who were privileged to know the man Maxwell will easily
recognise Mr Campbell's picture of the boy on his first appearance at school, — the home-
made garments more serviceable than fashionable, the rustic speech and curiously quaint
but often humorous manner of conveying his meaning, his bewilderment on first undergoing
the routine of schoolwork, and his Spartan conduct under various trials at the hands of
his schoolfellowa They will further feel how accurate is the sketch of the boy become
accustomed to his surroundings and rapidly assuming the place at school to which his
mental powers entitled him, while his superfluous energy finds vent privately in carrying
out mechanical contrivances and geometrical constructions, in reading and even trying his
hand at composing ballads, and in sending to his father letters richly embellished with
grotesquely elaborate borders and drawings.

An event of his school-days, worth recording, was his invention of a mechanical method
of drawing certain classes of Ovals. An account of this method was printed in the
Proceedings of the Royal Society of Edinburgh and forms the first of his writings
collected in the present work. The subject was introduced to the notice of the Society
by the celebrated Professor James Forbes, who from the first took the greatest possible
interest in Maxwell's progress. Professor Tait, another schoolfellow, mentions that at the
time when the paper on the Ovals was written, Maxwell had received no instruction in
hematic* beyond a little Euclid and Algebra.

PREFACE. XI

In 1847 Maxwell entered the University of Edinburgh where he remained for three
sessions. He attended the lectures of Kelland in Mathematics, Forbes in Natural Philosophy,
Gregory in Chemistry, Sir W. Hamilton in Mental Philosophy, Wilson (Christopher North)
in Moral Philosophy. The lectures of Sir W. Hamilton made a strong impression upon
him, in stimulating the love of speculation to which his mind was prone, but, as might
have been expected, it was the Professor of Natural Philosophy who obtained the chief share
of his devotion. The enthusiasm which so distinguished a man as Forbes naturally inspired
in young and ardent disciples, evoked a feeling of personal attachment, and the Professor, on
his part, took special interest in his pupil and gave to him the altogether unusual
privilege of working with his fine apparatus.

What was the nature of this experimental work we may conjecture from a perusal of
his paper on Elastic Solids, written at that time, in which he describes some experiments
made with the view of verifying the deductions of his theory in its application to Optics.
Maxwell would seem to have been led to the study of this subject by the following cir-
cumstance. He was taken by his uncle John Cay to see William Nicol, the inventor of
the polarising prism which bears his name, and was shewn by Nicol the colours of unan-
nealed glass in the polariscope. This incited Maxwell to study the laws of polarised light
and to construct a rough polariscope in which the polariser and analyser were simple glass
reflectors. By means of this instrument he was able to obtain the colour bands of unannealed
glass. These he copied on paper in water colours and sent to Nicol. It is gratifying to
find that this spirited attempt at experimenting on the part of a mere boy was duly
appreciated by Nicol, who at once encouraged and delighted him by a present of a couple of
his prisms.

The paper alluded to, viz. that entitled " On the Equilibrium of Elastic Solids," was
read to the Royal Society of Edinburgh in 1850. It forms the third paper which Maxwell
addressed to that Society. The first in 1846 on Ovals has been already mentioned. The
second, under the title " The Theory of Rolling Curves," was presented by Kelland in 1849.

It is obvious that a youth of nineteen years who had been capable of these efforts
must have been gifted with rare originality and with great power of sustained exertion.
But his singular self-concentration led him into habits of solitude and seclusion, the tendency
of which was to confirm his peculiarities of speech and of manner. He was shy and
reserved with strangers, and his utterances were often obscure both in substance and in
his manner of expressing himself, so many remote and unexpected allusions perpetually
obtruding themselves. Though really most sociable and even fond of society he was
essentially reticent and reserved. Mr Campbell thinks it is to be regretted that Maxwell
did not begin his Cambridge career earlier for the sake of the social intercourse which
he would have found it difficult to avoid there. It is a question, however, whether in
losing the opportunity of using Professor Forbes' apparatus he would not thereby have lost
what was perhaps the most valuable part of his early scientific training.

b2

Xll PREFACE.

It was originally intended that Maxwell should follow his father's profession of advocate,
but this intention was abandoned as soon as it became obvious that his tastes lay in a
direction so decidedly scientific. It was at length determined to send him to Cambridge
and accordingly in October, 1850, he commenced residence in Peterhouse, where however he
resided during the Michaelmas Term only. On December 14 of the same year he migrated
to Trinity College.

It may readily be supposed that his preparatory training for the Cambridge course
was far removed from the ordinary type. There had indeed for some time been practically
no restraint upon his plan of study and his mind had been allowed to follow its natural
bent towards science, though not to an extent so absorbing as to withdraw him from
other pursuits. Though he was not a sportsman, — indeed sport so called was always repugnant
to him — he was yet exceedingly fond of a country life. He was a good horseman and a
good swimmer. Whence however he derived his chief enjoyment may be gathered from the
account which Mr Campbell gives of the zest with which he quoted on one occasion the
lines of Burns which describe the poet finding inspiration while wandering along the banks
of a stream in the free indulgence of his fancies. Maxwell was not only a lover of poetry
but himself a poet, as the fine pieces gathered together by Mr Campbell abundantly testify.
He saw however that his true calling was Science and never regarded these poetical
efforts as other than mere pastime. Devotion to science, already stimulated by successful
endeavour, a tendency to ponder over philosophical problems and an attachment to English
literature, particularly to English poetry, — these tastes, implanted in a mind of singular
strength and purity, may be said to have been the endowments with which young Maxwell
began his Cambridge career. Besides this, his scientific reading, as we may gather from his
papers to the Royal Society of Edinburgh referred to above, was already extensive and
varied. He brought with him, says Professor Tait, a mass of knowledge which was really
immense for so young a man but in a state of disorder appalling to his methodical
private tutor.

Maxwell's undergraduate career was not marked by any specially notable feature. His
private speculations had in some measure to be laid aside in favour of more systematic
-tuily. Yet his mind was steadily ripening for the work of his later years. Among those
with whom he was brought into daily contact by his position, as a Scholar of Trinity
College, were some of the brightest and most cultivated young men in the University. In
thi- genial fellowship of the Scholars' table Maxwell's kindly humour found ready play, while
in the more select coterie of the Apostle Club, formed for mutual cultivation, he found a field
for the exercise of his love of speculation in essays on subjects beyond the lines of the
unliiiary University course. The composition of these essays doubtless laid the foundation
»f that literary finish which is one of the characteristics of Maxwell's scientific writings.

biographers have preserved several extracts on a variety of subjects chiefly of a specu-
lative character. They are remarkable mainly for the weight of thought contained in them
but occasionally also for smart epigrams and for a vein of dry and sarcastic humour.

PREFACE. xiii

These glimpses into Maxwell's character may prepare us to believe that, with all his
shyness, he was not without confidence in his own powers, as also appears from the account
which was given by the late Master of Trinity College, Dr Thompson, who was Tutor when
Maxwell personally applied to him for permission to migrate to that College. He appeared
to be a shy and diffident youth, but presently surprised Dr Thompson by producing a
bundle of papers, doubtless copies of those we have already mentioned, remarking " Perhaps
these may shew you that I am not unfit to enter at your College."

He became a pupil of the celebrated William Hopkins of Peterhouse, under whom his
course of study became more systematic. One striking characteristic was remarked by his
contemporaries. Whenever the subject admitted of it he had recourse to diagrams, though
his fellow students might solve the question more easily by a train of analysis. Many
illustrations of this manner of proceeding might be taken from his writings, but in
truth it was only one phase of his mental attitude towards scientific questions, which
led him to proceed from one distinct idea to another instead of trusting to symbols and
equations. /

Maxwell's published contributions to Mathematical Science during his undergraduate career
were few and of no great importance. He found time however to carry his investigations
into regions outside the prescribed Cambridge course. At the lectures of Professor Stokes*
he was regular in his attendance. Indeed it appears from the paper on Elastic Solids,
mentioned above, that he was acquainted with some of the writings of Stokes before he
entered Cambridge. Before 1850, Stokes had published some of his most important contri-
butions to Hydromechanics and Optics ; and Sir W. Thomson, who was nine years' Maxwell's
senior in University standing, had, among other remarkable investigations, called special
attention to the mathematical analogy between Heat-conduction and Statical Electricity.
There is no doubt that these authors as well as Faraday, of whose experimental researches
he had made a careful study, exercised a powerful directive influence on his mind.

In January, 1854, Maxwell's undergraduate career closed. He was second wrangler, but
shared with Dr Routh, who was senior wrangler, the honours of the First Smith's Prize.
In due course he was elected Fellow of Trinity and placed on the staff of College Lecturers.

No sooner was he released from the restraints imposed by the Trinity Fellowship
Examination than he plunged headlong into original work. There were several questions
he was anxious to deal with, and first of all he completed an investigation on the Trans-
formation of Surfaces by Bending, a purely geometrical problem. This memoir he presented
to the Cambridge Philosophical Society in the following March. At this period he also
set about an enquiry into the quantitative measurement of mixtures of colours and the
causes of colour-blindness. During his undergraduateship he had, as we have seen, found
time for the study of Electricity. This had already borne fruit and now resulted in the
first of his important memoirs on that subject, — the memoir on Faraday's Lines of Force.
* Now Sir George Gabriel Stokes, Bart., M.P. for the University.

IM:I:I \,

The number and importance of his papers, published in 1855 — 6, bear witness to his
assiduity during this period. With these labours, and in the preparation of his College
lectures, on which he entered with much enthusiasm, his mind was fully occupied and the
work was congenial. He had formed a number of valued friendships, and he had a variety of
interests, scientific and literary, attaching him to the University. Nevertheless, when the chair
of Natural Philosophy in Marischal College, Aberdeen, fell vacant, Maxwell became a candidate.
This step was probably taken in deference to his father's wishes, as the long summer
vacation of the Scottish College would enable him to reside with his father at Glenlair for
half the year continuously. He obtained the professorship, but unhappily the kind intentions
which prompted him to apply for it were frustrated by the death of his father, which took
place in April,

It is doubtful whether the change from the Trinity lectureship to the Aberdeen
professorship was altogether prudent. The advantages were the possession of a laboratory and
the long uninterrupted summer vacation. But the labour of drilling classes composed chiefly
of comparatively young and untrained lads, in the elements of mechanics and physics, was
not the work for which Maxwell was specially fitted. On the other hand, in a large college
like Trinity there could not fail to have been among its undergraduate members, some of the
most promising young mathematicians of the University, capable of appreciating his original
genius and immense knowledge, by instructing whom he would himself have derived ad-
vantage.

In 185G Max \\x-ll entered upon his duties as Professor of Natural Philosophy at Marischal
College, and two years afterwards he married Katharine Mary Dewar, daughter of the
Principal of the College. He in consequence ceased to be a Fellow of Trinity College,
but was afterwards elected an honorary Fellow, at the same time as Professor Cayley.

During the years 1856 — 60 he was still actively employed upon the subject of colour
sensation, to which he contributed a new method of measurement in the ingenious instru-
ment known as the colour-box. The most serious demands upon his powers and upon his
time were made by his investigations on the Stability of Saturn's Rings. This was the
subject chosen by the Examiners for the Adams Prize Essay to be adjudged in 1857, and
was advertised in the following terms: —

"The Problem may be treated on the supposition that the system of Rings is
exactly or very approximately concentric with Saturn and symmetrically disposed about
the plane of his equator and different hypotheses may be made respecting the physical
constitution of the Rings. It may be supposed (1) that they are rigid; (2) that they
are fluid and in part aeriform ; (3) that they consist of masses of matter not materially
coherent. The question will be considered to be answered by ascertaining on these
hypotheses severally whether the conditions of mechanical stability are satisfied by the
mutual attractions and motions of the Planet and the Rings."

PREFACE. XV

" It is desirable that an attempt should also be made to determine on which of
the above hypotheses the appearances both of the bright rings and the recently
discovered dark ring may be most satisfactorily explained; and to indicate any causes
to which a change of form such as is supposed from a comparison of modern with the
earlier observations to have taken place, may be attributed."

It is sufficient to mention here that Maxwell bestowed an immense amount of labour
in working out the theory as proposed, and that he arrived at the conclusion that "the
only system of rings which can exist is one composed of an indefinite number of unconnected
particles revolving round the planet with different velocities according to their respective
distances. These particles may be arranged in a series of narrow rings, or they may move
about through each other irregularly. In the first case the destruction of the system will be
very slow, in the second case it will be more rapid, but there may be a tendency towards
an arrangement in narrow rings which may retard the process."

Part of the work, dealing with the oscillatory waves set up in a ring of satellites,
was illustrated by an ingenious mechanical contrivance which was greatly admired when
exhibited before the Royal Society of Edinburgh.

This essay, besides securing the prize, obtained for its author great credit among
scientific men. It was characterized by Sir George Airy as one of the most remarkable
applications of Mathematics to Physics that he had ever seen.

The suggestion has been made that it was the irregular motions of the particles which
compose the Rings of Saturn resulting on the whole in apparent regularity and uni-
formity, which led Maxwell to the investigation of the Kinetic Theory of Gases, his first
contribution to which was read to the British Association in 1859. This is not unlikely,
but it must also be borne in mind that Bernoulli's Theory had recently been revived by
Herapath, Joule and Clausius whose writings may have drawn Maxwell's attention to the
subject.

In 1860 King's College and Marischal College were joined together as one institution,
now known as the University of Aberdeen. The new chair of Natural Philosophy thus
created was filled up by the appointment of David Thomson, formerly Professor at King's
College and Maxwell's senior. Professor Thomson, though not comparable to Maxwell as a
physicist, was nevertheless a remarkable man. He was distinguished by singular force of
character and great administrative faculty and he had been prominent in bringing about
the fusion of the Colleges. He was also an admirable lecturer and teacher and had done
much to raise the standard of scientific education in the north of Scotland. Thus the choice
made by the Commissioners, though almost inevitable, had the effect of making it appear
that Maxwell failed as a teacher. There seems however to be no evidence to support such
an inference. On the contrary, if we may judge from the number of voluntary students
attending his classes in his last College session, he would seem to have been as popular as a
professor as he was personally estimable.

xvi PKEFACE.

This is also borne out by the fact that he was soon afterwards elected Professor of
ural Philosophy and Astronomy in King's College, London. The new appointment had
the advantage of bringing him much more into contact with men in his own department
of science, especially with Faraday, with whose electrical work his own was so intimately
connected. In 1862 — 63 he took a prominent part in the experiments organised by a
Committee of the British Association for the determination of electrical resistance in
absolute measure and for placing electrical measurements on a satisfactory basis. In the
experiments which were conducted in the laboratory of King's College upon a plan due
to Sir W. Thomson, two long series of measurements were taken in successive years. In
thr first year, the working members were Maxwell, Balfour Stewart and Fleeming Jenkin ; in
th.- second, Charles Hockin took the place of Balfour Stewart. The work of this Committee
was communicated in the form of reports to the British Association and was afterwards
republished in one volume by Fleemiug Jenkin.

Maxwell was a professor in King's College from I860 to 1865, and this period of his
liti- is distinguished by the production of his most important papers. The second memoir
on Colours made its appearance in 1860. In the same year his first papers on the Kinetic
Theory of Gases were published. In 1861 came his papers on Physical Lines of Force
and in 1864 his greatest memoir on Electricity, — a Dynamical Theory of the Electro-
magnetic Field. He must have been occupied with the Dynamical Theory of Gases in l>S(i">.
as two important papers appeared in the following year, first the Bakerian lecture on the
Viscosity of Gases, and next the memoir on the Dynamical Theory of Gases.

The mental strain involved in the production of so much valuable work, combined
with the duties of his professorship which required his attention during nine months of
the year, seems to have influenced him in a resolution which in 1865 he at length
adopted of resigning his chair and retiring to his country seat. Shortly after this he had
a severe illness. On his recovery he continued his work on the Dynamical Theory of
Gases, to which reference has just been made. For the next few years he led a quiet
and secluded life at Glenlair, varied by annual visits to London, attendances at the British
Association meetings and by a tour in Italy in 1867. He was also Moderator or Examiner
in the Mathematical Tripos at Cambridge on several occasions, offices which entailed a fr\\
weeks' residence at the University in winter. His chief employment during those years
\\a^ the preparation of his now celebrated treatise on Electricity and Magnetism which,
however, was not published till 1873. He also wrote a treatise on Heat which was
published in 1871.

In 1871 Maxwell was, with some reluctance, induced to quit his retreat in the
country and to enter upon a new career. The University of Cambridge had recently
resolved to found a professorship of physical science, especially for the cultivation and
teaching of the subjects of Heat, Electricity and Magnetism. In furtherance of this
object her Chancellor, the Duke of Devonshire, had most generously undertaken to build
a laboratory and furnish it with the necessary apparatus. Maxwell was invited to fill the

PREFACE. xvii

new chair thus formed and to superintend the erection of the laboratory. In October,
1871, he delivered his inaugural lecture.

The Cavendish Laboratory, so called after its founder, the present venerable chief of
the family which produced the great physicist of the same name, was not completed
for practical work until 1874. In June of that year it was formally presented to the
University by the Chancellor. The building itself and the fittings of the several rooms
were admirably contrived mainly by Maxwell himself, but the stock of apparatus was
smaller than accorded with the generous intentions of the Chancellor. This defect must
be attributed to the anxiety of the Professor to procure only instruments by the best
makers and with such improvements as he could himself suggest. Such a defect therefore
required time for its removal and afterwards in great measure disappeared, apparatus being
constantly added to the stock as occasion demanded.

One of the chief tasks which Maxwell undertook was that of superintending and
directing the energies of such young Bachelors of Arts as became his pupils after
having acquired good positions in the University examinations. ' Several pupils, who have
since acquired distinction, carried out valuable experiments under the guidance of the
Professor. It must be admitted, however, that the numbers were at first small, but perhaps
this was only to be expected from the traditions of so many years. The Professor was
singularly kind and helpful to these pupils. He would hold long conversations with them,
opening up to them the stores of his mind, giving them hints as to what they might try
and what avoid, and was always ready with some ingenious remedy for the experimental
troubles which beset them. These conversations, always delightful and instructive, were,
according to the account of one of his pupils, a liberal education in themselves, and were
repaid in the minds of the pupils by a grateful affection rarely accorded to any teacher.

Besides discharging the duties of his chair, Maxwell took an active part in conducting
the general business of the University and more particularly in regulating the courses of
study in Mathematics and Physics.

For some years previous to 1866 when Maxwell returned to Cambridge as Moderator
in the Mathematical Tripos, the studies in the University had lost touch with the great
scientific movements going on outside her walls. It was said that some of the subjects most
in vogue had but little interest for the present generation, and loud complaints began to
be heard that while such branches of knowledge as Heat, Electricity and Magnetism, were
left out of the Tripos examination, the candidates were wasting their time and energy
upon mathematical trifles barren of scientific interest and of practical results. Into the
movement for reform Maxwell entered warmly. By his questions in 1866 and subsequent
years he infused new life into the examination ; he took an active part in drafting the
new scheme introduced in 1873 ; but most of all by his writings he exerted a powerful
influence on the younger members of the University, and was largely instrumental in
bringing about the change which has been now effected.

VOL. I. C

PREFACE.

In the first few years at Cambridge Maxwell was busy in giving the final touches
to his great work on Electricity and Magnetism and in passing it through the press.
This work was published in 1873, and it seems to have occupied the most of his attention
for the two previous years, as the few papers published by him during that period relate
chiefly to subjects forming part of the contents. After this publication his contributions to
scientific journals became more numerous, those on the Dynamical Theory of Oases being
perhaps the most important. He also wrote a great many short articles and reviews
which made their appearance in Nature and the Encyclopaedia Britannica. Some of these
essays are charming expositions of scientific subjects, some are general criticisms of the
works of contemporary writers and others are brief and appreciative biographies of fellow
workers in the same fields of research.

An undertaking in which he was long engaged and which, though it proved exceedingly
interesting, entailed much labour, was the editing of the "Electrical Researches" of the Hon.
Henry Cavendish. This work, published in 1879, has had the effect of increasing the
reputation of Cavendish, disclosing as it does the unsuspected advances which that acute
physicist had made in the Theory of Electricity, especially in the measurement of electrical
quantities. The work is enriched by a variety of valuable notes in which Cavendish's
views and results are examined by the light of modern theory and methods. Especially
valuable are the methods applied to the determination of the electrical capacities of con-
ductors and condensers, a subject in which Cavendish himself shewed considerable skill
both of a mathematical and experimental character.

The importance of the task undertaken by Maxwell in connection with Cavendish's
papers will be understood from the following extract from his introduction to them.

" It is somewhat difficult to account for the fact that though Cavendish had
prepared a complete description of his experiments on the charges of bodies, and had
even taken the trouble to write out a fair copy, and though all this seems to have
been done before 1774 and he continued to make experiments in Electricity till 1781
and lived on till 1810, he kept his manuscript by him and never published it."

" Cavendish cared more for investigation than for publication. He would under-
take the most laborious researches in order to clear up a difficulty which no one
but himself could appreciate or was even aware of, and we cannot doubt that the
result of his enquiries, when successful, gave him a certain degree of satisfaction.
But it did not excite in him that desire to communicate the discovery to others
which in the case of ordinary men of science, generally ensures the publication of
their results. How completely these researches of Cavendish remained unknown to
other men of science is shewn by the external history of electricity."

It will probably be thought a matter of some difficulty to place oneself in th.-
I position of a physicist of a century ago and to ascertain the exact bearing of hi<
l experiments. But Maxwell entered upon this undertaking with the utmost enthusiasm and

PREFACE. XIX

succeeded in completely identifying himself with Cavendish's methods. He shewed that
Cavendish had really anticipated several of the discoveries in electrical science which have been
made since his time. Cavendish was the first to form the conception of and to measure
Electrostatic Capacity and Specific Inductive Capacity; he also anticipated Ohm's law.

The Cavendish papers were no sooner disposed of than Maxwell set about preparing
a new edition of his work on Electricity and Magnetism; but unhappily in the summer
term of 1879 his health gave way. Hopes were however entertained that when he returned
to the bracing air of his country home he would soon recover. But he lingered through
the summer months with no signs of improvement and his spirits gradually sank. He was
finally informed by his old fellow-student, Professor Sanders, that he could not live more
than a few weeks. As a last resort he was brought back to Cambridge in October that he
might be under the charge of his favourite physician, Dr Paget*. Nothing however could
be done for his malady, and, after a painful illness, he died on the 5th of November, 1879,
in his 49th year.

Maxwell was thus cut off in the prime of his powers, and at a time when the depart-
ments of science, which he had contributed so much to develop, were being every day
extended by fresh discoveries. His death was deplored as an irreparable loss to science and
to the University, in which his amiable disposition was as universally esteemed as his genius
was admired.

It is not intended in this preface to enter at length into a discussion of 'the relation
which Maxwell's work bears historically to that of his predecessors, or to attempt to estimate
the effect which it has had on the scientific thought of the present day. In some of his
papers he has given more than usually copious references to the works of those by whom
he had been influenced; and in his later papers, especially those of a more popular nature
which appeared in the Encyclopedia Britannica, he has given full historical outlines of some
of the most prominent fields in which he laboured. Nor does it appear to the present
editor that the time has yet arrived when the quickening influence of Maxwell's mind on
modern scientific thought can be duly estimated. He therefore proposes to himself the duty
of recalling briefly, according to subjects, the most important speculations in which Maxwell
engaged.

His works have been arranged as far as possible in chronological order but they fall
naturally under a few leading heads; and perhaps we shall not be far wrong if we place
first in importance his work in Electricity.

His first paper on this subject bearing the title "On Faraday's Lines of Force" was
read before the Cambridge Philosophical Society on Dec. llth, 1855. He had been previously
attracted by Faraday's method of expressing electrical laws, and he here set before himself
the task of shewing that the ideas which had guided Faraday's researches were not incon-
sistent with the mathematical formulae in which Poisson and others had cast the laws of

* Now Sir George Edward Paget, K.C.B.

c2

XX PREFACE.

Electricity. His object, he says, is to find a physical analogy which shall help the mind
to grasp the results of previous investigations "without being committed to any theory
founded on the physical science from which that conception is borrowed, so that it is neither
drawn aside from the subject in the pursuit of analytical subtleties nor carried beyond the
truth by a favorite hypothesis."

The laws of electricity are therefore compared with the properties of an incompressible
fluid the motion of which is retarded by a force proportional to the velocity, and the fluid
is supposed to possess no inertia. He shews the analogy which the lines of flow of such
a fluid would have with the lines of force, and deduces not merely the laws of Statical
Electricity in a single medium but also a method of representing what takes place when the
action passes from one dielectric into another.

In the latter part of the paper he proceeds to consider the phenomena of Electro-
magnetism and shews how the laws discovered by Ampere lead to conclusions identical with
those of Faraday. In this paper three expressions are introduced which he identifies with
the components of Faraday's electrotonic state, though the author admits that he has not
been able to frame a physical theory which would give a clear mental picture of the
various connections expressed by the equations.

Altogether this paper is most important for the light which it throws on the principles
which guided Maxwell at the outset of his electrical work. The idea of the electrotonic
state had already taken a firm hold of his mind though as yet he had formed no physical
explanation of it. In the paper "On Physical Lines of Force" printed in the Philosophical
Magaziiie, Vol. XXI. he resumes his speculations. He explains that in his former paper he
had found the geometrical significance of the Electrotonic state but that he now proposes
"to examine magnetic phenomena from a mechanical point of view." Accordingly he propounds
his remarkable speculation as to the magnetic field being occupied by molecular vortices,
the axes of which coincide with the lines of force. The cells within which these vortices
rotate are supposed to be separated by layers of particles which serve the double purpose
of transmitting motion from one cell to another and by their own motions constituting ;in
electric current This theory, the parent of several working models which have been devised
to represent the motions of the dielectric, is remarkable for the detail with which it is
worked out and made to explain the various laws not only of magnetic and electromagnetic
action, but also the various forms of electrostatic action. As Maxwell subsequently gave a
more general theory of the Electromagnetic Field, it may be inferred that he did not desire
it to be supposed that he adhered to the views set forth in this paper in every particular ;
but there is no doubt that in some of its main features, especially the existence of
rotation round the lines of magnetic force, it expressed his permanent convictions. In his
treatise on " Electricity and Magnetism," Vol. II. p. 416, (2nd edition 427) after quoting from
Sir \V. Thomson on the explanation of the magnetic rotation of the plane of the polarisitinn
of light, he goes on to say of the present paper,

PREFACE. XXI

" A theory of molecular vortices which I worked out at considerable length was
published in the Phil. Mag. for March, April and May, 1861, Jan. and Feb. 1862."

" I think we have good evidence for the opinion that some phenomenon of rotation
is going on in the magnetic field, that this rotation is performed by a great number
of very small portions of matter, each rotating on its own axis, that axis being parallel
to the direction of the magnetic force, and that the rotations of these various vortices
are made to depend on one another by means of some mechanism between them."

" The attempt which I then made to imagine a working model of this mechanism
must be taken for no more than it really is, a demonstration that mechanism may
be imagined capable of producing a connection mechanically equivalent to the actual
connection of the parts of the Electromagnetic Field."

This paper is also important as containing the first hint of the Electromagnetic Theory
of Light which was to be more fully developed afterwards in his third great memoir
" On the Dynamical Theory of the Electromagnetic Field." This me/noir, which was presented
to the Royal Society on the 27th October, 1864, contains Maxwell's mature thoughts on a
subject which had so long occupied his mind. It was afterwards reproduced in his Treatise
with trifling modifications in the treatment of its parts, but without substantial changes
in its main features. In this paper Maxwell reverses the mode of treating electrical
phenomena adopted by previous mathematical writers ; for while they had sought to build
up the laws of the subject by starting from the principles discovered by Ampere, and
deducing the induction of currents from the conservation of energy, Maxwell adopts the
method of first arriving at the laws of induction and then deducing the mechanical
attractions and repulsions.

After recalling the general phenomena of the mutual action of currents and magnets
and the induction produced in a circuit by any variation of the strength of the field in
which it lies, the propagation of light through a luminiferous medium, the properties of
dielectrics and other phenomena which point to a medium capable of transmitting force
and motion, he proceeds. —

" Thus then we are led to the conception of a complicated mechanism capable
of a vast variety of motions but at the same time so connected that the motion of
one part depends, according to definite relations, on the motion of other parts, these
motions being communicated by forces arising from the relative displacement of their
connected parts, in virtue of their elasticity. Such a mechanism must be subject
to the laws of Dynamics."

On applying dynamical principles to such a connected system he attains certain general
propositions which, on being compared with the laws of induced currents, enable him to
identify certain features of the mechanism with properties of currents. The induction of
currents and their electromagnetic attraction are thus explained and connected.

XXII I'KITM 1.

In a subsequent part of the memoir he proceeds to establish from these premises
the gent-nil equations of the Field and obtains the usual formula; for the mechanical
force on currents, magnets and bodies possessing an electrostatic charge.

He also returns to and elaborates more fully the electromagnetic Theory of Light.
His equations shew that dielectrics can transmit only transverse vibrations, the speed of
propagation of which in air as deduced from electrical data comes out practically identical
with the known velocity of light. For other dielectrics the index of refraction is equal
to the square root of the product of the specific inductive capacity by the coefficient of
magnetic induction, which last factor is for most bodies practically unity. Various comparisons
have been made with the view of testing this deduction. In the case of paraffin wax and
some of the hydrocarbons, theory and experiment agree, but this is not the case with
glass and some other substances. Maxwell has also applied his theory to media which
are not perfect insulators, and finds an expression for the loss of light in passing through
a stratum of given thickness. He remarks in confirmation of his result that most good
conductors are opaque while insulators are transparent, but he also adds that electrolytes
which transmit a current freely are often transparent, while a piece of gold leaf whose
resistance was determined by Mr Hockin allowed far too great an amount of light to
pass. He observes however that it is possible "there is less loss of energy when the
electromotive forces are reversed with the rapidity of light than when they act for sensible
times as in our experiments." A similar explanation may be given of the discordance
between the calculated and observed values of the specific inductive capacity. Prof. J. J.
Thomson in the Proceedings of the Royal Society, Vol. 46, has described an experiment by
which he has obtained the specific inductive capacities of various dielectrics when acted
on by alternating electric forces whose frequency is 25,000,000 per second. He finds that
under these conditions the specific inductive capacity of glass is very nearly the same as
the square of the refractive index, and very much less than the value for slow rates of
reversals. In illustration of these remarks may be quoted the observations of Prof. Hertz who
has shewn that vulcanite and pitch are transparent for waves, whose periods of vibration are
about three hundred millionths of a second. The investigations of Hertz have shewn that
electro-dynamic radiations are transmitted in waves with a velocity, which, if not equal to, is
comparable with that of light, and have thus given conclusive proof that a satisfactory
thi-ory of Electricity must take into account in some form or other the action of the
dielectric. But this does not prove that Maxwell's theory is to be accepted in every
particular. A peculiarity of his theory is, as he himself points out in his treatise, that
the variation of the electric displacement is to be treated as part of the current as well
as the current of conduction, and that it is the total amount due to the sum of these
which flows as if electricity were an incompressible fluid, and which determines external
electrodynamic actions. In this respect it differs from the theory of Helmholtz which
al-o takes into account the action of the dielectric. Professor J. J. Thomson in his
Review of Electric Theories has entered into a full discussion of the points at issue

PREFACE. xxiii

between the two above mentioned theories, and the reader is referred to his paper for
further information*. Maxwell in the memoir before us has also applied his theory to
the passage of light through crystals, and gets rid at once of the wave of normal vibrations
which has hitherto proved the stumbling block in other theories of light.

The electromagnetic Theory of Light has received numerous developments at "the hands
of Lord Rayleigh, Mr Glazebrook, Professor J. J. Thomson and others. These volumes
also contain various shorter papers on Electrical Science, though perhaps the most complete
record of Maxwell's work in this department is to be found in his Treatise on Electricity
and Magnetism in which they were afterwards embodied.

Another series of papers of hardly less importance than those on Electricity are the
various memoirs on the Dynamical Theory of Gases. The idea that the properties of
matter might be explained by the motions and impacts of their ultimate atoms is as
old as the time of the Greeks, and Maxwell has given in his paper on "Atoms" a full
sketch of the ancient controversies to which it gave rise. The mathematical difficulties of
the speculation however were so great that it made little real progress till it was taken
up by Clausius and shortly afterwards by Maxwell. The first paper by Maxwell on the
subject is entitled "Illustrations of the Dynamical Theory of Gases" and was published
in the Philosophical Magazine for January and July, 1860, having been read at a meeting
of the British Association of the previous year. Although the methods developed in this
paper were afterwards abandoned for others, the paper itself is most interesting, as it indicates
clearly the problems in the theory which Maxwell proposed to himself for solution, and so far
contains the germs of much that was treated of in his next memoir. It is also epoch-making,
inasmuch as it for the first time enumerates various propositions which are characteristic
of Maxwell's work in this subject. It contains the first statement of the distribution of velo-
cities according to the law of errors. It also foreshadows the theorem that when two gases
are in thermal equilibrium the mean kinetic energy of the molecules of each system is the
same; and for the first time the question of the viscosity of gases is treated dynamically.

In his great memoir " On the Dynamical Theory of Gases " published in the Philo-
sophical Transactions of the Royal Society and read before the Society in May, 1866, he
returns to this subject and lays down for the first time the general dynamical methods
appropriate for its treatment. Though to some extent the same ground is traversed as in
his former paper, the methods are widely different. He here abandons his former hypothesis
that the molecules are hard elastic spheres, and supposes them to repel each other with
forces varying inversely as the fifth power of the distance. His chief reason for assuming
this law of action appears to be that it simplifies considerably the calculation of the
collisions between the molecules, and it leads to the conclusion that the coefficient of
viscosity is directly proportional to the absolute temperature. He himself undertook an
experimental enquiry for the purpose of verifying this conclusion, and, in his paper on the
Viscosity of Gases, he satisfied himself of its correctness. A re-examination of the numerical

* British Association Report, 1885.

XXIV PREFACE.

reductions made in the course of his work discloses however an inaccuracy which materially
affects the values of the coefficient of viscosity obtained. Subsequent experiments also seem
to shew that the concise relation he endeavoured to establish is by no means so near
the truth as he supposed, and it is more than doubtful whether the action between two
molecules can be represented by any law of so simple a character.

In the same memoir he gives a fresh demonstration of the law of distribution of
velocities, but though the method is of permanent value, it labours under the defect of
assuming that the distribution of velocities in the neighbourhood of a point is the same
in every direction, whatever actions may be taking place within the gas. This flaw in
the argument, first pointed out by Boltzmann, seems to have been recognised by Maxwell,
who in his next paper " On the Stresses in Rarefied Gases arising from inequalities of
Temperature," published in the Philosophical Transactions for 1879, Part I., adopts a form
of the distribution function of a somewhat different shape. The object of this paper was
to arrive at a theory of the effects observed in Crookes's Radiometer. The results of the
investigation are stated by Maxwell in the introduction to the paper, from which it would
appear that the observed motion cannot be explained on the Dynamical Theory, unless it
be supposed that the gas in contact with a solid can slide along the surface with a finite
velocity between places whose temperatures are different. In an appendix to the paper
he shews that on certain assumptions regarding the nature of the contact of the solid
and gas, there will be, when the pressure is constant, a flow of gas along the surface
from the colder to the hotter parts. The last of his longer papers on this subject is
one on Boltzmann's Theorem. Throughout these volumes will be found numerous shorter
essays on kindred subjects, published chiefly in Nature and in the Encyclopedia Britaimica.
Some of these contain more or less popular expositions of this subject which Maxwell
had himself in great part created, while others deal with the work of other writers in
the same field. They are profoundly suggestive in almost every page, and abound in acute
criticisms of speculations which he could not accept. They are always interesting; for
although the larger papers are sometimes difficult to follow, Maxwell's more popular writings
are characterized by extreme lucidity and simplicity of style.

The first of Maxwell's papers on Colour Perception is taken from the Transactions of
the Royal Scottish Society of Arts and is in the form of a letter to Dr G. Wilson dated
Jan. 4, 1855. It was followed directly afterwards by a communication to the Royal Society
of Edinburgh, and the subject occupied his attention for some years. The most important
of his subsequent work is to be found in the papers entitled "An account of Experiments
on the Perception of Colour" published in the Philosophical Magazine, Vol. xiv. and "On
the Theory of Compound Colours and its relation to the colours of the spectrum " in the
Philosophical Transactions for the year 1860. We may also refer to two lectures delivered
at the Royal Institution, in which he recapitulates and enforces his main positions in hi-
usual luminous style. Maxwell from the first adopts Young's Theory of Colour Sensation,
according to which all colours may ultimately be reduced to three, a red, a green and

PREFACE. XXV

a violet. This theory had been revived by Helmholtz who endeavoured to find for it a
physiological basis. Maxwell however devoted himself chiefly to the invention of accurate
methods for combining and recording mixtures of colours. His first method of obtaining
mixtures, that of the Colour Top, is an adaptation of one formerly employed, but in
Maxwell's hands it became an instrument capable of giving precise numerical results by
means which he added of varying and measuring the amounts of colour which were
blended in the eye. In the representation of colours diagrammatically he followed Young
in employing an equilateral triangle at the angles of which the fundamental colours were
placed. All colours, white included which may be obtained by mixing the fundamental
colours in any proportions will then be represented by points lying within the triangle.
Points without the triangle represent colours which must be mixed with one of the funda-
mental tints to produce a mixture of the other two, or with which two of them must be
mixed to produce the third.

In his later papers, notably in that printed in the Philosophical Transactions, he
adopts the method of the Colour Box, by which different parts 6f the spectrum may be
mixed in different proportions and matched with white, the intensity of which has been
suitably diminished. In this way a series of colour equations are obtained which can be
used to evaluate any colour in terms of the three fundamental colours. These observations
on which Maxwell expended great care and labour, constitute by far the most important
data regarding the combinations of colour sensations which have been yet obtained, and
are of permanent value whatever theory may ultimately be adopted of the physiology of the
perception of colour.

In connection with these researches into the sensations of the normal eye, may be
mentioned the subject of colour-blindness, which also engaged Maxwell's attention, and is
discussed at considerable length in several of his papers.

Geometrical Optics was another subject in which Maxwell took much interest. At an early
period of his career he commenced a treatise on Optics, which however was never completed.
His first paper "On the general laws of optical instruments," appeared in 1858, but a brief
account of the first part of it had been previously communicated to the Cambridge Philosophical
Society. He therein lays down the conditions which a perfect optical instrument must fulfil,
and shews that if an instrument produce perfect images of an object, i.e. images free from
astigmatism, curvature and distortion, for two different positions of the object, it will give
perfect images at all distances. On this result as a basis, he finds the relations between
the foci of the incident and emergent pencils, the magnifying power and other characteristic
quantities. The subject of refraction through optical combinations was afterwards treated
by him in a different manner, in three papers communicated to the London Mathematical
Society. In the first (1873), "On the focal lines of a refracted pencil," he applies Hamilton's
characteristic function to determine the focal lines of a thin pencil refracted from one
isotropic medium into another at any surface of separation. In the second (1874), "On
VOL. I. ^

XXVI PREFACE.

Hamilton's characteristic function for a narrow beam of light," he considers the more general
question of the passage of a ray from one isotropic medium into another, the two media
being separated by a third which may be of a heterogeneous character. He finds the most
general form of Hamilton's characteristic function from one point to another, the first being
in the medium in which the pencil is incident and the second in the medium in which
it is emergent, and both points near the principal ray of the pencil. This result is then
applied in two particular cases, viz. to determine the emergent pencil (1) from a spectroscope,
(2) from an optical instrument symmetrical about its axis. In the third paper (1875) he
resumes the last-mentioned application, discussing this case more fully under a somewhat
simplified analysis.

It may be remarked that all these papers are connected by the same idea, which was —
first to study the optical effects of the entire instrument without examining the mechanism
by which these effects are produced, and then, as in the paper in 1858, to supply whatever
data may be necessary by experiments upon the instrument itself.

Connected to some extent with the above papers is an investigation which was published
in 1868 " On the cyclide." As the name imports, this paper deals chiefly with the geometrical
properties of the surface named, but other matters are touched on, such as its conjugate
isothermal functions. Primarily however the investigation is on the orthogonal surfaces to
a system of rays passing accurately through two lines. In a footnote to this paper Maxwell
describes the stereoscope which he invented and which is now in the Cavendish Laboratory.

In 1868 was also published a short but important article entitled " On the best arrange-
ment for producing a pure spectrum on a screen."

The various papers relating to the stresses experienced by a system of pieces joined
together so as to form a frame and acted on by forces form an important group connected
with one another. The first in order was "On reciprocal figures and diagrams of forces,"
published in 1864. It was immediately followed by a paper on a kindred subject, " On
the calculation of the equilibrium and stiffness of frames." In the first of these Maxwell
demonstrates certain reciprocal properties in the geometry of two polygons which are related
to one another in a particular way, and establishes his well-known theorem in Graphical
Statics on the stresses in frames. In the second he employs the principle of work to
problems connected with the stresses in frames and structures and with the deflections
arising from extensions in any of the connecting pieces.

A third paper "On the equilibrium of a spherical envelope," published in 1867, may
here be referred to. The author therein considers the stresses set up in the envelope by
a system of forces applied at its surface, and ultimately solves the problem for two normal
forces applied at any two points. The solution, in which he makes use of the principle
<>f inversion as it is applied in various electrical questions, turns ultimately on the deter-
mination of a certain function first introduced by Sir George Airy, and called by Maxwell

PREFACE. XXvii

Airy's Function of Stress. The methods which in this paper were attended with so much
success, seem to have suggested to Maxwell a reconsideration of his former work, with the
view of extending the character of the reciprocity therein established. Accordingly in 1870
there appeared his fourth contribution to the subject, "On reciprocal figures, frames and
diagrams of forces." This important memoir was published in the Transactions of the Royal
Society of Edinburgh, and its author received for it the Keith Prize. He begins with a
remarkably beautiful construction for drawing plane reciprocal diagrams, and then proceeds
to discuss the geometry and the degrees of freedom and constraint of polyhedral frames,
his object being to lead up to the limiting case when the faces of the polyhedron become
infinitely small and form parts of a continuous surface. In the course of this work he
obtains certain results of a general character relating to inextensible surfaces and certain
others of practical utility relating to loaded frames. He then attacks the general problem of
representing graphically the internal stress of a body and by an extension of the meaning
of "Diagram of Stress," he gives a construction for finding a diagram which has mechanical
as well as geometrical reciprocal properties with the figure suppose^ to be under stress. It
is impossible with brevity to give an account of this reciprocity, the development of which
in Maxwell's hands forms a very beautiful example of analysis. It will be sufficient to
state that under restricted conditions this diagram of stress leads to a solution for the
components of stress in terms of a single function analogous to Airy's Function of Stress.
In the remaining parts of the memoir there is a discussion of the equations of stress, and
it is shewn that the general solution may be expressed in terms of three functions analogous
to Airy's single function in two dimensions. These results are then applied to special
cases, and in particular the stresses in a horizontal beam with a uniform load on its upper
surface are fully investigated.

On the subjects in which Maxwell's investigations were the most numerous it has
been thought necessary, in the observations which have been made, to sketch out briefly
the connections of the various papers on each subject with one another. • It is not how-
ever intended to enter into an account of the contents of his other contributions to science,
and this is the less necessary as the reader may readily obtain the information he may
require in Maxwell's own language. It was usually his habit to explain by way of
introduction to any paper his exact position with regard to the subject matter and to
give a brief account of the nature of the work he was contributing. There are however
several memoirs which though unconnected with others are exceedingly interesting in them-
selves. Of these the essay on Saturn's Rings will probably be thought the most important
as containing the solution of a difficult cosmical problem; there are also various papers on
Dynamics, Hydromechanics and subjects of pure mathematics, which are most useful con-
tributions on the subjects of which they treat.

The remaining miscellaneous papers may be classified under the following heads: (a)
Lectures and Addresses, (6) Essays or Short Treatises, (c) Biographical Sketches, (d) Criticisms

and Reviews.

d 2

.Ill I'ltKK.V

Class (a) comprises his addresses to the British Association, to the London Mathematical
Society, the Rede Lecture at Cambridge, his address at the opening of the Cavendish
Laboratory and his Lectures at the Royal Institution and to the Chemical Society.

Class (6) includes all but one of the articles which he contributed to the Encyclo-
pedia Britannica and several others of a kindred character to Nature.

Class (c) contains such articles as " Faraday " in the Encyclopedia Britannica and
"Helmholtz" in Nature.

Class (d) is chiefly occupied with the reviews of scientific books as they were pub-
lished. These appeared in Nature and the most important have been reprinted in these

In some of these writings, particularly those in class (6), the author allowed himself a
greater latitude in the use of mathematical symbols and processes than in others, as
for instance in the article " Capillary Attraction," which is in fact a treatise on that subject
treated mathematically. The lectures were upon one or other of the three departments
of Physics with which he had mainly occupied himself; — Colour Perception, Action through
a Medium, Molecular Physics; and on this account they are the more valuable. In the
whole series of these more popular sketches we find the same clear, graceful delineation of
principles, the same beauty in arrangement of subject, the same force and precision in
expounding proofs and illustrations. The style is simple and singularly free from any kind
of haze or obscurity, rising occasionally, as in his lectures, to a strain of subdued eloquence
when the emotional aspects of the subject overcome the purely speculative.

The books which were written or edited by Maxwell and published in his lifetime but
which are not included in this collection were the "Theory of Heat" (1st edition, 1871);
"Electricity and Magnetism" (1st edition, 1873); "The Electrical Researches of the Hon-
ourable Henry Cavendish, F.R.S., written between 1771 and 1781, edited from the original
manuscripts in the possession of the Duke of Devonshire, K.G." (1879). To these may be
added a graceful little introductory treatise on Dynamics entitled " Matter and Motion "
(published in 1876 by the Society for promoting Christian Knowledge). Maxwell also
contributed part of the British Association Report on Electrical Units which was afterwards
published in book form by Fleeming Jenkin.

The "Theory of Heat" appeared in the Text Books of Science series published by
Longmans, Green and Co., and was at once hailed as a beautiful exposition of a subject,
part of which, and that the most interesting part, the mechanical theory, had as yet but
commenced the existence which it owed to the genius and labours of Rankine, Thomson and
Clausius. There is a certain charm in Maxwell's treatise, due to the freshness and originality
of its expositions which has rendered it a great favourite with students of Heat.

After his death an " Elementary Treatise on Electricity," the greater part of which he
had written, was completed by Professor Garnett and published in 1881. The aim of this

PREFACE. Xxix

treatise and its position relatively to his larger work may be gathered from the following
extract from Maxwell's preface.

"In this smaller book I have endeavoured to present, in as compact a form as I
can, those phenomena which appear to throw light on the theory of electricity and to
use them, each in its place, for the development of electrical ideas in the mind of
the reader."

"In the larger treatise I sometimes made use of methods which I do not think
the best in themselves, but without which the student cannot follow the investigations
of the founders of the Mathematical Theory of Electricity. I have since become more
convinced of the superiority of methods akin to those of Faraday, and have therefore
adopted them from the first."

Of the "Electricity and Magnetism" it is difficult to predict the future, but there is
no doubt that since its publication it has given direction and colour to the study of
Electiical Science. It was the master's last word upon a subject t9 which he had devoted
several years of his life, and most of what he wrote found its proper place in the treatise.
Several of the chapters, notably those on Electromagnetism, are practically reproductions of
his memoirs in a modified or improved form. The treatise is also remarkable for the handling
of the mathematical details no less than for the exposition of physical principles, and is
enriched incidentally by chapters of much originality on mathematical subjects touched on
in the course of the work. Among these may be mentioned the dissertations on Spherical
Hannonics and Lagrange's Equations in Dynamics.

The origin and growth of Maxwell's ideas and conceptions of electrical action, cul-
minating in his treatise where all these ideas are arranged in due connection, form an
interesting chapter not only in the history of an individual mind but in the history of
electrical science. The importance of Faraday's discoveries and speculations can hardly be
overrated in their influence on Maxwell, who tells us that before he began the study of
electricity he resolved to read none of the mathematics of the subject till he had first
mastered the "Experimental Researches." He was also at first under deep obligations to
the ideas contained in the exceedingly important papers of Sir W. Thomson on the analogy
between Heat-Conduction and Statical Electricity and on the Mathematical Theory of
Electricity in Equilibrium. In his subsequent efforts we must perceive in Maxwell, possessed
of Faraday's views and embued with his spirit, a vigorous intellect bringing to bear on a
subject still full of obscurity the steady light of patient thought and expending upon it
all the resources of a never failing ingenuity.

ROYAL NAVAL COLLEGE,
GREENWICH,

August, 1890.

TABLE OF CONTENTS.

PAGE

I. On the Description of Oval Curves and those having a plurality of Foci; ivith

remarks by Professor Forbes ......... 1

II. On the Theory of Rolling Curves 4

III. On the Equilibrium of Elastic Solids 30

Solutions of Problems .......... ^ . 74

IV. On the Transformation of Surfaces by Bending 80

V. On a particular case of the descent of a heavy body in a resisting medium . 115

VI. On the Theory of Colours in relation to Colour-Blindness 119

VII. Experiments on Colour as perceived by the Eye, with remarks on Colour-Blindness 126

VIII. On Faraday's Lines of Force / . . . .155

IX. Description of a New Form of the Piatometer, an Instrument for measuring the

areas of Plane Figures drawn on paper 230

X. On the elementary theory of Optical Instruments ... . . 238

XI. On a method of drawing the Theoretical Forms of Faraday's Lines of Force

without calculation .... 241

XII. On the unequal sensibility of the Foramen Centrale to Light of different Colours 242

XIII. On the Theory of Compound Colours with reference to mixtures of Blue and

Yellow Light . ...... .243

XIV. On an instrument to illustrate Poinsot's Theory of Rotation . 246

XV. On a Dynamical Top, for exhibiting the phenomena of the motions of a body of

invariable form about a fixed point, with some suggestions as to the Earth's

motion 248

XVI. Account of Experiments on the Perception of Colour . . • 263

XVII. On the general laws of Optical Instruments . .271

XVIII. On Theories of the Constitution of Saturn's Rings • 286

XIX. On the stability of the motion of Saturn s Rings

XX. Illustrations of the Dynamical Theory of Gases . . • &••

XXI. On the Theory of Compound Colours and the Relations of the Colours of the

Spectrum • •

XXII. On the Theory of Three Primary Colours . • 445

XXIII. On Physical Lines of Force .... .451

XXIV. On Reciprocal Figures and Diagrams of Forces .

XXV. A Dynamical Theory of the Electromagnetic Field

XXVI. On the Calculation of the Equilibrium and Stiffness of Frames . 598

ERRATA.

Page 40. In the first of equations (12), second group of terms, read

instead of

with corresponding changes in the other two equations.

Page 153, five lines from bottom of page, read 127 instead of 276.

[From the Proceedings of the Royal Society of Edinburgh, Vol. II. April, 1846.]

I. On the Description of Oval Curves, and those having a plurality of Foci; with
remarks by Professor Forbes, Communicated by PROFESSOR FORBES.

MB CLERK MAXWELL ingeniously suggests the extension of the common
theory of the foci of the conic sections to curves of a higher degree of com-
plication in the following manner : —

(1) As in the ellipse and hyperbola, any point in the curve has the
sum, or difference of two lines drawn from two points or foci = a constant
quantity, so the author infers, that curves to a certain degree analogous, may
be described and determined by the condition that the simple distance from
one focus plus a multiple distance from the other, may be = a constant quantity;
or more generally, m times the one distance + n times the other = constant.

(2) The author devised a simple mechanical means, by the wrapping
of a thread round pins, for producing these curves. See Figs. 1 and 2. He

Fig. 1. Two Foci. Ratios 1, 2.

Fig. 2. Two Foci. Eatios 2, 3.

then thought of extending the principle to other curves, whose property
should be, that the sum of the simple or multiple distances of any point of
VOL. I. *

2 DESCRIPTION OF OVAL CURVES.

the curve from three or more points or foci, should be = a constant quantity ;
and this, too, he has effected mechanically, by a very simple arrangement of
a string of given length passing round three or more fixed pins, and con-
straining a tracing point, P. See Fig. 3. Farther, the author regards curves

Fig. 3. Three Foci. Ratios of Equality.

of the first kind as constituting a particular class of curves of the second
kind, two or more foci coinciding in one, a focus in which two strings meet
being considered a double focus; when three strings meet a treble focus, &c.

Professor Forbes observed that the equation to curves of the first class is
easily found, having the form

which is that of the curve known under the name of the First Oval of
Descartes*. Mr Maxwell had already observed that when one of the foci was
at an infinite distance (or the thread moved parallel to itself, and was confined
in respect of length by the edge of a board), a curve resembling an ellipse
was traced ; from which property Professor Forbes was led first to infer the
identity of the oval with the Cartesian oval, which is well known to have this
property. But the simplest analogy of all is that derived from the method of
description, r and r being the radients to any point of the curve from the -two

foci;

mr + nr = constant,

which in fact at once expresses on the undulatory theory of light the optical
character of the surface in question, namely, that light diverging from one
focus F without the medium, shall be correctly convergent at another point f

* Herachel, On Light, Art 232 ; Lloyd, On LigfU and Vision, Chap. vn.

DESCRIPTION OP OVAL CURVES. 3

tn

within it ; and in this case the ratio expresses the index of refraction of

tin

the medium*.

If we denote by the power of either focus the number of strings leading
to it by Mr Maxwell's construction, and if one of the foci be removed to an
infinite distance, if the powers of the two foci be equal the curve is a parabola;
if the power of the nearer focus be greater than the other, the curve is an
ellipse ; if the power of the infinitely distant focus be the greater, the curve
is a hyperbola. The first case evidently corresponds to the case of the reflection
of parallel rays to a focus, the velocity being unchanged after reflection ; the
second, to the refraction of parallel rays to a focus in a dense medium (in
which light moves slower) ; the third case to refraction into a rarer medium.

The ovals of Descartes were described in his Geometry, where he has also
given a mechanical method of describing one of themf, but only in a particular
case, and the method is less simple than Mr Maxwell's. The demonstration of
the optical properties was given by Newton in the Principia, Book i., prop. 97,
by the law of the sines; and by Huyghens in 1690, on the Theory of Undu-
lations in his Traite de la Lumiere. It probably has not been suspected that
so easy and elegant a method exists of describing these curves by the use of
a thread and pins whenever the powers of the foci are commensurable. For
instance, the curve, Fig. 2, drawn with powers 3 and 2 respectively, give the
proper form for a refracting surface of a glass, whose index of refraction is 1'50,
in order that rays diverging from f may be refracted to F.

As to the higher classes of curves with three or more focal points, we
cannot at present invest them with equally clear and curious physical properties,
but the method of drawing a curve by so simple a contrivance, which shall
satisfy the condition

mr + nr +pr" + &c. = constant,

is in itself not a little interesting; and if we regard, with Mr Maxwell, the
ovala above described, as the limiting case of the others by the coalescence
of two or more foci, we have a farther generalization of the same kind as that
so highly recommended by Montucla|, by which Descartes elucidated the conic
sections as particular cases of his oval curves.

* This was perfectly well shewn by Huyghens in his Traite de la Lumiere, p. 111. (1690.)

t Edit. 1683. Geometria, Lib. n. p. 54.

J Histoire des Mathematlques. First Edit. n. 102.

1—2

[From the Transactions of the Royal Society of Edinburgh, Vol. xvi. Pan \ .]

II. On the Tlicory of Rolling Curves. Communicated by the Rev. Professor

KELLAND.

THERE is an important geometrical problem which proposes to find a curve
having a given relation to a series of curves described according to a given
law. This is the problem of Trajectories in its general form.

The series of curves is obtained from the general equation to a curve by
the variation of its parameters. In the general case, this variation may change
the form of the curve, but, in the case which we are about to consider, the
curve is changed only in position.

This change of position takes place partly by rotation, and partly by trans-
ference through space. The rolling of one curve on another is an example of
this compound motion.

As examples of the way in which the new curve may be related to the
series of curves, we may take the following : —

1. The new curve may cut the series of curves at a given angle. When
this angle becomes zero, the curve is the envelope of the series of curves.

2. It may pass through corresponding points in the series of curves.
There are many other relations which may be imagined, but we shall confine
our attention to this, partly because it affords the means of tracing various
curves, and partly on account of the connection which it has with many
geometrical problems.

Therefore the subject of this paper will be the consideration of the relations
of three curves, one of which is fixed, while the second rolls upon it and
traces the third. The subject of rolling curves is by no means a new one.
The first idea of the cycloid is attributed to Aristotle, and involutes and
evolutes have been long known.

THE THEORY OF ROLLING CURVES. 5

In the History of the Royal Academy of Sciences for 1704, .page 97,
there is a memoir entitled "Nouvelle formation des Spirales," by M. Varignon,
in which he shews how to construct a polar curve from a curve referred to
rectangular co-ordinates by substituting the radius vector for the abscissa, and
a circular arc for the ordinate. After each curve, he gives the curve into
which it is "unrolled," by which he means the curve which the spiral must
be rolled upon in order that its pole may trace a straight line; but as this
is not the principal subject of his paper, he does not discuss it very fully.

There is also a memoir by M. de la Hire, in the volume for 1706, Part n.,
page 489, entitled " Methode generale pour rdduire toutes les Lignes courbes a
des Koulettes, leur generatrice ou leur base etant donne"e telle qu'on voudra."

M. de la Hire treats curves as if they were polygons, and gives geome-
trical constructions for finding the fixed curve or the rolling curve, the other
two being given ; but he does not work any examples. /

In the volume for 1707, page 79, there is a paper entitled, "Methode
generale pour determiner la nature des Courbes formers par le roulement de
toutes sortes de Courbes sur une autre Courbe quelconque." Par M. Nicole.

M. Nicole takes the equations of the three curves referred to rectangular
co-ordinates, and finds three general equations to connect them. He takes the
tracing-point either at the origin of the co-ordinates of the rolled curve or not.
He then shews how these equations may be simplified in several particular
cases. These cases are—

(1) When the tracing-point is the origin of the rolled curve.

(2) When the fixed curve is the same as the rolling curve.

(3) When both of these conditions are satisfied.

(4) When the fixed line is straight.

He then says, that if we roll a geometric curve on itself, we obtain a new
geometric curve, and that we may thus obtain an infinite number of geometric
curves.

The examples which he gives of the application of his method are all taken
from the cycloid and epicycloid, except one which relates to a parabola, rolling
on itself, and tracing a cissoid with its vertex. The reason of so small a
number of examples being worked may be, that it is not easy to eliminate
the co-ordinates of the fixed and rolling curves from his equations.

The case in which one curve rolling on another produces a circle is treated
of in Willis's Principles of Mechanism. Class C. Rolling Contact.

6 THE THEORY OP ROLLING CURVES.

He employs the same method of finding the one curve from the other
which is used here, and he attributes it to Euler (see the Acta Petropolitana,
Vol. v.).

Thus, nearly all the simple cases have been treated of by different authors;
but the subject is still far from being exhausted, for the equations have been
applied to very few curves, and we may easily obtain new and elegant proper-
ties from any curve we please.

Almost all the more notable curves may be thus linked together in a great
variety of ways, so that there are scarcely two curves, however dissimilar,
between which we cannot form a chain of connected curves.

This will appear in the list of examples given at the end of this paper.

Let there be a curve KAS, whose pole is at C.

THE THEORY OF ROLLING CURVES. 7

Let the angle DCA = 6l and CA=r^ and let

Let this curve remain fixed to the paper.

Let there be another curve BAT, whose pole is B.

Let the angle MBA = di} and BA = r3, and let

Let this curve roll along the curve KAS without slipping.
Then the pole B will describe a third curve, whose pole is C.
Let the angle DCB=6a, and CB = r3> and let

We have here six unknown quantities ^iff^fff, ; but we have only three
equations given to connect them, therefore the other three" must be sought for
in the enunciation.

But before proceeding to the investigation of these three equations, we must
premise that the three curves will be denominated as follows : —

The Fixed Curve, Equation, 61 = fa (r^.
The Rolled Curve, Equation, 01 = fa(r^.
The Traced Curve, Equation, 6, = fa (r3).

,

When it is more convenient to make use of equations between rectangular
co-ordinates, we shall use the letters xylf x^, x3y3. We shall always employ the
letters $,&,$» to denote the length of the curve from the pole, p^p3 for the per-
pendiculars from the pole on the tangent, and qtq.2q3 for the intercepted part of
the tangent.

Between these quantities, we have the following equations : —

x
= rsin 0,

s =

ydx — xdy

=

8 THE THEORY OF ROLLING CURVES.

rdr

~d8

/^7/^V'
V ' + (

We come now to consider the three equations of rolling which are involved
in the enunciation. Since the second curve rolls upon the first without slipping,
the length of the fixed curve at the point of contact is the measure of the
length of the rolled curve, therefore we have the following equation to connect
the fixed curve and the rolled curve —

s, = sr

Now, by combining this equation with the two equations

it is evident that from any of the four quantities Of&r, or x^ylx^t, we can
obtain the other three, therefore we may consider these quantities as known
functions of each other.

Since the curve rolls on the fixed curve, they must have a common tangent.

Let PA be this tangent, draw BP, CQ perpendicular to PA, produce CQ,
and draw BR perpendicular to it, then we have CA=rlt BA=rt, and CB = rt;
CQ=p>, PB=pt, and BN=pt; AQ = qlt AP = qt, and

Also r,>

=Pl* + 2plPt
r,' = r,1 + r*

Since the first curve is fixed to the paper, we may find the angle 6r
Thus

t + tan- — JV- + tan

THE THEORY OF ROLLING CURVES. 9

Thus we have found three independent equations, which, together with the
equations of the curves, make up six equations, of which each may be deduced
from the others. There is an equation connecting the radii of curvature of the
three curves which is sometimes of use.

The angle through which the rolled curve revolves during the description of
the element ds3, is equal to the angle of contact of the fixed curve and the
rolling curve, or to the sum of their curvatures,

ds3 dsl ds,

7~=7z;+7^-

But the radius of the rolled curve has revolved in the opposite direction
through an angle equal to d$lt therefore the angle between two successive posi-

ds
tions of rt is equal to ^ — cZ#2. Now this angle is the angle between two

**2

successive positions of the normal to the traced curve, therefore, if O be the
centre of curvature of the traced curve, it is the angle which ds3 or dsl subtends
at 0. Let OA = T, then

ds, _ ^j _ 3 ,. _ t , ,*
~~ ~~~' "'•" "

, _ ^ __
'ds, T~Rl+Ra ds,'

' r, \T

As an example of the use of this equation, we may examine a property
of the logarithmic spiral.

In this curve, p = mr, and R = — , therefore if the rolled curve be the

m

logarithmic spiral

1\ 1m

m _

T~^'

AO

therefore AO in the figure = mRl, and -p-=m.

•*%

Let the locus of O, or the evolute of the traced curve LYBH, be the
curve OZY, and let the evolute of the fixed curve KZAS be FEZ, and let
us consider FEZ as the fixed curve, and OZY as the traced curve.

VOL. I. 2

10 THE THEORY OF ROLLING CURVES.

Then in the triangles BPA, AOF, we have OAF=PBA, and ^p=m = ^jr

therefore the triangles are similar, and FOA = APE = - , therefore OF is perpen-

dicular to OA, the tangent to the curve OZY, therefore OF is the radius of
the curve which when rolled on FEZ traces OZY, and the angle which the
curve makes with this radius is OFA = PAB — sva.~1m, which is constant, there-
fore the curve, which, when rolled on FEZ, traces OZY, is the logarithmic
spiral. Thus we have proved the following proposition : " The involute of the
curve traced by the pole of a logarithmic spiral which rolls upon any curve,
is the curve traced by the pole of the same logarithmic spiral when rolled on
the involute of the primary curve."

It follows from this, that if we roll on any curve a curve having the
property 2)\s*miri> an^ ro^ another curve having pt = mtrt on the curve traced,
and so on, it is immaterial in what order we roll these curves. Thus, if we
roll a logarithmic spiral, in which p = mr, on the nth involute of a circle whose
radius is a, the curve traced is the w + lth involute of a circle whose radius

is >/l — m*.

Or, if we roll successively m logarithmic spirals, the resulting curve is the
n + mth involute of a circle, whose radius is

We now proceed to the cases in which the solution of the problem may
be simplified. This simplification is generally effected by the consideration that
the radius vector of the rolled curve is the normal drawn from the traced
curve to the fixed curve.

In the case in which the curve is rolled on a straight line, the perpen-
dicular on the tangent of the rolled curve is the distance of the tracing point
from the straight line; therefore, if the traced curve be defined by an equation
in xt and yu

THE THEORY OF ROLLING CURVES. 11

By substituting for rt in the first equation, its value, as derived from the
second, we obtain

Jdx^\(dx3Y I /dr.y

(dyJ LWJ+] W'

If we know the equation to the rolled curve, we may find |-^| in

\dffj

terms of ri} then by substituting for r2 its value in the second equation, we

have an equation containing x3 and -y-?, from which we find the value of ~

dy* dy,

in terms of x3; the integration of this gives the equation of the traced curve.

As an example, we may find the curve traced by the pole of a hyperbolic
spiral which rolls on a straight line.

The equation of the rolled curve is #3 = — , /

a?

dy

dx3 x3
' dy3~

This is the differential equation of the tractory of the straight line, which
is the curve traced by the pole of the hyperbolic spiral.
By eliminating x3 in the two equations, we obtain

drt _ idx3

This equation serves to determine the rolled curve when the traced curve
is given.

As an example we shall find the curve, which being rolled on a straight
line, traces a common catenary.

Let the equation to the catenary be

a

2—2

12 THE THEORY OF ROLLING CURVES.

Then

' ' 1 7//J / = ^«" 1,1*. \ » r* '

then by integration 0 =003"' I— — 1 J ,

2a

7* ~~

l + cos0'

This is the polar equation of the parabola, the focus being the pole ; there-
fore, if we roll a parabola on a straight line, its focus will trace a catenary.

The rectangular equation of this parabola is a? = <iay, and we shall now
consider what curve must be rolled along the axis of y to trace the parabola.

By the second equation (2),

T + l, but Xt=p»

but qt is the perpendicular on the normal, therefore the normal to the curve
always touches a circle whose radius is 2a, therefore the curve is the involute
of this circle.

Therefore we have the following method of describing a catenary by con-
tinued motion.

Describe a circle whose radius is twice the parameter of the catenary; roll a
straight line on this circle, then any point in the line will describe an involute

THE THEORY OF BOLLING CURVES. 13

of the circle ; roll this curve on a straight line, and the centre of the circle will
describe a parabola ; roll this parabola on a straight line, and its focus will trace
the catenary required.

We come now to the case in which a straight line rolls on a curve.

When the tracing-point is in the straight line, the problem becomes that
of involutes and evolutes, which we need not enter upon ; and when the tracing-
point is not in the straight line, the calculation is somewhat complex ; we shall
therefore consider only the relations between the curves described in the first
and second cases.

Definition. — The curve which cuts at a given angle all the circles of a
given radius whose centres are in a given curve, is called a tractory of the
given curve.

Let a straight line roll on a curve A, and let a point in the straight
line describe a curve B, and let another point, whose distance from the first
point is b, and from the straight line a, describe a curve C, then it is evident
that the curve B cuts the circle whose centre is in C, and whose radius is b,

at an angle whose sine is equal to T, therefore the curve B is a tractory of
the curve C.

When a = b, the curve B is the orthogonal tractory of the curve C. If
tangents equal to a be drawn to the curve B, they will be terminated in
the curve C; and if one end of a thread be carried along the curve C, the
other end will trace the curve B.

When a = 0, the curves B and C are both involutes of the curve A,
they are always equidistant from each other, and if a circle, whose radius is
6, be rolled on the one, its centre will trace the other.

If the curve A is such that, if the distance between two points measured
along the curve is equal to b, the two points are similarly situate, then the
curve B is the same with the curve C. Thus, the curve A may be a re-
entrant curve, the circumference of which is equal to b.

When the curve A is a circle, the curves B and C are always the same.
The equations between the radii of curvature become

11 r

14 THE THEORY OF ROLLING CURVES.

When a = 0, T=0, or the centre of curvature of the curve B is at the
point of contact. Now, the normal to the curve C passes through this point,
therefore —

"The normal to any curve passes through the centre of curvature of its
tractory."

In the next case, one curve, by rolling on another, produces a straight
line. Let this straight line be the axis of yt then, since the radius of the
rolled curve is perpendicular to it, and terminates in the fixed curve, and
since these curves have a common tangent, we have this equation,

rf<A_ .^0,
^c&T ' drt"

10

If the equation of the rolled curve be given, find -,-? in terms of rs, sub-

ar,

stitute xt for r,, and multiply by xlt equate the result to 3?-, and integrate.

Thus, if the equation of the rolled curve be

8 = Ar-" + &c. + Kr~* + Lr~l + M\og r + Nr + &c.

-f- = - nAx~* - &c. - 2Kx~* - Lx~l + M+ NX + &c. + nZxn,

. .

'• ^~ I lv "j" 1

which is the equation of the fixed curve.

If the equation of the fixed curve be given, find -r- in terms of x, sub-

stitute r for z, and divide by r, equate the result to -.-, and integrate.

Thus, if the fixed curve be the orthogonal tractory of the straight line,
whose equation is

alos

-
a + V a* — x1

dy _ N/a* — x*
dx~ x

THE THEORY OF ROLLING CURVES.

15

$

-ir I<t~
= cos l -- ./-=— 1,

a V f

this is the equation to the orthogonal tractory of a circle whose diameter is
equal to the constant tangent of the fixed curve, and its constant tangent
equal to half that of the fixed curve.

This property of the tractory of the circle may be proved geometrically,
thus — Let P be the centre of a circle whose radius is PD, and let CD be
a line constantly equal to the radius. Let BCP be the curve described by
the point (7 when the point D is moved along the circumference of the circle,
then if tangents equal to CD be drawn to the curve, their extremities will
be in the circle. Let ACH be the curve on which BCP' rolls, and let OPE
be the straight line traced by the pole, let CDE be the common tangent,
let it cut the circle in D, and the straight line in E.

Then CD = PD, .: LDCP=LDPC, and CP is perpendicular to OE,
.-. LCPE= LDCP+ LDEP. Take away LDCP= L DPC, and there remains
DPE=DEP, /

16 THE THEORY OF ROLLING CURVIX

Therefore the curve ACH has a constant tangent equal to the diameter of
the circle, therefore ACH is the orthogonal tractory of the straight line, which
is the tractrix or equitangential curve.

The operation of finding the fixed curve from the rolled curve is what
Sir John Leslie calls " divesting a curve of its radiated structure."

The method of finding the curve which must be rolled on a circle to
trace a given curve is mentioned here because it generally leads to a double
result, for the normal to the traced curve cuts the circle in two points, either
of which may be a point in the rolled curve.

Thus, if the traced curve be the involute of a circle concentric with the
given circle, the rolled curve is one of two similar logarithmic spirals.

If the curve traced be the spiral of Archimedes, the rolled curve may be
either the hyperbolic spiral or the straight line.

In the next case, one curve rolls on another and traces a circle.

Since the curve traced is a circle, the distance between the poles of the
fixed curve and the rolled curve is always the same ; therefore, if we fix the
rolled curve and roll the fixed curve, the curve traced will still be a circle,
and, if we fix the poles of both the curves, we may roll them on each other
without friction.

Let a be the radius of the traced circle, then the sum or difference of
the radii of the other curves is equal to a, and the angles which they make
with the radius at the point of contact are equal,

dd
If we know the equation between 0, and r,, we may find -, ' in terms of

substitute ± (a ± rt) for r,, multiply by '—^t and integrate.

***

Thus, if the equation between 0, and r, be

r, = a sec 0,,

THE THEORY OP ROLLING CURVES. 17

which is the polar equation of a straight line touching the traced circle whose
equation is r = a, then

r* ~ TI •Jrf-a?

gfl, _rt±a o_

<&• r* (r, ± a) Jr*±2r,a

a

2a 2a

Ts = — = .

Now, since the rolling curve is a straight line, and the tracing point is
not in its direction, we may apply to this example the observations which
have been made upon tractories.

Let, therefore, the curve r = -~ — - be denoted by A, its involute by B, and

v — I

the circle traced by C, then B is the tractory of C ; therefore the involute

2ct
of the curve r = ^—r is the tractory of the circle, the equation of which is

_.r la- __ . . 2a

0 = cos ] /— — I. The curve whose equation is r=7f. — - seems to be among

. a \J r" 6f — I

spirals what the catenary is among curves whose equations are between rec-
tangular co-ordinates ; for, if we represent the vertical direction by the radius
vector, the tangent of the angle which the curve makes with this line is
proportional to the length of the curve reckoned from the origin ; the point
at the distance a from a straight line rolled on this curve generates a circle,
and when rolled on the catenary produces a straight line ; the involute of this
curve is the tractory of the circle, and that of the catenary is the tractory
of the straight line, and the tractory of the circle rolled on that of the straight
line traces the straight line ; if this curve is rolled on the catenary, it produces
the straight line touching the catenary at its vertex; the method of drawing

VOL. I. 3

18 THE THEORY OF ROLLING CURVES.

tangents is the same as in the catenary, namely, by describing a circle whose
radius is a on the production of the radius vector, and drawing a tangent to the
circle from the given point.

In the next case the rolled curve is the same as the fixed curve. It is
evident that the traced curve will be similar to the locus of the intersection
of the tangent with the perpendicular from the pole ; the magnitude, however,
of the traced curve will be double that of the other curve ; therefore, if we
call r» = <&& the equation to the fixed curve, r1 = <f>l0l that of the traced curve,
we have

?o ^ '

also, £ = £'.

Similarly, rt = 2Pl = Zr, = 4 4r0 , 0, = 0, - 2 cos'1

r» rt \ro/

Similarly, r.= 2p._, - 2rn., & &c. = 2V0 (|?V ,

r* \roi

and £?=£•

Let en become 6J ; 0,, 0.1 and •&, . Let 0nl-0n = a,

a = $J - en = 0t>-0a-n cos'1 ^i + » cos'1

THE THEORY OP ROLLING CURVES. 19

Now, cos"1 ^ is the complement of the angle at which the curve cuts the

_. _ np fY) *

radius vector, and cos'1 — — cos"1 « is the variation of this angle when On varies
by an angle equal to a. Let this variation = <f> ; then if 00 - 00l = /3,

n n'
Now, if n increases, <j> will diminish ; and if n becomes infinite,

<j> = ^5- + ^5- = 0 when a and /8 are finite.

Therefore, when n is infinite, $ vanishes ; therefore the curve cuts the radius
vector at a constant angle ; therefore the curve is the logarithmic spiral.

Therefore, if any curve be rolled on itself, and the operation repeated an
infinite number of times, the resulting curve is the logarithmic spiral.

Hence we may find, analytically, the curve which, being rolled on itself,
traces itself.

For the curve which has this property, if rolled on itself, and the operation
repeated an infinite number of times, will still trace itself.

But, by this proposition, the resulting curve is the logarithmic spiral ;
therefore the curve required is the logarithmic spiral. As an example of a curve
rolling on itself, we will take the curve whose equation is

cos—

TT dr I v\ / 0Y1'1

Here — -~=2"a sin- cos- ;

a#0 V n \ nl

\ i \ f

m
cos

.'. r. = 2»n=2 — i= = ,

' i\--» / 6\ / 0\^-"

— 1 I cos —

nl \ n] \ n

\n+i

2"a

\ nl I

COS —

/

3—2

20 THE THEORY OF ROLLING CURVES.

n n

Now 0,-0.= -cos-l^= -cos'1 cos -* = -',

7",

• 0 -0
"'-

substituting this value of 0, in the expression for r,,

0 \"+1

similarly, if the operation be repeated m times, the resulting curve is

/ 6 \n+m
rm = 2n+ma ( cos — -

When n=l, the curve is

r = 2a cos 0,

the equation to a circle, the pole being in the circumference.
When w = 2, it is the equation to the cardioid

r = 4« (cos^) .
\ /

In order to obtain the cardioid from the circle, we roll the circle upon
itself, and thus obtain it by one operation ; but there is an operation which,
being performed on a circle, and again on the resulting curve, will produce a
cardioid, and the intermediate curve between the circle and cardioid is

?• =

As the operation of rolling a curve on itself is represented by changing n
into (n+1) in the equation, so this operation may be represented by changing n
into

Similarly there may be many other fractional operations performed upon
the curves comprehended under the equation

f)\

/ f)\n
=2"a (cos - ) .
\ V

We may also find the curve, which, being rolled ou itself, will produce a
given curve, by making n= —1.

THE THEOEY OF . EOLLING CUEVES. 21

We may likewise prove by the same method as before, that the result of
performing this inverse operation an infinite number of times is the logarithmic
spiral

As an example of the inverse method, let the traced line be straight, let
its equation be

r. = 2a sec ft,

, 0,

then P=i=& = to=

therefore suppressing the suffix,

r4

•-ar,

• \de) a.

1

dr

r I

.. /2a _ \
.'. ft = COS M 1 ),

2a
~l-cos0'

the polar equation of the parabola whose parameter is 4a.

The last case which we shall here consider affords the means of constructing
two wheels whose centres are fixed, and which shall roll on each other, so that
the angle described by the first shall be a given function of the angle described
by the second.

T A a ^ i d&s I",

-Let P, = <pc/i, then rl + r, = a, and -jy = — >

do, r.

_

d&l ~ a— »*i '

Let us take as an example, the pair of wheels which will represent the
angular motion of a comet in a parabola.

THE THEORY OF ROLLING CURVES.

0, d6t 1 r,
Here *, = tan-, ,.^ = __ = _

r,

' a 2 + COS0,'

therefore the first wheel is an ellipse, whose major axis is equal to $ of the
distance between the centres of the wheels, and in which the distance between
the foci is half the major axis.

Now since 0, = 2 tan'1 0, and rt = a-rv

which is the equation to the wheel which revolves with constant angular velocity.

Before proceeding to give a list of examples of rolling curves, we shall
state a theorem which is almost self-evident after what has been shewn pre-
viously.

Let there be three curves, A, B, and C. Let the curve A, when rolled
on itself, produce the curve B, and when rolled on a straight line let it
produce the curve C, then, if the dimensions of C be doubled, and B be
rolled on it, it will trace a straight line.

A Collection of Examples of Rotting Curves.

First. Examples of a curve rolling on a straight line.

Ex. 1. When the rolling curve is a circle whose tracing- point is in the
circumference, the curve traced is a cycloid, and when the point is not in the
circumference, the cycloid becomes a trochoid.

Ex. 2. When the rolling curve is the involute of the circle whose radius
is 2a, the traced curve is a parabola whose parameter is 4a.

THE THEORY OF ROLLING CURVES. 23

Ex. 3. When the rolled curve is the parabola whose parameter is 4a, the
traced curve is a catenary whose parameter is a, and whose vertex is distant
a from the straight line.

Ex. 4. When the rolled curve is a logarithmic spiral, the pole traces a
straight line which cuts the fixed line at the same angle as the spiral cuts
the radius vector.

Ex. 5. When the rolled curve is the hyperbolic spiral, the traced curve
is the tractory of the straight line.

Ex. 6. When the rolled curve is the polar catenary

the traced curve is a circle whose radius is a, and which touches the straight
line.

Ex. 7. When the equation of the rolled curve is

a
the traced curve is the hyperbola whose equation is

Second. In the examples of a straight line rolling on a curve, We shall
use the letters A, B, and C to denote the three curves treated of in page 22.

Ex. 1. When the curve A is a circle whose radius is a, then the curve B
is the involute of that circle, and the curve 0 is the spiral of Archimedes, r = a0.

Ex. 2. When the curve A is a catenary whose equation is

a ,

• — — I aa
' — 3TI C

the curve B is the tractory of the straight line, whose equation is

a* y.

y = a log - / + «/aa — X2,
^a + N/a'-x2

and C is a straight line at a distance a from the vertex of the catenary.

24 THE THEORY OF ROLLING CURVES.

Ex. 3. When the curve A is the polar catenary

the curve B is the tractory of the circle

and the curve C is a circle of which the radius is - .

Third. Examples of one curve rolling on another, and tracing a straight
line.

Ex. 1. The curve whose equation is

6 = Ar-" + &c. + Kr~* + Lr~l + Mlog
when rolled on the curve whose equation is

. .

traces the axis of y.

Ex. 2. The circle whose equation is r = acos# rolled on the circle whose
radius is a traces a diameter of the circle.

Ex. 3. The curve whose equation is

/2a . ... r

6— . -- 1— versm ' - ,
V r a

rolled on the circle whose radius is a, traces the tangent to the circle.

Ex. 4. If the fixed curve be a parabola whose parameter is 4a, and if we
roll on it the spiral of Archimedes r = a0, the pole will trace the axis of the
parabola.

Ex. 5. If we roll an equal parabola on it, the focus will trace the directrix
of the first parabola.

Ex. 6. If we roll on it the curve ^ = ja, the pole will trace the tangent
at the vertex of the parabola.

THE THEOEY OP ROLLING CURVES. 25

Ex. 7. If we roll the curve whose equation is

fa n
r = a cos I T 6

\b

on the ellipse whose equation is

the pole will trace the axis b.

Ex. 8. If we roll the curve whose equation is

on the hyperbola whose equation is

the pole will trace the axis b.

Ex. 9. If we roll the lituus, whose equation is

rf = 36'
on the hyperbola whose equation is

the pole will trace the asymptote.

Ex. 10. The cardioid whose equation is

r = a (1 +cos 6),
rolled on the cycloid whose equation is

y = a versin -1 - + Jzax — x1,
a

traces the base of the cycloid.

Ex. 11. The curve whose equation is

6 = versin"1 - + 2 /—-I,
a \j r

rolled on the cycloid, traces the tangent at the vertex.
VOL, I.

26 THE THEORY OF ROLLING CURVES.

Ex. 12. The straight line whose equation is

r = a sec 0,

rolled on a catenary whose parameter is a, traces a line whose distance from
the vertex is a.

Ex. 13. The part of the polar catenary whose equation is

rolled on the catenary, traces the tangent at the vertex.

Ex. 14. The other part of the polar catenary whose equation is

rolled on the catenary, traces a line whose distance from the vertex is equal to 2a.

Ex. 15. The tractory of the circle whose diameter is a, rolled on the
tractory of the straight line whose constant tangent is a, produces the straight
line.

Ex. 16. The hyperbolic spiral whose equation is

a
T=d>

rolled on the logarithmic curve whose equation is

x

i

a log - ,
°

a

traces the axis of y or the asymptote.

Ex. 17. The involute of the circle whose radius is a, rolled on an orthogonal
trajectory of the catenary whose equation is

x

traces the axis of y.

Ex. 18. The curve whose equation is

THE THEORY OF ROLLING CURVES. 27

rolled on the witch, whose equation is

traces the asymptote.

Ex. 19. The curve whose equation is

r = a tan 0,
rolled on the curve whose equation is

traces the axis of y.

Ex. 20. The curve whose equation is

rolled on the curve whose equation is

y?

, or r =

— of
traces the axis of y.

Ex. 21. The curve whose equation is

r = a (sec 9 — tan 6),
rolled on the curve whose equation is

traces the axis of y.

Fourth. Examples of pairs of rolling curves which have their poles at a fixed
distance = a.

ie straight line whose equation is ^=sec~1-.
Ex. 1. ' a

vThe polar catenary whose equation is 6 = ± . / 1

to

± r

Ex. 2. Two equal ellipses or hyperbolas centered at the foci.
Ex. 3. Two equal logarithmic spirals.

(Circle whose equation is r = 2a cos 6.

IT r

Curve whose equation is 6 = ^2 — 1+ versin J- .

4—2

28 THE THEORY OF ROLLING CURVES.

(Cardioid whose equation is r = 2a (1+ cos 0).

5- 1 r 1

[Curve whose equation is 0=sin~I- + log-^=

f Conchoid, r = a (sec 0 — I ).
Ex. 6. I -, r

(.Curve, 0 = A./ 1 — u +sec"1 -

V r* a

f Spiral of Archimedes, r = a0.

Ex- 7- r r

[Curve, 0 = - + log-.

ft ct

f Hyperbolic spiral, r = ^-

Ex. 8. j ff

(Curve, r = -£— .

9+1

fEllipse whose equation is r = a -n.

Ex. 9. \ 2 + cos0

{Involute of circle, 0 = 7-^—1 sec"1 -

V« a

i

Fifth. Examples of curves rolling on themselves.

Ex. 1. When the curve which rolls on itself is a circle, equation

r = a cos 0,
the traced curve is a cardioid, equation r = a(l + cos#).

Ex. 2. When it is the curve whose equation is

the equation of the traced curve is

B xn+l

(ft \n+i
008 -TT) •
n+l/

Ex. 3. When it is the involute of the circle, the traced curve is the spiral
of Archimedes.

THE THEOEY OF EOLLING CUEVES. 29

Ex. 4. "When it is a parabola, the focus traces the directrix, and the vertex
traces the cissoid.

Ex. 5. When it is the hyperbolic spiral, the traced curve is the tractory of
the circle.

Ex. 6. When it is the polar catenary, the equation of the traced curve is

a /2a • --i r

6 = / -- 1 — versm ' - .
V r a

Ex. 7. When it is the curve whose equation is

the equation of the traced curve is r = a(e6 — e~").

This paper commenced with an outline of the nature and history of the problem of rolling
curves, and it was shewn that the subject had been discussed previously, by several geometers,
amongst whom were De la Hire and Nicole in the Memoires de I'Academie, Euler, Professor
Willis, in his Principles of Mechanism, and the Rev. H. Holditch in the Cambridge Philosophical
Transactions.

None of these authors, however, except the two last, had made any application of their
methods ; and the principal object of the present communication was to find how far the general
equations could be simplified in particular cases, and to apply the results to practice.

Several problems were then worked out, of which some were applicable to the generation
of curves, and some to wheelwork ; while others were interesting as shewing the relations which
exist between different curves; and, finally, a collection of examples was added, as an illus-
tration of the fertility of the methods employed.

[From the Transactions of the Royal Society of Edinburgh, Vol. XX. Part I.]

III. — On the Equilibrium of Elastic Solids.

THERE are few parts of mechanics in which theory has differed more from
experiment than in the theory of elastic solids.

Mathematicians, setting out from very plausible assumptions with respect to
the constitution of bodies, and the laws of molecular action, came to conclusions
which were shewn to be erroneous by the observations of experimental philoso-
phers. The experiments of (Ersted proved to be at variance with the mathe-
matical theories of Navier, Poisson, and Lame and Clapeyron, and apparently
deprived this practically important branch of mechanics of all assistance from
mathematics.

The assumption on which these theories were founded may be stated thus : —

Solid bodies are composed of distinct molecules, which are kept at a certain
i/'.ttance from each other by the opposing principles of attraction and heat. When
the distance between two molecules is changed, they act on each other with a force
whose direction is in the line joining the centres of the molecules, and whose
magnitude is equal to the change of distance multiplied into a function of the
distance which vanishes when that distance becomes sensible.

The equations of elasticity deduced from this assumption contain only one
coefficient, which varies with the nature of the substance.

The insufficiency of one coefficient may be proved from the existence of
bodies of different degrees of solidity.

No effort is required to retain a liquid in any form, if its volume remain
unchanged ; but when the form of a solid is changed, a force is called into
action which tends to restore its former figure ; and this constitutes the differ-

THE EQUILIBRIUM OF ELASTIC SOLIDS. 31

ence between elastic solids and fluids. Both tend to recover their volume, but
fluids do not tend to recover their shape.

Now, since there are in nature bodies which are in every intermediate state
from perfect solidity to perfect liquidity, these two elastic powers cannot exist
in every body in the same proportion, and therefore all theories which assign to
them an invariable ratio must be erroneous.

I have therefore substituted for the assumption of Navier the following
axioms as the results of experiments.

If three pressures in three rectangular axes be applied at a point in an
elastic solid, —

1. The sum of the three pressures is proportional to the sum of the com-
pressions which they produce.

2. The difference between two of the pressures is proportional to the differ-
ence of the compressions which they produce.

The equations deduced from these axioms contain two coefficients, and differ
from those of Navier only in not assuming any invariable ratio between the
cubical and linear elasticity. They are the same as those obtained by Professor
Stokes from his equations of fluid motion, and they agree with all the laws of
elasticity which have been deduced from experiments.

In this paper 2)ressures are expressed by the number of units of weight to
the unit of surface ; if in English measure, in pounds to the square inch, or
in atmospheres of 15 pounds to the square inch.

Compression is the proportional change of any dimension of the solid caused
by pressure, and is expressed by the quotient of the change of dimension divided
by the dimension compressed "*.

Pressure will be understood to include tension, and compression dilatation ;
pressure and compression being reckoned positive.

Elasticity is the force which opposes pressure, and the equations of elasticity
are those which express the relation of pressure to compression f.

Of those who have treated of elastic solids, some have confined themselves
to the investigation of the laws of the bending and twisting of rods, without

* The laws of pressure and compression may be found in the Memoir of Lam<j and Clapeyrou. See
note A.

t See note B.

THE EQUILIBRIUM OF ELASTIC SOLIDS.

considering the relation of the coefficients which occur in these two cases;
while others have treated of the general problem of a solid body exposed to
any forces.

The investigations of Leibnitz, Bernoulli, Euler, Varignon, Young, La Hire,
and Lagrange, are confined to the equilibrium of bent rods; but those of
Navier, Poisson, Lanit? and Clapeyron, Cauchy, Stokes, and Wertheim, are
principally directed to the formation and application of the general equations.

The investigations of Navier are contained in the seventh volume of the
Memoirs of the Institute, page 373 ; and in the Annales de Chimie et de
PI' Unique, 2' S^rie, xv. 264, and xxxvm. 435 ; U Application de la Mecanique,
Tom. i.

Those of Poisson in M6m. de I'Institut, vni. 429 ; Annales de Chimie, 2."
Serie, xxxvr. 334; xxxvu. 337; xxxvrn, 338; XLII. Journal de I'Ecole
Polytechnique, cahier xx., with an abstract in Annales de Chimie for 1829.

The memoir of MM. Lamd and Clapeyron is contained in Crelle's Math<-
matical Journal. Vol. vn. ; and some observations on elasticity are to be found
in Lamp's Cours de Physique.

M. Cauchy's investigations are contained in his Exercices d? Analyse, Vol. m.
p. 180, published in 1828.

Instead of supposing each pressure proportional to the linear compression
which it produces, he supposes it to consist of two parts, one of which is pro-
portional to the linear compression in the direction of the pressure, while the
other is proportional to the diminution of volume. As this hypothesis admits
two coefficients, it differs from that of this paper only in the values of the
coefficients selected. They are denoted by K and k, and K=p. — ^m, k = m.

The theory of Professor Stokes is contained in Vol. vni. Part 3, of the
Cambridge Philosophical Transactions, and was read April 14, 1845.

He states his general principles thus : — " The capability which solids possess
of being put into a state of isochronous vibration, shews that the pressures
called into action by small displacements depend on homogeneous functions of
those displacements of one dimension. I shall suppose, moreover, according to
the general principle of the superposition of small quantities, that the pressures
due to different displacements are superimposed, and, consequently, that the
pressures are linear functions of the displacements."

THE EQUILIBEIUM OF ELASTIC SOLIDS. 33

Having assumed the proportionality of pressure to compression, he proceeds
to define his coefficients. — "Let -AS be the pressures corresponding to a uniform
linear dilatation 8 when the solid is in equilibrium, and suppose that it becomes
mAS, in consequence of the heat developed when the solid is in a state of rapid
vibration. Suppose, also, that a displacement of shifting parallel to the plane
xy, for which Sx = kx, Sy = -ky, and Sz = 0, calls into action a pressure -Bk
on a plane perpendicular to the axis of x, and a pressure Bk on a plane
perpendicular to the axis of y ; the pressure on these planes being equal and
of contrary signs; that on a plane perpendicular to z being zero, and the tan-
gential forces on those planes being zero." The coefficients A and B, thus

defined, when expressed as in this paper, are A = 3p., B =

m
~2 '

Professor Stokes does not enter into the solution of his equations, but gives
their results in some particular cases.

1. A body exposed to a uniform pressure on its whole surface.

2. A rod extended in the direction of its length.

3. A cylinder twisted by a statical couple.

He then points out the method of finding A and B from the last two cases.

While explaining why the equations of motion of the luminiferous ether are
the same as those of incompressible elastic solids, he has mentioned the property
of plasticity or the tendency which a constrained body has to relieve itself
from a state of constraint, by its molecules assuming new positions of equi-
librium. This property is opposed to linear elasticity ; and these two properties
exist in all bodies, but in variable ratio.

M. Wertheim, in Annales de Chimie, 3e Sdrie, XXIIL, has given the results
of some experiments on caoutchouc, from which he finds that K= k, or p. = f TO ;
and concludes that k = K in all substances. In his equations, /A is therefore
made equal to f m.

The accounts of experimental researches on the values of the coefficients
are so numerous that I can mention only a few.

Canton, Perkins, (Ersted, Aime, Colladon and Sturm, and Kegnault, have
determined the cubical compressibilities of substances ; Coulomb, Duleau, and
Giulio, have calculated the linear elasticity from the torsion of wires ; and a
great many observations have been made on the elongation and bending of beams.

VOL. I. 5

•'J4 THK EQUILIBRIUM OF ELASTIC SOLIDS.

I have found no account of any experiments on the relation between the
doubly refracting power communicated to glass and other elastic solids by com-
pression, and the pressure which produces it* ; but the phenomena of bent glass
seem to prove, that, in homogeneous singly-refracting substances exposed to
pressures, the principal axes of pre&sure coincide with the principal axes of
double refraction ; and that the difference of pressures in any two axes is
proportional to the difference of the velocities of the oppositely polarised rays
whose directions are parallel to the third axis. On this principle I have
calculated the phenomena seen by polarised light in the cases where the solid
is bounded by parallel planes.

In the following pages I have endeavoured to apply a theory identical
with that of Stokes to the solution of problems which have been selected on
account of the possibility of fulfilling the conditions. I have not attempted to
extend the theory to the case of imperfectly elastic bodies, or to the laws of
permanent bending and breaking. The solids here considered are supposed not
to be compressed beyond the limits of perfect elasticity.

The equations employed in the transformation of co-ordinates may be found
in Gregory's Solid Geometry.

I have denoted the displacements by &x, 8y, 8z. They are generally denoted
by a, ft, y ; but as I had employed these letters to denote the principal axes
at any point, and as this had been done throughout the paper, I did not alter
a notation which to me appears natural and intelligible.

The laws of elasticity express the relation between the changes of the
dimensions of a body and the forces which produce them.

These forces are called Pressures, and their effects Compressions. Pressures
are estimated in pounds on the square inch, and compressions in fractions of the
dimensions compressed.

Let the position of material points in space be expressed by their co-ordinates
x, y, and z, then any change in a system of such points is expressed by giving
to these co-ordinates the variations Sx, 8y, 8z, these variations being functions of
x, y, «•

* See note C.

THE EQUILIBRIUM OF ELASTIC SOLIDS. 35

The quantities 8x, Sy, Sz, represent the absolute motion of each point in
the directions of the three co-ordinates ; but as compression depends not on
absolute, but on relative displacement, we have to consider only the nine
quantities —

dSx
dx '

dSy

dSx
dy>

<%

dSx

dz '

dSy
~dz '

dSz

dx '
dSz

dy '
dSz

dx'

dy'

dz •

Since the number of these quantities is nine, if nine other independent
quantities of the same kind can be found, the one set may be found in terms
of the other. The quantities which we shall assume for this purpose are —

1. Three compressions, -, -jr, •— , in the directions of three principal
axes a, ft, y.

2. The nine direction-cosines of these axes, with the six connecting equa-
tions, leaving three independent quantities. (See Gregory's Solid Geometry.)

3. The small angles of rotation of this system of axes about the axes of
x, y, z.

The cosines of the angles which the axes of x, y, z make with those of
a, ft, y are

cos (aOx) — a, , cos (ftOx) = ^ , cos (yOx) = cl ,
cos (aOy) = aa, cos (ftOy) = bs, cos (yOy) = c2)
cos (aOz) =as, cos (/JOz) = ba, cos (yOz) = c3.

These direction-cosines are connected by the six equations,

t , ,3 2c3 = 0,

«32 + &/ + cs2 = 1, «,«, + &A + c3c, = 0.

The rotation of the system of axes a, ft, y, round the axis of

x, from y to z, =S#1(
y, from z to x, =&&.,,
z, from x to y, =80.,;

5—2

36

THE EQUILIBRIUM OF ELASTIC SOLIDS.

By resolving the displacements Sa, 8)8, Sy, 80!, 80,, 80,, in the directions
of the axes x, y, z, the displacements in these axes are found to be

Bx = «,8a + 6,8/J + c,8y — 80,z + 80,y ,
8y = a,8a + 6,8)3 + c,8y — 80,x + 80,2,
Bz = a,8a + 6,8)8 + c,Sy - S0,?/ + 80,*.

a B y

and a = a,z + a,y + a,z, B = b^c + bj/ + bf, and y = c,x + c^ + c,z.

Substituting these values of 8a, 8)8, and 8y in the expressions for Bx, Sy,
82, and differentiating with respect to x, y, and z, in each equation, we obtain
the equations

dBx _ Ba , 8/8 , , 8y ,
dx a 1 B 1 y '

dSy Sa 8/8 . . Sy
-,'/=--a1&'-c1

Sa

y

.+fe.

x Sa 8)8 7 7 Sy ,

= = a A + 6i&> + CA "

_ So, 8^8 Sy

~ " \JvtfJL* ^" ^»~ ^2^1 "T "' Q¥^\ """" ^

* • Aj »w

£ rt C*

op 7 7 oy
p y

= 8a a 8jtbb Sy

Sa
a

Sa
a

Sa

.(1).

Equations of
compression.

(2).

Equations of the eq»i'i;/,n'>im of an element of the solid.
The forces which may act on a particle of the solid are :

1. Three attractions in the direction of the axes, represented by X, Y, Z.

2. Six pressures on the six faces.

THE EQUILIBRIUM OF ELASTIC SOLIDS.

3. Two tangential actions on each face.

37

Let the six faces of the small parallelepiped be denoted by xlt yl} z1} £2, yt,
and z,, then the forces acting on. #, are : —

1. A normal pressure pt acting in the direction of x on the area dydz.

2. A tangential force q3 acting in the direction of y on the same area.

3. A tangential force qj acting in the direction of z on the same area,
and so on for the other five faces, thus : —

Forces which act in the direction of the axes of

x y z

On the face a;,

-p.dydz

— q3dydz

-qjdydz

x,

* • • • • • y\

dp,.

(q3 + -^ dx) dydx
—p^dzdx

in 1 _L -^2 s7rY\ diifl"

— qjdzdx

^" dx ' y

— q^dzdx

. * • ... 7/3

dq\, .,,

( p.2 -\- j dy) dzdx
dy

I n \ ti'Jti fi"siT

\ '/ a ~l~ 7 W1 V / tt'-wCvJ./

dy

\ (7 1 | 7 Us If t Cv^W/t*/

dy

... * * * \

— qfixdy

— qldxdy

-padxdy

* * * * • • ^2

fe+^4)4rf,

(q1 + -^- dz) dxdy

( n\ A, -^3 s?<7\ /Iv/lii

1 I/Q T^ j \JUtv 1 UUiAAjVtf

Attractions,

pXdxdydz

p Ydxdydz

pZdxdydz

Takmg the moments of these forces round the axes of the particle, we find

and then equating the forces in the directions of the three axes, and dividing
by dx, dy, dz, we find the equations of pressures,

+ + +
dx dy dz

p, + qi + q1+
dy dz dx ^

dpt dq dq,
~

_
-

Equations of Pressures.

,(3).

38

Till: Kt,'t ll.IBRIUM OF ELASTIC SOLIDS.

The resistance which the solid opposes to these pressures is called Elasticity,
aud is of two kinds, for it opposes either change of volume or change of Jvjim-.
These two kinds of elasticity have no necessary connection, for they are possessed
in very different ratios by different substances. Thus jelly has a cubical elas-
ticity little different from that of water, and a linear elasticity as small as we
please ; while corkt whose cubical elasticity is very small, has a much greater
linear elasticity than jelly.

Hooke discovered that the elastic forces are proportional to the changes
that excite them, or as he expressed it, " Ut tensio sic vis."

To fix our ideas, let us suppose the compressed body to be a parallelepiped,
and let pressures Plt Pt, Pt act on its faces in the direction of the axes
a> ft, y, which will become the principal axes of compression, and the com-

pressions

.„ , Sa S/S Sy
will be - - , -£ , -f .

The fundamental assumption from which the following equations are deduced
is an extension of Hooke's law, and consists of two parts.

I. The sum of the compressions is proportional to the sum of the pressures.

II. The difference of the compressions is proportional to the difference of
the pressures.

These laws are expressed by the following equations : —

(4).

II.

Equations of Elasticity.

(5).

The quantity p. is the coefficient of cubical elasticity, and m that of linear
elasticity.

THE EQUILIBEIUM OF ELASTIC SOLIDS.

By solving these equations, the values of the pressures P,, P,, P3, and the

s\ o 5?
compressions --, -J-, - may be found.

8/3

Sa

Sy

(6).

a

From these values of the pressures in the axes a, /3, y, may be obtained
the equations for the axes x, y, z, by resolutions of pressures and compressions""".

For

and

q = aaP, + lbP3 + ccP3 ;

dSz\ dSx

d8z

dS

d8x

dSx

dSz

_ m id8y dSz
=~"'~

m idSz dSx
^ \~dx~ ~

_ in idSx dSy\
=~

(8).

.(9).

* See the Memoir of Lame and Clapeyron, and note A.

40

THE EQUILIBRIUM OF ELASTIC SOLIDS.

1 1

d8z

(10).

-«-%*-±ft

dz

dx

dz

m

.(11).

By substituting in Equations (3) the values of the forces given in Equa-
tions (8) and (9), they become

d /<

z/J 2
d8z\ m/c?

d&x

These are the general equations of elasticity, and are identical with those
of M. Cauchy, in his Exercices d'Analyse, Vol. in., p. 180, published in 1828,

m

where k stands for m, and K for p — —, and those of Mr Stokes, given in the

2

Cambi-idge Philosophical Transactions, Vol. vin., part 3, and numbered (30) ;

in

in his equations A = 3/x, B = .

8

If the temperature is variable from one part to another of the elastic
solid, the compressions , , , - , -r- , at any point will be diminished by a

quantity proportional to the temperature at that point. This principle is applied
in Cases X. and XI. Equations (10) then become

THE EQUILIBRIUM OF ELASTIC SOLIDS.

dSx [ 1 1 \ , 1

41

dSx I 1 1 \ ,

3?" for Mi) (A

dSy / 1 1

-4 = fe ~ 3

(13),

c,v being the linear expansion for the temperature v.

Having found the general equations of the equilibrium of elastic solids, I
proceed to work some examples of their application, which afford the means of
determining the coefficients /*, m, and <a, and of calculating the stiffness of
solid figures. I begin with those cases in which the elastic solid is a hollow
cylinder exposed to given forces on the two concentric cylindric surfaces, and
the two parallel terminating planes. /

In these cases the co-ordinates x, y, z are replaced by the co-ordinates
x = x, measured along the axis of the cylinder.
y — r, the radius of any point, or the distance from the axis.
z = rO, the arc of a circle measured from a fixed plane passing
through the axis.

-j— , p1 = o, are the compression and pressure in the direction of the
axis at any point.

y-75

T—

radius.

-j—

y

~T = -T— , pt=p, are the compression and pressure in the direction of the

dSz
dz

7C> f\ CS

= — , pt = q, are the compression and pressure in the direction of the

tangent.
Equations (9) become, when expressed in terms of these co-ordinates-

m

m

.(14).

m d8x

The length of the cylinder is b, and the two radii a, and a, in every case.
VOL. i. • G

42 THE EQUILIBRIUM OF ELASTIC SOLIDS.

CASE I.

The first equation is applicable to the case of a hollow cylinder, of which
the outer surface is fixed, while the inner surface is made to turn through
a small angle 86, by a couple whose moment is M.

The twisting force M is resisted only by the elasticity of the solid, and
therefore the whole resistance, in every concentric cylindric surface, must be equal
to M.

The resistance at any point, multiplied into the radius at which it acts, is
expressed by

m

rq, = — T -
2 d'i

Therefore for the whole cylindric surface

—j— mirru = M.
dr

M (I l
= - — , — -- •

Whence 80

M fl i

and m = 7TTsa ~ -- i

2irbo6 Vet," a,8

The optical effect of the pressure of any point is expressed by

T i Mb

7=0,^ = 0,. — ............................ (15).

Tlierefore, if the soh'd be viewed by polarized light (transmitted parallel to
the axis), the difference of retardation of the oppositely polarized rays at any
point in the solid will be inversely proportional to the square of the distance from
the axis of the cylinder, and the planes of polarization of these rays will be
inclined 45° to the radius at that point.

The general appearance is therefore a system of coloured rings arranged
oppositely to the rings in uniaxal crystals, the tints ascending in the scale as
they approach the centre, and the distance between the rings decreasing towards
the centre. The whole system is crossed by two dark bands inclined 45° to the
plane of primitive polarization, when the plane of the analysing plate is perpen-
dicular to that of the first polarizing plate.

THE EQUILIBRIUM OF ELASTIC SOLIDS. 43

A jelly of isinglass poured when hot between two concentric cylinders forms,
when cold, a convenient solid for this experiment ; and the diameters of the rings
may be varied at pleasure by changing the force of torsion applied to the interior
cylinder.

By continuing the force of torsion while the jelly is allowed to dry, a hard
plate of isinglass is obtained, which still acts in the same way on polarized light,
even when the force of torsion is removed.

It seems that this action cannot be accounted for by supposing the interior
parts kept in a state of constraint by the exterior parts, as in unannealed and
heated glass ; for the optical properties of the plate of isinglass are such as
would indicate a strain preserving in every part of the plate the direction of
the original strain, so that the strain on one part of the plate cannot be main-
tained by an opposite strain on another part.

Two other uncrystallised substances have the power of retaining the polariz-
ing structure developed by compression. The first is a mixture of wax and resin
pressed into a thin plate between two plates of glass, as described by Sir David
Brewster, in the Philosophical Transactions for 1815 and 1830.

When a compressed plate of this substance is examined with polarized light,
it is observed to have no action on light at a perpendicular incidence ; but when
inclined, it shews the segments of coloured rings. This property does not belong
to the plate as a whole, but is possessed by every part of it. It is therefore
similar to a plate cut from a uniaxal crystal perpendicular to the axis.

I find that its action on light is like that of a positive crystal, while that
of a plate of isinglass similarly treated would be negative.

The other substance which possesses similar properties is gutta percha. This
substance in its ordinary state, when cold, is not transparent even in thin films;
but if a thin film be drawn out gradually, it may be extended to more than
double its length. It then possesses a powerful double refraction, which it
retains so strongly that it has been used for polarizing light*. As one of its
refractive indices is nearly the same as that of Canada balsam, while the other
is very different, the common surface of the gutta percha and Canada balsam
will transmit one set of rays much more readily than the other, so that a film
of extended gutta percha placed between two layers of Canada balsam acts like

* By Dr Wright, I believe.

6—2

44 THE EQUILIBRIUM OF ELASTIC SOLIDS.

a plate of nitre treated in the same way. That these films are in a state of
constraint may be proved by heating them slightly, when they recover their
original dimensions.

As all these permanently compressed substances have passed their limit of
perfect elasticity, they do not belong to the class of elastic solids treated of in
this paper ; and as I cannot explain the method by which an uncrystallised body
maintains itself in a state of constraint, I go on to the next case of twisting,
which has more practical importance than any other. This is the case of a
cylinder fixed at one end, and twisted at the other by a couple whose moment
is .1 / .

CASE II.

In this case let 80 be the angle of torsion at any point, then the resistance
to torsion in any circular section of the cylinder is equal to the twisting force M.

The resistance at any point in the circular section is given by the second
Equation of (14).

_m dS0
qt~~2r~fa-

This force acts at the distance r from the axis ; therefore its resistance to torsion
will be qtr, and the resistance in a circular annulus will be

_ , , ,

qtr2irrdr = mtn3 -?— dr

and the whole resistance for the hollow cylinder will be expressed by

,, ftnt dW ,

T 7E"fa *fl

88
720 M

m =

ir* n

In this equation, m is the coefficient of linear elasticity ; a, and o, are the
radii of the exterior and interior surfaces of the hollow cylinder in inches ; M is
the moment of torsion produced by a weight acting on a lever, and is expressed

THE EQUILIBRIUM OF ELASTIC SOLIDS. 45

by the product of the number of pounds in the weight into the number of inches
in the lever; 6 is the distance of two points on the cylinder whose angular
motion is measured by means of indices, or more accurately by small mirrors
attached to the cylinder ; n is the difference of the angle of rotation of the two
indices in degrees.

This is the most accurate method for the determination of m independently
of p, and it seems to answer best with thick cylinders which cannot be used
with the balance of torsion, as the oscillations are too short, and produce a
vibration of the whole apparatus.

CASE III.

/

A hollow cylinder exposed to normal pressures only. When the pressures
parallel to the axis, radius, and tangent are substituted for plt p2, and pt,
Equations (10) become

9/i 3m/ ' m

(20).

d(rd) " r \9/* Sm^
By multiplying Equation (20) by r, differentiating with respect to r, and

75

comparing this value of —j— with that of Equation (19),

p-q_f 1 _LW^> , dp + ^2\_I^2
rm " \9/x 3m/ \dr dr dr) m dr '

The equation of the equilibrium of an element of the solid is obtained by
considering the forces which act on it in the direction of the radius. By
equating the forces which press it outwards with those pressing it inwards, we
find the equation of the equilibrium of the element,

p_ ..(21).

dr

46 THE EQUILIBRIUM OF ELASTIC SOLIDS.

By comparing this equation with the last, we find

(L L\<b + (L AW^ + ^

\9/t 3m/ dr \9ft 3m/ \dr dr)
Integrating,

Since o, the longitudinal pressure, is supposed constant, we may assume

Cl ~

c,

i+'-i

9/x 37?i
Therefore q—p = ct — 2p, therefore by (21),

dr r r '
a linear equation, which gives

The coefficients c, and c, must be found from the conditions of the surface
of the solid. If the pressure on the exterior cylindric surface whose radius is a,
be denoted by hlt and that on the interior surface whose radius is aa by ^,,

then p = h1 when r = a,
and p=ht when r = aa
and the general value of p is

P= V'-a/ ' ^ a?11"? ^22)>

,'A, — a*h ct/a,' A., — j

.(24).

r* aa'-<!

This last equation gives the optical effect of the pressure at any point. The
law of the magnitude of this quantity is the inverse square of the radius, as in

THE EQUILIBRIUM OF ELASTIC SOLIDS. 47

Case I. ; but the direction of the principal axes is different, as in this case they
are parallel and perpendicular to the radius. The dark bands seen by polarized
light will therefore be parallel and perpendicular to the plane of polarization, in-
stead of being inclined at an angle of 45°, as in Case I.

By substituting hi Equations (18) and (20), the values of p and q given in
(22) and (23), we find that when r = aj,

o + 2
2

+2

?\ ~ n I • " Vi^l — "P*-! / i 5 \ t\

9ft 3m/ ' a? — a? \9ju, 3m/ .

.(25).

When r = als :- = f o + 2-^^p +Q , ,

r 9/t\ a.'-a.1 / 9m\ of -of / , , .

lzb;-

(i i \ i /o™ » /» s 4. Q« "X r»3 / 9

I 1 \ 7 1 / ^(.t-j (*! "T 01*,, \ 7 tt-q / &

__^_ _^ I I h ____________ I _ I _ I „_ rt _______^^__ I __^_ I

9/* 3m/ ' ttj2 — aa2 \ 9/x 3m / 3a,a — «2a \9^i 3m/ j

From these equations it appears that the longitudinal compression of cylin-
dric tubes is proportional to the longitudinal pressure referred to unit of surface
when the lateral pressures are constant, so that for a given pressure the com-
pression is inversely as the sectional area of the tube.

These equations may be simplified in the following cases : —

1. When the external and internal pressures are equal, or hl = hi.

2. When the external pressure is to the internal pressure as the square of
the interior diameter is to that of the exterior diameter, or when a^h1 = a^h.,.

3. When the cylinder is solid, or when aa = 0.

4. When the solid becomes an indefinitely extended plate with a cylindric
hole in it, or when aa becomes infinite.

5. When pressure is applied only at the plane surfaces of the solid cylinder,
and the cylindric surface is prevented from expanding by being inclosed in a

Sr
strong case, or when — = 0.

6. When pressure is applied to the cylindric surface, and the ends are

Sx

retained at an invariable distance, or when - — = 0.

oc

48 THE EQUILIBRIUM OF ELASTIC SOLIDS.

1. When A, = A,, the equations of compression become

"3m'°~ ''

(27).

When A, = A, = o, then

x r

The compression of a cylindrical vessel exposed on all sides to the same
hydrostatic pressure is therefore independent of m, and it may be shewn that
the same is true for a vessel of any shape.

2. When a^/t^a,1^,,

x
Sr

3m

(28).

In this case, when o = 0, the compressions are independent of /A.

3. In a solid cylinder, a, = 0,

p = q = h1.

S £

The expressions for ^ and - - are the same as those in the first case, when

X i

ht.

When the longitudinal pressure o vanishes,

x

.
3m'

THE EQUILIBRIUM OF ELASTIC SOLIDS.

When the cylinder is pressed on the plane sides only,

8x_ /_!_ 2 \
~~x = ° \9ft + 3m/ '

8r_ /_! J_\

r = ° \9/A 3m/ '

4. When the solid is infinite, or when aa is infinite,

7 -^- •> / 7 7 \

r 9/i

5. When 8r = 0 in a solid cylinder,

So; 3o

6. When

x 2m + 3/i

Sx 8r_
x~ ' r ~

49

(29).

.(30).

Since the expression for the effect of a longitudinal strain is

Sx
x

3m/ '

if we make

VOL. I.

„ 9ma ,1 8x 1
E=- •—- , then -=0-^,
m + 6/x 33 £>

(31).

7

50 IHE EQUILIBRIUM OF ELASTIC SOLIDS.

The quantity E may be deduced from experiment on the extension of wires
or rods of the substance, and \L is given in terms of m and E by the equation,

<32>>

(33),

P being the extending force, b the length of the rod, s the sectional area,
and Sx the elongation, which may be determined by the deflection of a wire,
as in the apparatus of S' Gravesande, or by direct measurement.

CASE IV.

The only known direct method of finding the compressibility of liquids is
that employed by Canton, (Ersted, Perkins, Aimd, &c.

The liquid is confined in a vessel with a narrow neck, then pressure is
applied, and the descent of the liquid in the tube is observed, so that the
difference between the change of volume of liquid and the change of internal
capacity of the vessel may be determined.

Now, since the substance of which the vessel is formed is compressible, a
change of the internal capacity is possible. If the pressure be applied only to
the contained liquid, it is evident that the vessel will be distended, and the
compressibility of the liquid will appear too great. The pressure, therefore, is
commonly applied externally and internally at the same time, by means of a
hydrostatic pressure produced by water compressed either in a strong vessel or
in the depths of the sea.

As it does not necessarily follow, from the equality of the external and
internal pressures, that the capacity does not change, the equilibrium of the
vessel must be determined theoretically. (Ersted, therefore, obtained from Poisson
his solution of the problem, and applied it to the case of a vessel of lead.
To find the cubical elasticity of lead, he applied the theory of Poisson to the
numerical results of Tredgold. As the compressibility of lead thus found was
greater than that of water, (Ersted expected that the apparent compressibility
of water in a lead vessel would be negative. On making the experiment the
apparent compressibility was greater in lead than in glass. The quantity found

THE EQUILIBRIUM OF ELASTIC SOLIDS. 51

by Tredgold from the extension of rods was that denoted by E, and the value
of IJL deduced from E alone by the formulae of Poisson cannot be true, unless

— = $•; and as - for lead is probably more than 3, the calculated compressi-
bility is much too great.

A similar experiment was made by Professor Forbes, who used a vessel of
caoutchouc. As in this case the apparent compressibility vanishes, it appears
that the cubical compressibility of caoutchouc is equal to that of water.

Some who reject the mathematical theories as unsatisfactory, have conjec-
tured that if the sides of the vessel be sufficiently thin, the pressure on both
sides being equal, the compressibility of the vessel will not affect the result.
The following calculations shew that the apparent compressibility of the liquid
depends on the compressibility of the vessel, and is independent of the thickness
when the pressures are equal.

A hollow sphere, whose external and internal radii are a, and «2, is acted
on by external and internal normal pressures h± and A2, it is required to deter-
mine the equilibrium of the elastic solid.

The pressures at any point in the solid are : —

1. A pressure p in the direction of the radius.

2. A pressure q in the perpendicular plane.

These pressures depend on the distance from the centre, which is denoted
by r.

The compressions at any point are —r- in the radial direction, and - in
the tangent plane, the values of these compressions are : —

_ -- p ...................... (34).

dr \9/i 3m/ v m *

L\(p + 2g) + -tf ............... (35).

3m/ ^ *' m *

Multiplying the last equation by r, differentiating with respect to r, and
equating the result with that of the first equation, we find

7—2

52 THE EQUILIBRIUM OP ELASTIC SOLIDS.

Since the forces which act on the particle in the direction of the radius
must balance one another, or

2qdrd0 +p (rd0y = lp + ^ drj (r + dr)*e,

therefore y"~P = <j ~f~ (3^).

Substituting this value of q— p in the preceding equation, and reducing,

rJ.n rlfi

therefore

Integrating,

But

and the equation becomes

^_

dr^ J r r

therefore p = ct -j + ^ .

Since p = h^ when r = al, and p — ht when r = att the value of p at any
distance is found to be

_ afh^afh, _ ofo* _V-A . .

o, ,-«„ «+i^gL i- « /38\

a^-a,' i* a,' -a,'

8F .. 8r a.'hi — afh. 1 _ 0,'ci,' ^, — A, 1

^_ Q ^_ * ^L J . j__. —

— ==- — O - - "^ • 8 g

Whenr=al; ^ = ^"^ I+f^'Az*-. i 1

r a. —a, u. a, —at m.

, (39).

~ a,1 — a,' \ /u, 2wi/ a,' — a,1 \ji 2m/

When the external and internal pressures are equal

, SF ^
hl = ht=p = q, and -=^.- = — v4(J/»

Jr /^

THE EQUILIBRIUM OF ELASTIC SOLIDS. 53

the change of internal capacity depends entirely on the cubical elasticity of the
vessel, and not on its thickness or linear elasticity.

When the external and internal pressures are inversely as the cubes of the
radii of the surfaces on which they act,

a3 , . a '

fmtli"u *l~ ~ 2 — r "i

SF _ a,3 At , ,

~T~' ~%^»m

— = -t-1

In this case the change of capacity depends on the linear elasticity alone.

M. Regnault, in his researches on the theory of the steam engine, has
given an account of the experiments which he made in / order to determine
with accuracy the compressibility of mercury.

He considers the mathematical formulas very uncertain, because the theories
of molecular forces from which they are deduced are probably far from the
truth ; and even were the equations free from error, there would be much
uncertainty in the ordinary method by measuring the elongation of a rod of
the substance, for it is difficult to ensure that the material of the rod is the
same as that of the hollow sphere.

He has, therefore, availed himself of the results of M. Lam4 for a hollow
sphere in three different cases, in the first of which the pressure acts on the
Ulterior and exterior surface at the same time, while in the other two cases
the pressure is applied to the exterior or interior surface alone. Equation (39)
becomes in these cases, —

5\ V 7i

1. When &i = hj, -^ = — and the compressibility of the enclosed liquid being

H1

fa, and the apparent diminution of volume S'F,

(42).

* v-i

2. When ^ = 0,

±} (43).

2W

54 THE EQUILIBRIUM OF ELASTIC SOLIDS.

3. When A, = 0,

o/\
f '

87 A, fa,' , .a,'

, , .,

17-= — -- » 1 — +f —
F '-' *

m

M. Lamd's equations differ from these only in assuming that p. = ^m. If
this assumption be correct, then the coefficients p., m, and fi,, may be found
from two of these equations ; but since one of these equations may be derived
from the other two, the three coefficients cannot be found when fi is supposed
independent of m. In Equations (39), the quantities which may be varied at
pleasure are A, and ht, and the quantities which may be deduced from the
apparent compressions are,

therefore some independent equation between these quantities must be found,
and this cannot be done by means of the sphere alone ; some other experiment
must be made on the liquid, or on another portion of the substance of which
the vessel is made.

The value of JA,, the elasticity of the liquid, may be previously known.

The linear elasticity m of the vessel may be found by twisting a rod of
the material of which it is made ;

Or, the value of E may be found by the elongation or bending of the

1 1 2
rod, and -£ = - + — .

We have here five quantities, which may be determined by experiment.

(43) 1. c,= (- + ^— j by external pressure

on sphere.

A 1\
(42) 2. c,= ( ---- J equal pressures.

\r" fa/

(31) 3. = + ) bv elongation of a rod.

(17) 4. m by twisting the rod.

5. /t, the elasticity of the liquid.

THE EQUILIBRIUM OF ELASTIC SOLIDS.

55

When the elastic sphere is solid, the internal radius a3 vanishes, and

, 8F h,
A, =p = q, and -== =

When the case becomes that of a spherical cavity in an infinite solid, the
external radius a^ becomes infinite, and

J.

m

SV

P-

m

(44).

The effect of pressure on the surface of a spherical cavity on any other part
of an elastic solid is therefore inversely proportional to the cube of its distance
from the centre of the cavity.

When one of the surfaces of an elastic hollow sphere has its radius rendered
invariable by the support of an incompressible sphere, whose radius is Oj, then

Sr

therefore

p=hti

3a,V

= 0, when r = av

+ A2-

q = i

8r

— = ft,
r

3a//i

a'a*

m

r3

a/ , afaf

—3 "2 — j- ;

Ain. 8F j, 3a,'- 3d?

Whenr = as, -fr=hin —T-

' V ' O/v 3m _1_ Q/-» 8<i

(45).

CASE V.

On the equilibrium of an elastic beam of rectangular section uniformly

bent.

By supposing the bent beam to be produced till it returns into itself, we
may treat it as a hollow cylinder.

56 THE EQUILIBRIUM OF ELASTIC SOLIDS.

Let a rectangular elastic beam, whose length is Zirc, be bent into a circular
form, so as to be a section of a hollow cylinder, those parts of the beam which
lie towards the centre of the circle will be longitudinally compressed, while the
opposite parts will be extended.

The expression for the tangential compression is therefore

oV r— c

Comparing this value of - • with that of Equation (20),
r — c

and by (21), q

wr

(1 2 \
— + — ], the equation

becomes

dp Zm + Spp 9w

— — -J- — ^Z

UT m + 6/x T
a linear differential equation, which gives

(7, may be found by assuming that when r=aiy p = hlt and q may be found
from p by equation (21).

As the expressions thus found are long and cumbrous, it is better to use
the following approximations : —

(48)-

In these expressions a is half the depth of the beam, and y is the distance
of any part of the beam from the neutral surface, which in this case is a cylin-
dric surface, whose radius is c.

These expressions suppose c to be large compared with a, since most sub-
stances break when -- exceeds a certain small quantity.

THE EQUILIBRIUM OF ELASTIC SOLIDS. 57

Let b be the breadth of the beam, then the force with which the beam
resists flexure = M.

which is the ordinary expression for the stiffness of a rectangular beam.

The stiffness of a beam of any section, the form of which is expressed by
an equation between x and y, the axis of x being perpendicular to the plane of
flexure, or the osculating plane of the axis of the beam at any point, is ex-
pressed by

Mc = E ly'dx ................................. (50),

M being the moment of the force which bends the beam, and c the radius of
the circle into which it is bent.

CASE VI. ,

At the meeting of the British Association in 1839, Mr James Nasmyth
described his method of making concave specula of silvered glass by bending.

A circular piece of silvered plate-glass was cemented to the opening of an
iron vessel, from which the air was afterwards exhausted. The mirror then
became concave, and the focal distance depended on the pressure of the air.

Buffon proposed to make burning-mirrors in this way, and to produce the
partial vacuum by the combustion of the air in the vessel, which was to be
effected by igniting sulphur in the interior of the vessel by means of a burn-
ing-glass. Although sulphur evidently would not answer for this purpose, phos-
phorus might; but the simplest way of removing the air is by means of the
air-pump. The mirrors which were actually made by Buffon, were bent by
means of a screw acting on the centre of the glass.

To find an expression for the curvature produced in a flat, circular, elastic
plate, by the difference of the hydrostatic pressures which act on each side
of it, —

Let t be the thickness of the plate, which must be small compared with
its diameter.

Let the form of the middle surface of the plate, after the curvature is
produced, be expressed by an equation between r, the distance of any point
from the axis, or normal to the centre of the plate, and x the distance of
the point from the plane in which the middle of the plate originally was, and let

VOL I.

8

58 THE EQUILIBRIUM OF ELASTIC 8OUDa

Let A, be the pressure on one side of the plate, and A, that on the other.

Let p and q be the pressures in the plane of the plate at any point, p
acting in the direction of a tangent to the section of the plate by a plane
passing through the axis, and q acting in the direction perpendicular to that
plane.

By equating the forces which act on any particle in a direction parallel to
the axis, we find

dr dx , dp dx d*x . . dr

By making p = Q when r = 0 in this equation, when integrated,

The forces perpendicular to the axis are

fdr\ dpdr d*r n ,. dx

Substituting for p its value, the equation gives

(A.-&.) (dr dr dx\
1 t \dsdx+ds)'i

The equations of elasticity become

'dr ds d*x ds d*r\ ,_v

r»-a- j.' A.- d&r d fir \ j • XT.-

Uinerentiatmg -^— = -5- ( — rl, and in this case

d8r _ dr dr

dr ds ds ds '

By a comparison of these values of — ,

.. _P +L?-2 + ^£4,-/ ' -1^* +

9p 3m/\P 2 )+m + dsm*r<) 3m/ dr + dr

ds

THE EQUILIBRIUM OF ELASTIC SOLIDS. 59

To obtain an expression for the curvature of the plate at the vertex, let a
be the radius of curvature, then, as an approximation to the equation of the
plate, let

= —

~'

By substituting the value of x in the values of p and q, and in the equa-
tion of elasticity, the approximate value of a is found to be

.9

3m/ m

t -18m/* 1 z - , .

"- '"

Since the focal distance of the mirror, or - , depends on the difference of

Zi

pressures, a telescope on Mr Nasmyth's principle would act as an aneroid baro-
meter, the focal distance varying inversely as the pressure of the atmosphere.

CASE VII.

To find the conditions of torsion of a cylinder composed of a great number
of parallel wires bound together without adhering to one another.

Let x be the length of the cylinder, a its radius, r the radius at any point,
80 the angle of torsion, M the force producing torsion, 8x the change of length,
and P the longitudinal force. Each of the wires becomes a helix whose radius
is r, its angular rotation 80, and its length along the axis x — Bx.

I ( O«;C

Its length is therefore / (rodY + y? ( 1 — '•

V v \ x.

{/ / O#*\ 2
1 — 1(1 )+i
V V •/

This force, resolved parallel to the axis, is

^^t/Ff^'l' :

8—2

60 THE EQUILIBRIUM OF ELASTIC SOLIDS.

t> CVrt

and since - - and r - - are small, we may assume

X X

The force, when resolved in the tangential direction, is approximately

1 d d ( SdSx rVStfYl

• -ja j- M= Er \r - -- — \ ,
rdQdr \ x x 2\a5/J

By eliminating - : between (54) and (55) we have

...................... (56).

2 x 24

When P = 0, -Jf depends on the sixth power of the radius and the cube
of the angle of torsion, when the cylinder is composed of separate filaments.

Since the force of torsion for a homogeneous cylinder depends on the
fourth power of the radius and the first power of the angle of torsion, the
torsion of a wire having a fibrous texture will depend on both these laws.

The parts of the force of torsion which depend on these two laws may be
found by experiment, and thus the difference of the elasticities in the direction
of the axis and in the perpendicular directions may be determined.

A calculation of the force of torsion, on this supposition, may be found in
Young's Mathematical Principles of Natural Philosophy; and it is introduced
here to account for the variations from the law of Case II., which may be
observed in a twisted rod.

CASE VIII.

It is well known that grindstones and fly-wheels are often broken by the
centrifugal force produced by their rapid rotation. I have therefore calculated
the strains and pressure acting on an elastic cylinder revolving round its axis,
and acted on by the centrifugal force alone.

THE EQUILIBRIUM OF ELASTIC SOLIDS.

61

The equation of the equilibrium of a particle [see Equation (21)], becomes

dp

q—p = r -r —
dr

"

^

where q and p are the tangential and radial pressures, k is the weight in
pounds of a cubic inch of the substance, g is twice the height in inches that
a body falls in a second, t is the time of revolution of the cylinder in seconds.

By substituting the value of q and -^ in Equations (19), (20), and neglect-

ng o,

which gives

j^P.
m V* dr

d"p

r

-4 + -
l

3

(57).

If the radii of the surfaces of the hollow cylinder be ai and c
pressures acting on them h,, and A2, then the values of ct and c2 are

E\

m

and the

(58).

When a, = 0, as in the case of a solid cylinder, cx = 0, and

When A, = 0, and r =

(60).

When 5 exceeds the tenacity of the substance hi pounds per square inch,
the cylinder will give way ; and by making q equal to the number of pounds
which a square inch of the substance will support, the velocity may be found
at which the bursting of the cylinder will take place.

62 THE EQUILIBRIUM OF ELASTIC SOLIDS.

Since 7= but (q — p) = — — ( -- 2j&r*, a transparent revolving cylinder, when

polarized light is transmitted parallel to the axis, will exhibit rings whose
diameters are as the square roots of an arithmetical progression, and brushes
parallel and perpendicular to the plane of polarization.

CASE IX.

A hollow cylinder or tube is surrounded by a medium of a constant
temperature while a liquid of a different temperature is made to flow through
it. The exterior and interior surfaces are thus kept each at a constant tem-
perature till the transference of heat through the cylinder becomes uniform.

Let v be the temperature at any point, then when this quantity has
reached its limit ,

rdv

(61).

Let the temperatures at the surfaces be 0, and 0,, and the radii of the
surfaces a, and c^, then

,' logOj-loga,

Let the coefficient of linear dilatation of the substance be c,, then the
proportional dilatation at any point will be expressed by ctv, and the equations
of elasticity (18), (19), (20), become

dSx I 1 1

d8r
8r

The equation of equilibrium is

and since the tube is supposed to be of a considerable length

d8x

-j— *=ct a constant quantity.

»-*+••£ <«).

THE EQUILIBRIUM OP ELASTIC SOLIDS.

From these equations we find that

,_fJ__ J_\ (zp+r&}

\9fji 3m/ \ * dr]

63

._

3m

and hence v = c,logr + ct, p may be found in terms of r.

Hence

Since

v

1

/= 60, - = &o, + -

2 1 \
+ — J

(

the rings seen in this case will differ from those described' in Case III. only
by the addition of a constant quantity.

When no pressures act on the exterior and interior surfaces of the tube
^=5^ = 0, and

2 . _1_\ "
+3m/

'

ofo* log «. - log a, a.* log a. - a,8 log aA
-- --

,/2 J^X-1
6fe 3m/ ^"

a1ia>Moga1-logai\

a.'-a/

.
'

...(62).

There will, therefore, be no action on polarized light for the ring whose
radius is r when

a,2a,2 , a,
r» = 2- —Jog-.
«i — a/ & a,

CASE X.

Sir David Brewster has observed (Edinburgh Transactions, Vol. vm.), that
when a soHd cyUnder of glass is suddenly heated at the cylindric surface a
polarizing force is developed, which is at any point proportional to the square
of the distance from the axis of the cylinder; that is to say, that the dif-

64 THE EQUILIBRIUM OF ELASTIC SOLIDS.

ference of retardation of the oppositely polarized rays of light is proportional
to the square of the radius r, or

/= "bcjtav* = l)oi (q —p) = boir -£- ,

dp

.-. = Cr .: =

Since if a be the radius of the cylinder, p = 0 when r = a,

P-f (<-*).
Hence q=^- (Sr'-a').

a

By substituting these values of p and q in equations (19) and (20), and
d 8r d8r , ,, ,

r = Ifind'

c4 being the temperature of the axis of the cylinder, and c, the coefficient of
linear expansion for glass.

CASE XI.

Heat is passing uniformly through the sides of a spherical vessel, such as
the ball of a thermometer, it is required to determine the mechanical state of
the sphere. As the methods a¥e nearly the same as in Case IX., it will be
sufficient to give the results, using the same notation.

j dv ci

r-j-sCj, .'. v = ct — -,
dr r

When &,=A, = 0 the expression for p becomes
/2 IN"1 /a a. ( a,'a,,' 1 0,0, 1 , a^-a,' 1 lc.\

7)=(— - + - C,(0,-0,H i -- — H«i««? - TT^ —Kf ...... (64)'

\9/i 3w/ i; \a,' - a,' r* a, - a» r («,-«*) (a,1 - a/) J

From this value of p the other quantities may be found, as in Case IX.,
from the equations of Case IV.

THE EQUILIBRIUM OF ELASTIC SOLIDS. 65

CASE XII.

When a long beam is bent into the form of a closed circular ring (as in
Case V.), all the pressures act either parallel or perpendicular to the direction
of the length of the beam, so that if the beam were divided into planks, there
would be no tendency of the planks to slide on one another.

But when the beam does not form a closed circle, the planks into which it
may be supposed to be divided will have a tendency to slide on one another,
and the amount of sliding is determined by the linear elasticity of the sub-
stance. The deflection of the beam thus arises partly from the bending of the
whole beam, and partly from the sliding of the planks ; and since each of these
deflections is small compared with the length of the beam/ the total deflection
will be the sum of the deflections due to bending and sliding.

Let A = Mc = E\xy*dy (65).

A is the strfihess of the beam as found in Case V., the equation of the
transverse section being expressed in terms of x and y, y being measured from
the neutral surface.

Let a horizontal beam, whose length is 2l, and whose weight is 2w, be
supported at the extremities and loaded at the middle with a weight W.

Let the deflection at any point be expressed by ty, and let this quantity
be small compared with the length of the beam.

At the middle of the beam, S,?/ is found by the usual methods to be

8$ = -. (-faVw + %l3 W) (66).

A

Let B= — \xdy = -fr (sectional area) (67).

2 J 2i

B is the resistance of the beam to the sliding of the planks. The de-
flection of the beam arising from this cause is

cs /i T/T7A / fi Q ^

VOL. I. 9

66 THE EQUILIBRIUM OF ELASTIC SOLIDS.

This quantity is small compared with 8,y, when the depth of the beam is
small compared with its length.

The whole deflection &y = S,y + Sy

(69).

CASE XIII.

Wlien the values of the compressions at any point have been found, when
two different seta of forces act on a solid separately, the compressions, when
the forces act at the same time, may be found by the composition of com-
pressions, because the small compressions are independent of one another.

It appears from Case L, that if a cylinder be twisted as there described,
the compressions will be inversely proportional to the square of the distance
from the centre.

If two cylindric surfaces, whose axes are perpendicular to the plane of an
indefinite elastic plate, be equally twisted in the same direction, the resultant
compression in any direction may be found by adding the compression due to
each resolved in that direction.

The result of this operation may be thus stated geometrically. Let Al and
At (Fig. 1) be the centres of the twisted cylinders. Join A^At, and bisect AtAt
in O. Draw OBC at right angles, and cut off OBl and OBa each equal to OA,.

Then the difference of the retardation of oppositely polarized rays of light
passing perpendicularly through any point of the plane varies directly as the
product of its distances from Bl and B3, and inversely as the square of the
product of its distances from Al and At.

The isochromatic lines are represented in the figure.

The retardation is infinite at the points .4, and At ; it vanishes at £,
and B, ; and if the retardation at O be taken for unity, the isochromatic curves
2, 4, surround A1 and As; that in which the retardation is unity has two
loops, and passes through 0; the curves £, £ are continuous, and have points
of contrary flexure ; the curve £ has multiple points at Cl and (7,, where

THE EQUILIBRIUM OF ELASTIC SOLIDS.

67

AlG^ = A-LAt, and two loops surrounding B^ and B2; the other curves, for which
-^=lV> ^r, &c., consist each of two ovals surrounding Br and Bz, and an
exterior portion surrounding all the former curves.

Fig. 1.

I have produced these curves in the jelly of isinglass described in Case I.
They are best seen by using circularly polarized light, as the curves are then
seen without interruption, and their resemblance to the calculated curves is
more apparent. To avoid crowding the curves toward the centre of the figure,
I have taken the values of / for the different curves, not in an arithmetical,
but in a geometrical progression, ascending by powers of 2.

9—2

68

THE EQUILIBRIUM OF ELASTIC SOLIDS.

CASE XIV.

On the determination of the pressures which act in the interior of trans-
parent solids, from observations of the action of the solid on polarized light.

Sir David Brewster has pointed out the method by which polarized light
might be made to indicate the strains in elastic solids ; and his experiments on
bent glass confirm the theories of the bending of beams.

The phenomena of heated and unannealed glass are of a much more complex
nature, and they cannot be predicted and explained without a knowledge of the
laws of cooling and solidification, combined with those of elastic equilibrium.

In Case X. I have given an example of the inverse problem, in the case
of a cylinder in which the action on light followed a simple law ; and I now
go on to describe the method of determining the pressures in a general case,
applying it to the case of a triangle of unannealed plate-glass.

D n

The lines of equal intensity of the action on light are seen without
interruption, by using circularly polarized light. They are represented in Fig. 2,
where A, BBS, DDD are the neutral points, or points of no action on light,
and CCC, EEE are the points where that action is greatest ; and the intensity

THE EQUILIBRIUM OF ELASTIC SOLIDS. 69

of the action at any other point is determined by its position with respect to
the isochromatic curves.

The direction of the principal axes of pressure at any point is found by
transmitting plane polarized light, and analysing it in the plane perpendicular
to that of polarization. The light is then restored in every part of the triangle,
except in those points at which one of the principal axes is parallel to the
plane of polarization. A dark band formed of all these points is seen, which
shifts its position as the triangle is turned round in its own plane. Fig. 3
represents these curves for every fifteenth degree of inclination. They correspond
to the lines of equal variation of the needle in a magnetic chart.

From these curves others may be found which shall indicate, by their own
direction, the direction of the principal axes at any point. These curves of
direction of compression and dilatation are represented in Fig. 4 ; the curves
whose direction corresponds to that of compression are concave toward the
centre of the triangle, and intersect at right angles the curves of dilatation.

Let the isochromatic lines in Fig. 2 be determined by the equation

k(x,y) = I- = «>(q-p)-,

where / is the difference of retardation of the oppositely polarized rays, and
q and p the pressures in the principal axes at any point, z being the thick-
ness of the plate.

Let the lines of equal inclination be determined by the equation

0 bemg the angle of inclination of the principal axes ; then the differential
equation of the curves of direction of compression and dilatation (Fig. 4) is

By considering any particle of the plate as a portion of a cylinder whose
axis passes through the centre of curvature of the curve of compression, we find

' ' ' - =

70 THE EQUILIBRIUM OF ELASTIC SOLIDS.

Let R denote the radius of curvature of the curve of compression at any
point, and let S denote the length of the curve of dilatation at the same
point,

and since (q —p), R and S are known, and since at the surface, where <£, (a;, y) = 0,
p = 0, all the data are given for determining the absolute value of p by inte-
gration.

Though this is the best method of finding p and q by graphic construc-
tion, it is much better, when the equations of the curves have been found, that
is, when <£, and <£, are known, to resolve the pressures in the direction of the
axes.

The new quantities are plf pt, and qt ; and the equations are

Pi~ Pi

It is therefore possible to find the pressures from the curves of equal tint
and equal inclination, in any case in which it may be required. In the mean-
time the curves of Figs. 2, 3, 4 shew the correctness of Sir John Herschell's
ingenious explanation of the phenomena of heated and unannealed glass.

NOTE A.

As the mathematical laws of compressions and pressures have been very thoroughly
investigated, and as they are demonstrated with great elegance in the very complete and
elaborate memoir of MM. Lame" and Clapeyron, I shall state as briefly as possible their results.

Let a solid be subjected to compressions or pressures of any kind, then, if through any
point in the solid lines be drawn whose lengths, measured from the given point, are pro-
portional to the compression or pressure at the point resolved in the directions in which the
lines are drawn, the extremities of such lines will be in the surface of an ellipsoid, whose
centre is the given point.

The properties of the system of compressions or pressures may be deduced from those
of the ellipsoid.

THE EQUILIBRIUM OP ELASTIC SOLIDS. 71

There are three diameters having perpendicular ordinates, which are called the principal
axes of the ellipsoid.

Similarly, there are always three directions in the compressed particle in which there
is no tangential action, or tendency of the parts to slide on one another. These directions
are called the principal axes of compression or of pressure, and in homogeneous solids they
always coincide with each other.

The compression or pressure in any other direction is equal to the sum of the products
of the compressions or pressures in the principal axes multiplied into the squares of the
cosines of the angles which they respectively make with that direction.

NOTE B.

The fundamental equations of this paper differ from those of Navier, Poisson, &c., only
in not assuming an invariable ratio between the linear and the cubical elasticity; but since
I have not attempted to deduce them from the laws of molecular action, some other reasons
must be given for adopting them.

The experiments from which the laws are deduced are —

1st. Elastic solids put into motion vibrate isochronously, so that the sound does not
vary with the amplitude of the vibrations.

2nd. Regnault's experiments on hollow spheres shew that both linear and cubic com-
pressions are proportional to the pressures.

3rd. Experiments on the elongation of rods and tubes immersed in water, prove that
the elongation, the decrease of diameter, and the increase of volume, are proportional to the
tension.

4th. In Coulomb's balance of torsion, the angles of torsion are proportional to the
twisting forces.

It would appear from these experiments, that compressions are always proportional to
pressures.

Professor Stokes has expressed this by making one of his coefficients depend on the
cubical elasticity, while the other is deduced from the displacement of shifting produced by
a given tangential force.

M. Cauchy makes one coefficient depend on the linear compression produced by a force
acting in one direction, and the other on the change of volume produced by the same force.

Both of these methods lead to a correct result ; but the coefficients of Stokes seem to
have more of a real signification than those of Cauchy; I have therefore adopted those of
Stokes, using the symbols m and p, and the fundamental equations (4) and (5), which define
them.

7'J THE EQUILIBRIUM OF ELASTIC SOLIDS,

NOTE C.

As the coefficient o>, which determines the optical effect of .pressure on a substance,
varies from one substance to another, and is probably a function of the linear elasticity, a
determination of its value in different substances might lead to some explanation of tli<-
action of media on light

This paper commenced by pointing out the insufficiency of all theories of elastic solids,
in which the equations do not contain two independent constants deduced from experiments.
One of these constants is common to liquids and solids, and is called the modulus of cubical
elasticity. The other is peculiar to solids, and is here called the modulus of linear elasticity.
The equations of Navier, Poisson, and Lame* and Clapeyron, contain only one coefficient ;
and Professor G. G. Stokes of Cambridge, seems to have formed the first theory of elastic
solids which recognised the independence of cubical and linear elasticity, although M. Cauchy
seems to have suggested a modification of the old theories, which made the ratio of linear
to cubical elasticity the same for all substances. Piofessor Stokes has deduced the theory
of elastic solids from that of the motion of fluids, and his equations are identical with those
of this paper, which are deduced from the two following assumptions.

In an element of an elastic solid, acted on by three pressures at right angles to one
another, as long as the compressions do not pass the limits of perfect elasticity —

1st. The sum of the pressures, in three rectangular axes, is proportional to the sum
of the compressions in those axes.

2nd. The difference of the pressures in two axes at right angles to one another, is
proportional to the difference of the compressions in those axes.

Or, in symbols:

..

2.

p being the modulus of cubical, and m that of linear elasticity.

These equations are found to be very convenient for the solution of problems, some
of which were given in the latter part of the paper.

THE EQUILIBRIUM OF ELASTIC SOLIDS. 73

These particular cases were —

That of an elastic hollow cylinder, the exterior surface of which was fixed, while the
interior was turned through a small angle. The action of a transparent solid thus twisted
on polarized light, was calculated, and the calculation confirmed by experiment.

The second case related to the torsion of cylindric rods, and a method was given by

Q-Wiw

which m may be found. The quantity E= —T~^~ was found by elongating, or by bending
the rod used to determine TO, and p. is found by the equation,

Em

/* =

9m -6E'

The effect of pressure on the surfaces of a hollow sphere or cylinder was calculated,
and the result applied to the determination of the cubical compressibility of liquids and
solids.

An expression was found for the curvature of an elastic plate exposed to pressure on
one side ; and the state of cylinders acted on by centrifugal fqrce and by heat was
determined.

The principle of the superposition of compressions and pressures was applied to the case of
a bent beam, and a formula was given to determine E from the deflection of a beam
supported at both ends and loaded at the middle.

The paper concluded with a conjecture, that as the quantity to (which expresses the
relation of the inequality of pressure in a solid to the doubly-refracting force produced) is
probably a function of m, the determination of these quantities for different substances
might lead to a more complete theory of double refraction, and extend our knowledge of the
laws of optics.

VOL. I. 10

[Extracted from the Cambridge and Dublin Mathematical Journal, Vol. vili. p. 188,

February, 1854.]

Solutions of Problems.

1. If from a point in the circumference of a vertical circle two heavy particles be suc-
cessively projected along the curve, their initial velocities being equal and either in the same
or in opposite directions, the subsequent motion will be such that a straight line joining
the particles at any instant will touch a circle.

Note. The particles are supposed not to interfere with each other's motion.

THE direct analytical proof would involve the properties of elliptic integrals,
but it may be made to depend upon the following geometrical theorems.

(1) If from a point in one of two circles a right line be drawn cutting
the other, the rectangle contained by the segments so formed is double of the
rectangle contained by a line drawn from the point perpendicular to the rmln-nl
axis of the two circles, and the line joining their centres.

The radical axis is the line joining the points of intersection of the two
circles. It is always a real line, whether the points of intersection of the circles
be real or imaginary, and it has the geometrical property — that if from any point
on the radical axis, straight lines be drawn cutting the circles, the rectangle con-
tained by the segments formed by one of the circles is equal to the rectangle
contained by the segments formed by the other.

The analytical proof of these propositions is very simple, and may be resorted
to if a geometrical proof does not suggest itself as soon as the requisite figure
is constructed.

If A, B be the centres of the circles, P the given point in the circle whose
centre is A, a line drawn from P cuts the first circle in p, the second in Q

SOLUTIONS OP PBOBLEMS. 75

and q, and the radical axis in R. If PH be drawn perpendicular to the radical
axis, then

PQ.Pq = 2AB.HP.

COR. If the line be drawn from P to touch the circle in T, instead of
cutting it in Q and q, then the square of the tangent PT is equal to the
rectangle 2AB . HP.

Similarly, if ph be drawn from p perpendicular to the radical axis

pF = 2AB.hp.

Hence, if a line be drawn touching one circle in T, and cutting the other
in P and p, then

(PT)2 : (pT? :: HP : hp.

(2) If two straight lines touching one circle and cutting another be made
to approach each other indefinitely, the small arcs intercepted by their inter-
sections with the second circle will be ultimately proportional to their distances
from the point of contact.

This result may easily be deduced from the properties of the similar
triangles FTP and p'pT.

COR. If particles P, p be constrained to move in the circle A, while
the line Pp joining them continually touches the circle B, then the velocity
of P at any instant is to that of p as PT to pT; and conversely, if the
velocity of P at any instant be to that of P as PT to pT, then the line
Pp will continue to be a tangent to the circle B.

Now let the plane of the circles be vertical and the radical axis horizontal,
and let gravity act on the particles P, p. The particles were projected from
the same point with the same velocity. Let this velocity be that due to the
depth of the point of projection below the radical axis. Then the square of
the velocity at any other point will be proportional to the perpendicular from
that point on the radical axis ; or, by the corollary to (l), if P and p be at
any time at the extremities of the line PTp, the square of the velocity of P
will be to the square of the velocity of p as PH to ph, that is, as (PT)3 to
(pT)\ Hence, the velocities of P and p are in the proportion of PT to pT,
and therefore, by the corollary to (2), the line joining them will continue a
tangent to the circle B during each instant, and will therefore remain a tangent
during the motion.

10—2

76 SOLUTIONS OF PROBLEMS.

The circle A, the radical axis, and one position of the line Pp, are given
by the circumstances of projection of P and p. From these data it is easy to
determine the circle B by a geometrical construction.

It is evident that the character of the motion will determine the position
of the circle B. If the motion is oscillatory, B will intersect A. If P and p
make complete revolutions in the same direction, B will lie entirely within A,
but if they move in opposite directions, B will lie entirely above the radical axis.

If any number of such particles be projected from the same point at equal
intervals of time with the same direction and velocity, the lines joining successive
particles at any instant will be tangents to the same circle ; and if the time
of a complete revolution, or oscillation, contain n of these intervals, then these
lines will form a polygon of n sides, and as this is true at any instant, any
number of such polygons may be formed.

Hence, the following geometrical theorem is true :

" If two circles be such that n lines can be drawn touching one of them
and having their successive intersections, including that of the last and first,
on the circumference of the other, the construction of such a system of lines
will be possible, at whatever point of the first circle we draw the first tangent."

2. A transparent medium is such that the path of a ray of light within it is a given
circle, the index of refraction being a function of the distance from a given point in the
plane of the circle.

Find the form of this function and shew that for light of the same refrangibility —

(1) The path of every ray within the medium is a circle.

(2) All the rays proceeding from any point in the medium will meet accurately in
another point.

(3) If rays diverge from a point without the medium and enter it through a spherical
surface having that point for its centre, they will be made to converge accurately to a point
within the medium.

LEMMA I. Let a transparent medium be so constituted, that the refractive
index is the same at the same distance from a fixed point, then the path of
any ray of light within the medium will be in one plane, and the perpen-

SOLUTIONS OF PROBLEMS. 77

dicular from the fixed point on the tangent to the path of the ray at any
point will vary inversely as the refractive index of the' medium at that point.

We may easily prove that when a ray of light passes through a spherical
surface, separating a medium whose refractive index is ^ from another where
it is p.2, the plane of incidence and refraction passes through the centre of
the sphere, and the perpendiculars on the direction of the ray before and after
refraction are in the ratio of p, to p.r Since this is true of any number of
spherical shells of different refractive powers, it is also true when the index of
refraction varies continuously from one shell to another, and therefore the
proposition is true.

LEMMA II. If from any fixed point in the plane of a circle, a perpen-
dicular be drawn to the tangent at any point of the circumference, the rectangle
contained by this perpendicular and the diameter of the circle is equal to the
square of the line joining the point of contact with the fixed point, together
with the rectangle contained by the segments of any chord through the fixed
point.

Let APS be the circle, 0 the fixed point ; then

OY.PR=OP*+AO.OB.

Produce PO to- Q, and join QR, then the triangles 0 YP, PQR are similar ;

therefore

OY.PR=OP.PQ

.: OY.PR=OP* + AO.
If we put in this expression AO . 0-8 = a2,

P0 = r, OY=p,
it becomes 2pp = r^ + a\

7^ SOLUTIONS OF PROBLEMS.

To find the law of the index of refraction of the medium, so that a ray
from A may describe the circle APB, /* must be made to vary inversely as p
by Lemma I.

Let AO = r,, and let the refractive index at A=p^; then generally

_C _ 2Cp
^^"oM^5

2Cp
but at A /*! = •

therefore P- = PL\ ~TT 3 •

a +?

The value of p at any point is therefore independent of p, the radius of
the given circle ; so that the same law of refractive index will cause any other
ray to describe another circle, for which the value of a' is the same. The

value of OB is — , which is also independent of p ; so that every ray which
proceeds from A must pass through B.

Again, if we assume /A, as the value of p. when r = 0,

therefore = .

a result independent of r,. This shews that any point A' may be taken as
the origin of the ray instead of A, and that the path of the ray will still be
circular, and will pass through another point R on the other side of 0, such that

OR= a

Next, let CP be a ray from C, a point without the medium, falling at P
on a spherical surface whose centre is C.

Let 0 be the fixed point in the medium as before. Join PO, and produce

a*
to Q till OQ = 7jp- Through Q draw a circle touching CP in P, and cutting

CO in A and B ; then PBQ is the path of the ray within the medium.

SOLUTIONS OF PROBLEMS. 79

Since CP touches the circle, we have

CP*=CA.CB,

= (CO-OA)(CO+OB);

but OA = -Q£ ;

therefore CP2 = CO1 + CO I OB - ^) - a2,

\ /

an equation whence OB may be found, B being the point in the medium
through which all rays from C pass.

NOTE. The possibility of the existence of a medium of this kind possessing
remarkable optical properties, was suggested by the contemplation of the structure
of the crystalline lens in fish ; and the method of searching for these properties
was deduced by analogy from Newton's Principia, Lib. I. Prop. vn.

It would require a more accurate investigation into the law of the refractive
index of the different coats of the lens to test its agreement with the supposed
medium, which is an optical instrument theoretically perfect for homogeneous
light, and might be made achromatic by proper adaptation of the dispersive
power of each coat.

On the other hand, we find that the law of the index of refraction which
would give a minimum of aberration for a sphere of this kind placed in water,
gives results not discordant with facts, so far as they can be readily ascertained.

[From the Transactions of the Cambridge Philosophical Society, VoL IX. Part IV.]

IV. On the Transformation of Surfaces by Bending.

EUCLID has given two definitions of a surface, which may be taken as
examples of the two methods of investigating their properties.

That in the first book of the Elements is—

" A superficies is that which has only length and breadth."

The superficies differs from a line in having breadth as well as length,
and the conception of a third dimension is excluded without being explicitly
introduced.

In the eleventh book, where the definition of a solid is first formally
given, the definition of the superficies is made to depend on that of the solid—
" That which bounds a solid is a superficies."

Here the conception of three dimensions in space is employed in forming
a definition more perfect than that belonging to plane Geometry.

In our analytical treatises on geometry a surface is defined by a function
of three independent variables equated to zero. The surface is therefore the
boundary between the portion of space in which the value of the function is
positive, and that in which it is negative ; so that we may now define a
surface to be the boundary of any assigned portion of space.

Surfaces are thus considered rather with reference to the figures which they
limit than as having any properties belonging to themselves.

But the conception of a surface which we most readily form is that of
a portion of matter, extended in length and breadth, but of which the thick-

TRANSFORMATION OF SURFACES BY BENDING. 81

ness may be neglected. By excluding the thickness altogether, we arrive at
Euclid's first definition, which we may state thus —

"A surface is a lamina of which the thickness is diminished so as to become
evanescent."

We are thus enabled to consider a surface by itself, without reference to
the portion of space of which it is a boundary. By drawing figures on the
surface, and investigating their properties, we might construct a system of
theorems, which would be true independently of the position of the surface in
space, and which might remain the same even when the form of the solid of
which it is the boundary is changed.

When the properties of a surface with respect to space are changed, while
the relations of lines and figures in the surface itself are unaltered, the surface
may be said to preserve its identity, so that we may consider it, after the
change has taken place, as the same surface.

When a thin material lamina is made to assume a new form it is said
to be bent. In certain cases this process of bending is called development, and
when one surface is bent so as to coincide with another it is said to be
applied to it.

By considering the lamina as deprived of rigidity, elasticity, and other
mechanical properties, and neglecting the thickness, we arrive at a mathemati-
cal definition of this kind of transformation.

" The operation of bending is a continuous change of the form of a surface,
without extension or contraction of any part of it."

The following investigations were undertaken with the hope of obtaining
more definite conceptions of the nature of such transformations by the aid of
those geometrical methods which appear most suitable to each particular case.
The order of arrangement is that in which the different parts of the subject
presented themselves at first for examination, and the methods employed form
parts of the original plan, but much assistance in other matters has been
derived from the works of Gauss *, Liouvillef, BertrandJ, Puiseux§, &c., references
to which will be given in the course of the investigation.

* Disquititiones generates circa superficies curvas. Presented to the Royal Society of Gottingen,
8th October, 1827. Commentationes Recentiores, Tom. vi.

f Liouville's Journal, xn. J Ibid. xnr. § Ibid.

VOL. I.

82 TRANSFORMATION OF SURFACES BY BENDING.

I.

On the Bending of Surfaces generated by the motion of a strain/lit line in space.

If a straight line can be drawn in any surface, we may suppose that
part of the surface which is on one side of the straight line to be fixed,
while the other part is turned about the straight line as an axis.

In this way the surface may be bent about any number of generating lines
as axes successively, till the form of every part of the surface is altered.

The mathematical conditions of this kind of bending may be obtained in
the following manner.

Let the equations of the generating line be expressed so that the constants
involved in them are functions of one independent variable u, by the variation of
which we pass from one position of the line to another.

If in the equations of the generating line A a, u = ul, then in the equations
of the line Bb we may put u = ui, and from the equations of these lines we
may find by the common methods the equations of the shortest line PQ between
Aa and Bb, and its length, which we may call 8£. We may also find the
angle between the directions of A a and Bb, and let this angle be 80.

In the same way from the equations of
Cc, in which u = us, we may deduce the equa-
tio'ns of RS, the shortest line between Bb and
Cc, its length S£,> and the angle 80.; between
the directions of Bb and Cc. We may also
find the value of QR, the distance between
the points at which PQ and RS cut Bb.
Let QR = 8cr, and let the angle between the
directions of PQ and RS be 8$.

Now suppose the part of the surface between the lines Aa and Bb to be
fixed, while the part between Bb and Cc is turned round Bb as an axis. The
line RS will then revolve round the point R, remaining perpendicular to Bl>,
and Cc will still be at the same distance from Bb, and will make the same
angle with it. Hence of the four quantities S&,, S03, So- and 8<f>, 8<f> alone will
be changed by the process of bending. S<£, however, may be varied in a
perfectly arbitrary manner, and may even be made to vanish.

TRANSFORMATION OF SURFACES BY BENDING. 83

For, PQ and RS being both perpendicular to Bb, RS may be turned
about Bb till it is parallel to PQ, in which case S<£ becomes = 0.

By repeating this process, we may make all the " shortest lines" parallel to
one another, and then all the generating lines will be parallel to the same
plane.

We have hitherto considered generating lines situated at finite distances from
one another ; but what we have proved will be equally true when their distances
are indefinitely diminished. Then in the limit

becomes -j- ,

Ut — tt,

W d0_

ut — tt, du '

Scr dcr

u — M du '

d(f>

du '

All these quantities being functions of u, £, 0, cr and <}>, are functions of u
and of each other ; and if the forms of these functions be known, the positions
of all the generating lines may be successively determined, and the equation
to the surface may be found by integrating the equations containing the values
of £, 0, a- and <f>.

When the surface is bent in any manner about the generating lines, £, 0,
and cr remain unaltered, but (j> is changed at every point.

The form of <f> as a function of u will depend on the nature of the
bending ; but since this is perfectly arbitrary, <£ may be any arbitrary function
of u. In this way we may find the form of any surface produced by bending
the given surface along its generating lines.

By making <j> = 0, we make all the generating lines parallel to the same
plane. Let this plane be that of xy, and let the first generating line coincide
with the axis of x, then £ will be the height of any other generating line
above the plane of xy, and 0 the angle which its projection on that plane
makes with the axis of x. The ultimate intersections of the projections of the
generating lines on the plane of xy will form a curve, whose length, measured
from the axis of x, will be cr.

11—2

84 TRANSFORMATION OF SURFACES BY BENDING.

Since in this case the quantities £, 6, and cr are represented by distinct
geometrical quantities, we may simplify the consideration of all surfaces generated
by straight lines by reducing them by bending to the case in which those lines
are parallel to a given plane.

In the class of surfaces in which the generating lines ultimately intersect,

-1^ = 0, and £ constant. If these surfaces be bent so that <f> = Q, the whole of

' f '/

the generating lines will lie in one plane, and their ultimate intersections will
form a plane curve. The surface is thus reduced to one plane, and therefore
belongs to the class usually described as "developable surfaces." The form of a
developable surface may be defined by means of the three quantities 6, a- and
^. The generating lines form by their ultimate intersections a curve of double
curvature to which they are all tangents. This curve has been called tl it-
cuspidal edge. The length of this curve is represented by cr, its absolute

curvature at any point by • - , and its torsion at the same point by -7- .

When the surface is developed, the cuspidal edge becomes a plane curve,
and every part of the surface coincides with the plane. But it does not follow
that every part of the plane is capable of being bent into the original form
of the surface. This may be easily seen by considering the surface when the
position of the cuspidal edge nearly coincides with the plane curve but is not
confounded with it. It is evident that if from any point in space a tangent
can be drawn to the cuspidal edge, a sheet of the surface passes through that
point. Hence the number of sheets which pass through one point is the same
as the number of tangents to the cuspidal edge which pass through that
point ; and since the same is true in the limit, the number of sheets which
coincide at any point of the plane is the same as the number of tangents
which can be drawn from that point to the plane curve. In constructing a
developable surface of paper, we must remove those parts of the sheet from
which no real tangents can be drawn, and provide additional sheets where more
than one tangent can be drawn.

In the case of developable surfaces we see the importance of attending to
the position of the lines of bending; for though all developable surfaces may
be produced from the same plane surface, their distinguishing properties depend
on the form of the plane curve which determines the lines of bending.

TRANSFORMATION OF SURFACES BY BENDING.

85

II.

On the Betiding of Surfaces of Revolution.

In the cases previously considered, the bending in one part of the surface
may take place independently of that in any other part. In the case now
before us the bending must be simultaneous over the whole surface, and its
nature must be investigated by a different method.

The position of any point P on a surface of revolution may be deter-
mined by the distance PV from the vertex, measured
along a generating line, and the angle AVO which
the plane of the generating line makes with a fixed
plane through the axis. Let PV=s and AVO = 0.
Let r be the distance (Pp) of P from the axis; r
will be a function of s depending on the form of the
generating curve.

Now consider the small rectangular element of the surface at P. Its length
PIZ = 8s, and its breadth PQ = rS0, where r is a function of s.

If in another surface of revolution r is some other function of s, then the
length and breadth of the new element will be 8s and r'86', and if

r' = fir, and ff = - 6,
P-

and the dimensions of the two elements will be the same.

Hence the one element may be applied to the other, and the one surface
may be applied to the other surface, element to element, by bending it. To
effect this, the surface must be divided by cutting it along one of the generating
lines, and the parts opened out, or made to overlap, according as p, is greater
or less than unity.

To find the effect of this transformation on the form of the surface we
must find the equation to the original form of the generating line in terms of
s and r, then putting r' = p.r, the equation between s and r' will give the form
of the generating line after bending.

86 TRANSFORMATION OF SURFACES HY l.KM'IMJ.

When p. is greater than 1 it may happen that for some values of *, -j- is

creater than . In this case
P-

dr <!>• .

= /i-y-; is greater than 1 ;

a result which indicates that the curve becomes impossible for such values of
s and p..

The transformation is therefore impossible for the corresponding part of
the surface. If, however, that portion of the original surface be removed, the
remainder may be subjected to the required transformation.

The theory of bending when applied to the case of surfaces of revolution
presents no geometrical difficulty, and little variety ; but when we pass to
the consideration of surfaces of a more general kind, we discover the insufficiency
of the methods hitherto employed, by the vagueness of our ideas with respect
to the nature of bending in such cases. In the former case the bending is
of one kind only, and depends on the variation of one variable ; but the
surfaces we have now to consider may be bent in an infinite variety of ways,
depending on the variation of three variables, of which we do not yet know the
nature or interdependence.

We have therefore to discover some method sufficiently general to be appli-
cable to every possible case, and yet so definite as to limit each particular case
to one kind of bending easily understood.

The method adopted in the following investigations is deduced from the
consideration of the surface as the limit of the inscribed polyhedron, when the
size of the sides is indefinitely diminished, and their number indefinitely increased.

A method is then described by which such a polyhedron may be inscribed
in any surface so that all the sides shall be triangles, and all the solid angles
composed of six plane angles.

The problem of the bending of such a polyhedron is a question of trigo-
nometry, and equations might be found connecting the angles of the different
edges which meet in each solid angle of the polyhedron. It will be shewn that

TRANSFORMATION OF SURFACES BY BENDING. 87

the conditions thus obtained would be equivalent to three equations between
the six angles of the edges belonging to each solid angle. Hence three addi-
tional conditions would be necessary to determine the value of every such angle,
and the problem would remain as indefinite as before. But if by any means
we can reduce the number of edges meeting in a point to four, only one con-
dition would be necessary to determine them all, and the problem would be
reduced to the consideration of one kind of bending only.

This may be done by drawing the polyhedron in such a manner that the
planes of adjacent triangles coincide two and two, and form quadrilateral facets,
four of which meet in every solid angle. The bending of such a polyhedron
can take place only in one way, by the increase of the angles of two of the
edges which meet in a point, and the diminution of the angles of the other two.

The condition of such a polyhedron being inscribed in any surface is then
found, and it is shewn that when two forms of the same surface are given,
a perfectly definite rule may be given by which two corresponding polyhedrons
of this kind may be inscribed, one in each surface.

Since the kind of bending completely defines the nature of the quadrilateral
polyhedron which must be described, the lines formed by the edges of the
quadrilateral may be taken as an indication of the kind of bending performed
on the surface.

These lines are therefore defined as " Lines of Bending."

When the lines of bending are given, the forms of the quadrilateral facets
are completely determined ; and if we know the angle which any two adjacent
facets make with one another, we may determine the angles of the three edges
which meet it at one of its extremities. From each of these we may find the
angles of three other edges, and so on, so that the form of the polyhedron
after bending will be completely determined when the angle of one edge is given.
The bending is thus made to depend on the change of one variable only.

In this way the angle of any edge may be calculated from that of any
given edge ; but since this may be done in two different ways, by passing
along two different sets of edges, we must have the condition that these results
may be consistent with each other. This condition is satisfied by the method
of inscribing the polyhedron. Another condition will be necessary that the
change of the angle of any edge due to a small change of the given angle,
produced by bending, may be the same by both calculations. This is the con-
dition of ' Instantaneous Lines of Bending." That this condition may continue

88 TRANM«>I;MATION OF SURFACES BY LKSLINO.

to be satisfied during the whole process we must have another, which is the
condition for " Permanent Lines of Bending."

The use of these lines of bending in simplifying the theory of surfaces is
the only part of the present method which is new, although the investigations
connected with them naturally led to the employment of other methods which
had been used by those who have already treated of this subject. A state-
ment of the principal methods and results of these mathematicians will save
repetition, and will indicate the different points of view under which the
subject may present itself.

The first and most complete memoir on the subject is that of M. Gauss,
already referred to.

The method which he employs consists in referring every point of the
.surface to a corresponding point of a sphere whose radius is unity. Normals
are drawn at the several points of the surface toward the same side of it,
then lines drawn through the centre of the sphere in the direction of each of
these normals intersect the surface of the sphere in points corresponding to
those points of the original surface at which the normals were drawn.

If any line be drawn on the surface, each of its points will have a
corresponding point on the sphere, so that there will be a corresponding line
on the sphere.

If the line on the surface return into itself, so as to enclose a finite area
of the surface, the corresponding curve on the sphere will enclose an area on
the sphere, the extent of which will depend on the form of the surface.

This area on the sphere has been defined by M. Gauss as the measure of
the " entire curvature " of the area on the surface. This mathematical quantity
is of great use in the theory of surfaces, for it is the only quantity connected
with curvature which is capable of being expressed as the sum of all its parts.

The sum of the entire curvatures of any number of areas is the entire
curvature of their sum, and the entire curvature of any area depends on the
form of its boundary only, and is not altered by any change in the form of
the surface within the boundary line.

The curvature of the surface may even be discontinuous, so that we may
speak of the entire curvature of a portion of a polyhedron, and calculate its
amount.

If the dimensions of the closed curve be diminished so that it may be
treated as an element of the surface, the ultimate ratio of the entire curvature

TRANSFORMATION OF SURFACES BY BENDING. 89

to the area of the element on the surface is taken as the measure of the
" specific curvature " at that point of the surface.

The terms " entire " and " specific " curvature when used in this paper are
adopted from M. Gauss, although the use of the sphere and the areas on its
surface formed an essential part of the original design. The use of these terms
will save much explanation, and supersede several very cumbrous expressions.

M. Gauss then proceeds to find several analytical expressions for the measure
of specific curvature at any point of a surface, by the consideration of three
points very near each other.

The co-ordinates adopted are first rectangular, x and y, or x, y and z, being
regarded as independent variables.

Then the points on the surface are referred to two systems of curves drawn
on the surface, and their position is defined by the values of two independent
variables p and q, such that by varying p while q remains /constant, we obtain
the different points of a line of the first system, while p constant and q
variable defines a line of the second system.

By means of these variables, points on the surface may be referred to lines
on the surface itself instead of arbitrary co-ordinates, and the measure of cur-
vature may be found in terms of p and q when the surface is known.

In this way it is shewn that the specific curvature at any point is the
reciprocal of the product of the principal radii of curvature at that point, a
result of great interest.

From the condition of bending, that the length of any element of the
curve must not be altered, it is shewn that the specific curvature at any point
is not altered by bending.

The rest of the memoir is occupied with the consideration of particular
modes of describing the two systems of lines. One case is when the lines of
the first system are geodesic, or " shortest " lines having their origin in a point,
and the second system is drawn so as to cut off equal lengths from the curves
of the first system.

The angle which the tangent at the origin of a line of the first system
makes with a fixed line is taken as one of the co-ordinates, and the distance
of the point measured along that line as the other.

It is shewn that the two systems intersect at right angles, and a simple
expression is found for the specific curvature at any point.

M. Liouville (Journal, Tom. xu.) has adopted a different mode of simpli-

VOL. I. 12

90 TRANSFORMATION OF SURFACES BY BENDING.

tying the problem. He has shewn that on every surface it is possible to find
two systems of curves intersecting at right angles, such that the length and
breadth of every element into which the surface is thus divided shall be equal,
und that an infinite number of such systems may be found. By means of these
curves he has found a much simpler expression for the specific curvature than
that given by M. Gauss.

He has also given, in a note to his edition of Monge, a method of testing
two given surfaces in order to determine whether they are applicable to one
another. He first draws on both surfaces lines of equal specific curvature, and
determines the distance between two corresponding consecutive lines of curvature
in both surfaces.

If by assuming the origin properly these distances can be made equal for
every part of the surface, the two surfaces can be applied to each other. He
has developed the theorem analytically, of which this is only the geometrical
interpretation.

When the lines of equal specific curvature are equidistant throughout their
whole length, as in the case of surfaces of revolution, the surfaces may be
applied to one another in an infinite variety of ways.

When the specific curvature at every point of the surface is positive and
equal to a', the surface may be applied to a sphere of radius a, and when the
specific curvature is negative = — a" it may be applied to the surface of revo-
lution which cuts at right angles all the spheres of radius a, and whose centres
are in a straight line.

M. Bertrand has given in the XHIth Vol. of Liouville's Journal a very
simple and elegant proof of the theorem of M. Gauss about the product of
the radii of curvature.

He supposes one extremity of an inextensible thread to be fixed at a point
in a surface, and a closed curve to be described on the surface by the other
extremity, the thread being stretched all the while. It is evident that the
length of such a curve cannot be altered by bending the surface. He then
calculates the length of this curve, considering the length of the thread small,
and finds that it depends on the product of the principal radii of curvature
of the surface at the fixed point. His memoir is followed by a note of
M. Diguet, who deduces the same result from a consideration of the area of
the same curve ; and by an independent memoir of M. Puiseux, who seems to
</ive the same proof at somewhat greater length.

TRANSFORMATION OF SURFACES BY BENDING. 91

NOTE. Since this paper was written, I have seen the Rev. Professor Jellett's Memoir, On
the Properties of Inextensible Surfaces. It is to be found in the Transactions of the Royal Irish
Academy, Vol. XXII. Science, &c., and was read May 23, 1853.

Professor Jellett has obtained a system of three partial differential equations which express
the conditions to which the displacements of a continuous inextensible membrane are subject.
From these he has deduced the two theorems of Gauss, relating to the invariability of the product
of the radii of curvature at any point, and of the " entire curvature" of a finite portion of the
surface.

He has then applied his method to the consideration of cases in which the flexibility of the
surface is limited by certain conditions, and he has obtained the following results : —

If the displacements of an inextensible surface be all parallel to the same plane, the surface
moves as a rigid body,

Or, more generally,

If the movement of an inextensible surface, parallel to any one line, be,' that of a rigid body, the
entire movement is that of a rigid body.

The following theorems relate to the case in which a curve traced on the surface is rendered
rigid :—

// any curve be traced upon an inextensible surface whose principal radii of curvature are finite
and of the same sign, and if this curve be rendered immoveable, the entire surface will become
immoveable also.

In a developable surface composed of an inextensible membrane, any one of its rectilinear
sections may be fixed without destroying the flexibility of the membrane.

In convexo-concave surfaces, there are two directions passing through every point of the
surface, such that the curvature of a normal section taken in these directions vanishes. We
may therefore conceive the entire surface to be crossed by two series of curves, such that
a tangent drawn to either of them at any point shall coincide with one of these direc-
tions. These curves Professor Jellett has denominated Curves of Flexure, from the following
properties : —

Any curve of flexure may be fixed without destroying the flexibility of the surface.

If an arc of a curve traced upon an inextensible surface be rendered fixed or rigid, the entire of
the quadrilateral, formed by drawing the two curves of flexure through each extremity of the curve,
becomes fixed or rigid also.

Professor Jellett has also investigated the properties of partially inextensible surfaces, and
of thin material laminae whose extensibility is small, and in a note he has demonstrated the
following theorem : —

If a closed oval surface be perfectly inextensible, it is also perfectly rigid.

A demonstration of one of Professor Jellett's theorems will be found at the end of this paper.

J. C. M.
Aug. 30, 1854.

12—2

92

TRANSFORMATION OF SURFACES BY BENDING.

ON THE PROPERTIES OF A SURFACE CONSIDERED AS THE LIMIT OF THE INSCRIBED

POLYHEDRON.

1. To inscribe a polyhedron in a given surface, all whose sides shatt be
triangles, and all whose solid angles shall be hexahedral.

On the given surface describe a series of curves
according to any assumed law. Describe a second series
intersecting these in any manner, so as to divide the
whole surface into quadrilaterals. Lastly, describe a
third series (the dotted lines in the figure), so as to
pass through all the intersections of the first and second
series, forming the diagonals of the quadrilaterals.

The surface is now covered with a network of curvilinear triangles. The
plane triangles which have the same angular points will form a polyhedron
fulfilling the required conditions. By increasing the number of the curves in
each series, and diminishing their distance, we may make the polyhedron
approximate to the surface without limit. At the same time the polygons
formed by the edges of the polyhedron will approximate to the three systems
of intersecting curves.

2. To find the ineasure of the " entire curvature " of a solid angle of the
polyhedron, and of a finite portion of its surface.

From the centre of a sphere whose radius is unity draw perpendiculars to
the planes of the six sides forming the solid angle. These lines will meet the
surface in six points on the same side of the centre, which being joined by
arcs of great circles will form a hexagon on the surface of the sphere.

The area of this hexagon represents the entire curvature of the solid angle.

It is plain by spherical geometry that the angles of this hexagon are the
supplements of the six plane angles which form the solid angle, and that the
arcs forming the sides are the supplements of those subtended by the angles
of the six edges formed by adjacent sides.

The area of the hexagon is equal to the excess of the sum of its angles
above eight right angles, or to the defect of the sum of the six plane angles
from four right angles, which is the same thing. Since these angles are

TRANSFORMATION OF SURFACES BY BENDING. 93

invariable, the bending of the polyhedron cannot alter the measure of curvature
of each of its solid angles.

If perpendiculars be drawn to the sides of the polyhedron which contain
other solid angles, additional points on the sphere will be found, and if these
be joined by arcs of great circles, a network of hexagons will be formed on
the sphere, each of which corresponds to a solid angle of the polyhedron and
represents its " entire curvature."

The entire curvature of any assigned portion of the polyhedron is the sum
of the entire curvatures of the solid angles it contains. It is therefore repre-
sented by a polygon on the sphere, which is composed of all the hexagons
corresponding to its solid angles.

If a polygon composed of the edges of the polyhedron be taken as the
boundary of the assigned portion, the sum of its exterior angles will be the
same as the sum of the exterior angles of the polygon oil the sphere ; but
the area of a spherical polygon is equal to the defect of the sum of its
exterior angles from four right angles, and this is the measure of entire curva-
ture.

Therefore the entire curvature of the portion of the polyhedron enclosed
by the polygon is equal to the defect of the sum of its exterior angles from
four right angles.

Since the entire curvature of each solid angle is unaltered by bending,
that of a finite portion of the surface must be also invariable.

3. On the " Conic of Contact," and its use in determining the curvature
of normal sections of a surface.

Suppose the plane of one of the triangular facets of the polyhedron to
be produced till it cuts the surface. The form of the curve of intersection
will depend on the nature of the surface, and when the size of the triangle
is indefinitely diminished, it will approximate to the form of a conic section.

For we may suppose a surface of the second order constructed so as to
have a contact of the second order with the given surface at a point within
the angular points of the triangle. The curve of intersection with this surface
will be the conic section to which the other curve of intersection approaches.
This curve will be henceforth called the " Conic of Contact," for want of a better
name.

'.'I

TRANSFORMATION OF SURFACES BY BKMHN'.

/ //in/ (hi- ifidt'uft <)/' atnftture of a normal section
<>f the surface.

Let ARa be the conic of contact, C its centre, and
< '/' perpendicular to its plane. rPR a normal section, and
0 its centre of curvature, then

PIP

PO -
~

Clf
= \-fTp in the limit, when CR and P7? coincide.

r/?
-t cp>

or calling CP the "sagitta," we have this theorem :

"The radius of curvature of a normal section is equal to the square of
the corresponding diameter of the conic of contact divided by eight times the
sagitta."

4. To inscribe a polyhedron in a given surface, all wJiose side* */<«// be
/ilnnc (quadrilaterals, and all whose solid angles shall be tctmln-ilral.

Suppose the three systems of curves drawn as described in sect. (1), then
each of the quadrilaterals formed by the intersection of the first and second
systems is divided into two triangles by the third system. If the planes of
these two triangles coincide, they form a plane quadrilateral, and if every such
pair of triangles coincide, the polyhedron will satisfy the required condition.

Let abc be one of these triangles, and acd the
other, which is to be in the same plane with abc.
Then if the plane of abc be produced to meet the
surface in the conic of contact, the curve will pass
through abc and d. Hence abed must be a quad-
rilateral inscribed in the conic of contact.

But since ab and dc belong to the same system of curves, they will be
ultimately parallel when the size of the facets is diminished, and for a similar
reason, ad and be will be ultimately parallel. Hence abed will become a paral-
lelogram, but the sides of a parallelogram inscribed in a conic are parallel to
conjugate diameters.

TRANSFORMATION OF SURFACES BY BENDING. 95

Therefore the directions of two curves of the first and second system at
their point of intersection must be parallel to two conjugate diameters of the
conic of contact at that point in order that such a polyhedron may be inscribed.

Systems of curves intersecting in this manner will be referred to as "conju-
gate systems."

5. On the elementally conditions of the applicability of two surfaces.

It is evident, that if one surface is capable of being applied to another by
bending, every point, line, or angle in the first has its corresponding point, line,
or angle in the second.

If the transformation of the surface be effected without the extension or
contraction of any part, no line drawn on the surface can experience any change
in its length, and if this condition be fulfilled, there can be no extension or
contraction.

Therefore the condition of bending is, that if any line /whatever be drawn
on the first surface, the corresponding curve on the second surface is equal to it
in length. All other conditions of bending may be deduced from this.

6. If two curves on the first surface intersect, the corresponding curves on the
second surface intersect at the same angle.

On the first surface draw any curve, so as to form a triangle with the
curves already drawn, and let the sides of this triangle be indefinitely dimin-
ished, by making the new curve approach to the intersection of the former
curves. Let the same thing be done on the second surface. We shall then
have two corresponding triangles whose sides are equal each to each, by (5),
and since their sides are indefinitely small, we may regard them as straight
lines. Therefore by Euclid I. 8, the angle of the first triangle formed by the
intersection of the two curves is equal to the corresponding angle of the second.

7. At any given point of the first surface, two directions can be found, which
are conjugate to each other with respect to the conic of contact at that point, and
continue to be conjugate to each other when the first surface is transformed into the
second.

For let the first surface be transferred, without changing its form, to a
position such that the given point coincides with the corresponding point of the
second surface, and the normal to the first surface coincides with that of the

96

TRANSFORMATION OF SURFACES BY BENDING.

second at the same point. Then let the first surface be turned about the normal
as an axis till the tangent of any line through the point coincides with the
tangent of the corresponding line in the second surface.

Then by (6) any pair of corresponding lines passing through the point will
have a common tangent, and will therefore coincide in direction at that point.

If we now draw the conies of contact belonging to each surface we shall
have two conies with the same centre, and the problem is to determine a pair
of conjugate diameters of the first which coincide with a pair of conjugate
diameters of the second. The analytical solution gives two directions, real,
coincident, or impossible, for the diameters required.

In our investigations we can be concerned only with the case in which these
directions are real.

When the conies intersect in four points, P, Q, R, S, PQRS is a parallelo-
gram inscribed in both coriics, and the axes CA, CB,
parallel to the sides, are conjugate in both conies.

If the conies do not intersect, describe, through any
point P of the second conic, a conic- similar to and con-
centric with the first. If the conies intersect in four
points, we must proceed as before; if they touch in two
points, the diameter through those points and its conju-
gate must be taken. If they intersect in two points only,
then the problem is impossible ; and if they coincide
altogether, the conies are similar and similarly situated,
and the problem is indeterminate.

8. Two surfaces being given as before, one pair of conjugate systems of
curves may be drawn on the first surface, which shall correspond to a pair of
conjugate systems on the second surface.

By article (7) we may find at every point of the first surface two
directions conjugate to one another, corresponding to two conjugate directions on
the second surface. These directions indicate the directions of the two systems
of curves which pass through that point.

Knowing the direction which every curve of each system must have at every
point of its course, the systems of curves may be either drawn by some direct
geometrical method, or constructed from their equations, which may be found by
solving their differential equations.

TRANSFORMATION OF SURFACES BY BENDING. 97

Two systems of curves being drawn on the first surface, the corresponding
systems may be drawn on the second surface. These systems being conjugate
to each other, fulfil the condition of Art. (4), and may therefore be made the
means of constructing a polyhedron with quadrilateral facets, by the bending of
which the transformation may be effected.

These systems of curves will be referred to as the "first and second systems
of Lines of Bending."

9. General considerations applicable to Lines of Bending.

It has been shewn that when two forms of a surface are given, one of
which may be transformed into the other by bending, the nature of the lines
of bending is completely determined. Supposing the problem reduced to its
analytical expression, the equations of these curves would appear under the
form of double solutions of differential equations of the fir^t order and second
degree, each of which would involve one arbitrary quantity, by the variation of
which we should pass from one curve to another of the same system.

Hence the position of any curve of either system depends on the value
assumed for the arbitrary constant ; to distinguish the systems, let us call one
the first system, and the other the second, and let all quantities relating to
the second system be denoted by accented letters.

Let the arbitrary constants introduced by integration be u for the first
system, and u' for the second.

Then the value of u will determine the position of a curve of the first
system, and that of u a curve of the second system, and therefore u and u' will
suffice to determine the point of intersection of these two curves.

Hence we may conceive the position of any point on the surface to be
determined by the values of u and u for the curves of the two systems which
intersect at that point.

By taking into account the equation to the surface, we may suppose x, y,
and z the co-ordinates of any point, to be determined as functions of the two
variables u and u. This being done, we shall have materials for calculating
everything connected with the surface, and its lines of bending. But before
entering on such calculations let us examine the principal properties of these lines
which we must take into account.

Suppose a series of values to be given to u and u, and the corresponding
curves to be drawn on the surface.

VOL. i. 13

98 TRANSFORMATION OF SURFACES BY BENDING.

The surface will then be covered with a system of quadrilaterals, the size
of which may be diminished indefinitely by interpolating values of u and u
between those already assumed; and in the limit each quadrilateral may be
regarded as a parallelogram coinciding with a facet of the inscribed polyhedron.

The length, the breadth, and the angle of these parallelograms will vary at
different parts of the surface, and will therefore depend on the values of «
and u'.

The curvature of a line drawn on a surface may be investigated by consider-
ing the curvature of two other lines depending on it.

The first is the projection of the line on a tangent plane to the surface at
a given point in the line. The curvature of the projection at the point of
contact may be called the tangential curvature of the line on the surface. It
has also been called the geodesic curvature, because it is the measure of its
deviation from a geodesic or shortest line on the surface.

The other projection necessary to define the curvature of a line on the
surface is on a plane passing through the tangent to the curve and the normal
to the surface at the point of contact. The curvature of this projection at that
point may be called the normal curvature of the line on the surface.

It is easy to shew that this normal curvature is the same as the curvature
of a normal section of the surface passing through a tangent to the curve at
the same point.

10. General considerations applicable to tlie inscribed polyhedron.

When two series of lines of bending belonging to the first and second systems
have been described on the surface, we may proceed, as in Art. (1), to describe
a third series of curves so as to pass through all their intersections and form
the diagonals of the quadrilaterals formed by the first pair of systems.

Plane triangles may then be constituted within the surface, having these
points of intersection for angles, and the size of the facets of this polyhedron may
be diminished indefinitely by increasing the number of curves in each series.

But by Art. (8) the first and second systems of lines of bending are conju-
gate to each other, and therefore by Art. (4) the polygon just constructed will
have every pair of triangular facets in the same plane, and may therefore be

TRANSFORMATION OF SURFACES BY BENDING.

considered as a polyhedron with plane quadrilateral facets all whose solid angles
are formed by four of these facets meeting in a point.

When the number of curves in each system is increased and their distance
diminished indefinitely, the plane facets of the polyhedron will ultimately coincide
with the curved surface, and the polygons formed by the successive edges between
the facets, will coincide with the lines of bending.

These quadrilaterals may then be considered as parallelograms, the length
of which is determined by the portion of a curve of the second system inter-
cepted between two curves of the first, while the breadth is the distance of
two curves of the second system measured along a curve of the first. The
expressions for these quantities will be given when we come to the calculation of
our results along with the other particulars which we only specify at present.

The angle of the sides of these parallelograms will be ultimately the same
as the angle of intersection of the first and second systems, which we may
call <f> ; but if we suppose the dimensions of the facets to be small quantities
of the first order, the angles of the four facets which meet in a point will differ
from the angle of intersection of the curves at that point by small angles of
the first order depending on the tangential curvature of the lines of bending.
The sum of these four angles will differ from four right angles by a small
angle of the second order, the circular measure of which expresses the entire
curvature of the solid angle as in Art. (2).

The angle of inclination of two adjacent facets will depend on the normal
curvature of the lines of bending, and will be that of the projection of two con-
secutive sides of the polygon of one system on a plane perpendicular to a side
of the other system.

11. Explanation of the Notation to be employed in calculation.

Suppose each system of lines of bend-
ing to be determined by an equation con-
taining one arbitrary parameter.

Let this parameter be u for the first
system, and u for the second.

Let two curves, one from each system,
be selected as curves of reference, and let
their parameters be u, and u\.

13—2

100 TRANSFORMATION OF SURFACES BY BENDING.

Let ON and OM in the figure represent these two curves.

Let PM be any curve of the first system whose parameter is u, and PN
any curve of the second whose parameter is u', then their intersection P may
be defined as the point (u, u'), and all quantities referring to the point P may
be expressed as functions of u and u'.

Let PN, the length of a curve of the second system (u), from N (w0) to P
(u), be expressed by s, and PM the length of the curve (u) from («',) to (u), by
/, then s and s' will be functions of u and u.

Let (u + 8u) be the parameter of the curve QV of the first system consecu-
tive to PM. Then the length of PQ, the part of the curve of the second system
intercepted between the curves (u) and (u + 8u), will be

— 8u
du

Similarly PR may be expressed by

— 8u'

These values of PQ and PR will be the ultimate values of the length and
breadth of a quadrilateral facet.

The angle between these lines will be ultimately equal to <}>, the angle of
intersection of the system ; but when the values of Stt and 8u' are considered as
finite though small, the angles a, b, c, d of the facets which form a solid angle
will depend on the tangential curvature of the two systems of lines.

Let r be the tangential curvature of a curve of the first system at the
given point measured in the direction in which u increases, and let ?•', that of the
second system, be measured in the direction in which u' increases.

Then we shall have for the values of the four plane angles which meet at P,

1 ds 1 ds

I _, 1 ds ^ , I ds ~

, 1 <&' s , !_ * s

, , 1 ds' _ , 1 ds ^

TRANSFORMATION OF SURFACES BY BENDING. 101

These values are correct as far as the first order of small quantities. Those
corrections which depend on the curvature of the surface are of the second order.

Let p be the normal curvature of a curve of the first system, and p' that
of a curve of the second, then the inclination I of the plane facets a and b,
separated by a curve of the second system, will be

ds'

' I 7 / Vw »

p sin <p du
as far as the first order of small angles, and the inclination I' of b and c will be

7, 1 ds *
I =-7—. — 1 -^-8u
p siu<j> du

to the same order of exactness.

12. On the corresponding polygon on the surface of the sphere of reference.

By the method described in Art. (2) we may
find a point on the sphere corresponding to each
facet of the polyhedron.

In the annexed figure, let a, b, c, d be the
points on the sphere corresponding to the four facets
which meet at the solid angle P. Then the area
of the spherical quadrilateral a, b, c, d will be the
measure of the entire curvature of the solid angle P.

This area is measured by the defect of the sum of the exterior angles
from four right angles ; but these exterior angles are equal to the four angles
a, b, c, d, which form the solid angle P, therefore the entire curvature is
measured by

Since a, b, c, d are invariable, it is evident, as in Art. (2), that the entire
curvature at P is not altered by bending.

By the last article it appears that when the facets are small the angles b
and d are approximately equal to $, and a and c to (IT — <j>), and since the sides
of the quadrilateral on the sphere are small, we may regard it as approximately
a plane parallelogram whose angle bad = <f>.

The sides of this parallelogram will be I and I', the supplements of the
angles of the edges of the polyhedron, and we may therefore express its area
as a plane parallelogram

k = IV sin <f>.

102

TRANSFORMATION OF SURFACES BY BEXDIKO.

By the expression for I and T in the last article, we find

1 ds ds' s s ,

K = — j -j- -j-, OU OU

pp sin 9 (fit au
for the entire curvature of one solid angle.

Since the whole number of solid angles is equal to the whole number of
facets, we may suppose a quarter of each of the facets of which it is composed
to be assigned to each solid angle. The area of these will be the same as that

of one whole facet, namely,

. ,dsds'~ ~ ,
sin d> -j- -j—, ou ou ;
^ du du

therefore dividing the expression for k by this quantity, we find for the value
of the specific curvature at P

which gives the specific curvature in terms of the normal curvatures of the
lines of bending and their angle of intersection.

13. Further reduction of this expression by means of the " Conic of Con-
tact," as defined in Art. (3).

Let a and b be the semiaxes of the conic of contact, and h the sagitta
or perpendicular to its plane from the centre to the surface.

Let CP, CQ be semidiameters parallel to the
lines of bending of the first and second systems, and
therefore conjugate to each other.

By (Art. 3),

and p =

and the expression for p in Art. (12), becomes

But CP . CQ sin <£ is the area of the parallelogram CPRQ, which is one
quarter of the circumscribed parallelogram, and therefore by a well-known
theorem

TRANSFORMATION OF SURFACES BY BENDING. 103

and the expression for p becomes

or if the area of the circumscribing parallelogram be called A,

I6h3

P--JT-

The principal radii of curvature of the surface are parallel to the axes of
the conic of contact. Let R and R denote these radii, then

£ = *! and % = \\;

and therefore substituting in the expression for p,

1

or the specific curvature is the reciprocal of the product of , the principal radii
of curvature.

This remarkable expression was introduced by Gauss in the memoir referred
to in a former part of this paper. His method of investigation, though not
so elementary, is more direct than that here given, and will shew how this
result can be obtained without reference to the geometrical methods necessary
to a more extended inquiry into the modes of bending.

14. On the variation of normal curvature of the lines of bending as we pass
from one point of the surface to another.

We have determined the relation between the normal curvatures of the
lines of bending of the two systems at their points of intersection ; we have
now to find the variation of normal curvature when we pass from one line of
the first system to another, along a line of the second.

In analytical language we have to find the value of

-

du

Referring to the figure in Art. (11), we shall see that this may be done
if we can determine the difference between the angle of inclination of the
facets a and b, and that of c and d : for the angle I between a and b is

7 1 ds' >
I = . . -V-, Su ,
psunp au

104

TRANSFORMATION OF SURFACES BY BKNDIXG.

and therefore the difference between the angle of a and b and that of c and d is

*, dl * d I 1 «Y\ £ c ,
S( = -,- 5u = j- ( — : — j j— ,) 8«5u ;
au au \pH\ntp an J

whence the differential of p with respect to M may be found.

We must therefore find 8/, and this is done by means of the quadrilateral
on the sphere described in Art. (12).

15. To find tJie values of 81 and 81'.

In the annexed figure let abed repre-
sent the small quadrilateral on the surface
of the sphere. The exterior angles a, b,
c, d are equal to those of the four facets
which meet at the point P of the surface,
and the sides represent the angles which
the planes of those facets make with each
other ; so that

and the problem is to determine 81 and 81' in terms of the sides I and I' and
the angles o, b, c, d.

On the sides ba, be complete the parallelogram abed.

Produce ad to p, so tiiat op = a8. Join S^>.
Make cq = cd and join dq.
then. 8l = cd—ab,
= cq — c8,
= - (qo + 08).

Now qo = qd tan qdo

= cd sin qcd cot qod,
but cd = l nearly, sin qcd = qcd = (c + 6 - tr) and

TRANSFORMATION OF SURFACES BY BENDING. 105

Also 08 = . ., —
sin Sop

1

= aS(Sop) -,-

Substituting the values of a, b, c, d from Art. (11),
81= -

, 1 ds ,~ v 1 ds' 1 £ ,

= — I — -y- COt <}>OU — I -- j— ; — - r OU ,

r du r du sin <p

Finally, substituting the values of I, I', and 8Z from Art. (14),
d I 1 cfe'\ ,, 5. , cot $ ds' \ ds ~ ~. , 1 efe 1 efe'

—5 — , 3-7) OU OM = -- r-Z-i -j—, — -j-bubu -- , . -5 -- -y— ,

du \p sin (p du / p sin <f> du r du p sin $ du r du

which may be put under the more convenient form

d n * d , /I cfo'\ , 1 cfo . p 1 ds 1

j- (log p) = -j- log -r— f -j-, H -- 7 T- COt <A + -S - j -- : - 7 J

du ^ du 3 \sin $ du) r du p r du sin <£

and from the value of SZ' we may similarly obtain

d ,. d , /I <M , 1 ds' , a* 1 ds' 1

-T~, (lOff P ) = I"/ l°ff — - 7 J- + - J~? COt * + — — -j-, -. - 7 .

du ' dtt ' \sm ^ dw/ r du p r du sin (j>

We may simplify these equations by putting p for the specific curvature of
the surface, and q for the ratio -^ , which is the only quantity altered by bending.
We have then

PP

whence p* = <
and the equations become

-j- (log a) = -T- log ( p -j- ) + —, -j- cot <4 H — -v — 7 q,
du v 5 -1' rftt ' \* c?w 1 1 r du r du sin 9

rf , / ~ds^\ 2 dsT 2 ds' 1 1

In this way we may reduce the problem of bending a surface to the
consideration of one variable q, by means of the lines of bending.

14

VOL. I.

IOC TRANSFORMATION OF SURFACES BY BENDING.

16. To obtain the condition of Instantaneous lines of bending.

We have now obtained the values of the differential coefficients of q with
respect to each of the variables u, «'.

From the equation

we might find an equation which would give certain conditions of lines of
bending. These conditions however would be equivalent to those which we have
already assumed when we drew the systems of lines so as to be conjugate to
each other.

To find the true conditions of bending we must suppose the form of the
surface to vary continuously, so as to depend on some variable t which we
may call the time.

Of the different quantities which enter into our equations, none are changed
by the operation of bending except q, so that in differentiating with respect
to t all the rest may be considered constant, q being the only variable.

Differentiating the equations of last article with respect to t, we obtain
d* n , 2 ds 1 d

eP /, 2 ds 1 1 d

^(log2) =

Whence

Zdsl d x d , , 2 cfcs 1 d

rd^^ + rTu 2E?

and

d /2 ds' 1 \ 2 ds' 1 d , \ 1 d ., 2 ds' 1 1 d

^ VW ^/ ~ rWshT^ 10g ? J ? cfc (1°g ^

..

two independent values of the same quantity, whence the required conditions
may be obtained.

TRANSFORMATION OF SURFACES BY BENDING. 107

Substituting in these equations the values of those quantities which occur
in the original equations, we obtain

1 ds f d . I ds . ,\ 2ds' .}
Q--J- -T-, log (pr j- sin <b )+ - j— , cot <& >
* r du {du = \f du v / r du r J

1 1 ds' ( d . / ,ds ,\ 2 ds ,}
= — > -J-, •{ -j—, log pr -j—, sui <b ) + — -j- cot (4 >- ,
q r du \du = \* du YJ r du rj '

which is the condition which must hold at every instant during the process of
bending for the lines about which the bending takes place at that instant.
When the bending is such that the position of the lines of bending on the
surface alters at every instant, this is the only condition which is required.
It is therefore called the condition of Instantaneous lines of bending.

17. To find the condition of Permanent lines of bending.

Since q changes with the time, the equation of last article will not be
satisfied for any finite time unless both sides are separately equal to zero. In
that case we have the two conditions

d , / ds ,\ 2 ds'

or - 3- = 0.
r du

d , I ,ds . \ 2 ds

Q

or -, -T-, = 0.
r du

If the lines of bending satisfy these conditions, a finite amount of bending
may take place without changing the position of the system on the surface.
Such lines are therefore called Permanent lines of bending.

The only case in which the phenomena of bending may be exhibited by
means of the polyhedron with quadrilateral facets is that in which permanent
lines of bending are chosen as the boundaries of the facets. In all other cases
the bending takes place about an instantaneous system of lines which is con-
tinually in motion with respect to the surface, so that the nature of the poly-
hedron would need to be altered at every instant.

14—2

108 TRANSFORMATION OF SURFACES BY BENDING.

We are now able to determine whether any system of lines drawn on a
given surface is a system of instantaneous or permanent lines of bending.

We are also able, by the method of Article (8), to deduce from two con-
secutive forms of a surface, the lines of bending about which the transformation
must have taken place.

. If our analytical methods were sufficiently powerful, we might apply our
results to the determination of such systems of lines on any known surface, but
the necessary calculations even in the simplest cases are so complicated, that,
even if useful results were obtained, they would be out of place in a paper of
this kind, which is intended to afford the means of forming distinct conceptions
rather than to exhibit the results of mathematical labour.

18. On the application of the ordinary methods of analytical geometry to the
consideration of lines of bending.

It may be interesting to those who may hesitate to accept results derived
from the consideration of a polyhedron, when applied to a curved surface, to
inquire whether the same results may not be obtained by some independent
method.

As the following method involves only those operations which are most
familiar to the analyst, it will be sufficient to give the rough outline, which may
be filled up at pleasure.

The proof of the invariability of the specific curvature may be taken from
any of the memoirs above referred to, and its value in terms of the equation of
the surface will be found in the memoir of Gauss.

Let the equation to the surface be put under the form

then the value of the specific curvature is

d*z d*z d\ I*
chi? dy* dxdy\

dy

The definition of conjugate systems of curves may be rendered independent
of the reasoning formerly employed by the following modification.

TRANSFORMATION OF SURFACES BY BENDING. 109

Let a tangent plane move along any line of the first system, then if the line
of ultimate intersection of this plane with itself be always a tangent to some line
of the second system, the second system is said to be conjugate to the first.

It is easy to show that the first system is also conjugate to the second.

Let the system of curves be projected on the plane of xy, and at the point
(x, y) let a be the angle which a projected curve of the first system makes with
the axis of x, and )8 the angle which the projected curve of the second system
which intersects it at that point makes with the same axis. Then the condition
of the systems being conjugate will be found to be

d'z „ d'z

—

a and /8 being known as functions of x and y, we may determine the nature
of the curves projected on the plane of xy.

Supposing the surface to touch that plane at the origin, the length and
tangential curvature of the lines on the surface near the point of contact may
be taken the same as those of their projections on the plane, and any change
of form of the surface due to bending will not alter the form of the projected
lines indefinitely near the point of contact. We may therefore consider z as the
only variable altered by bending; but in order to apply our analysis with facility,
we may assume

^ = PQ sin1 a + PQ-1 sin' &

= — PQ sin a cos a — PQ~ ' sin /3 cos /3,

dxdy

cPz
3? =

It will be seen that these values satisfy the condition last given. Near the
origin we have .

d'z d'z d'z *

^_

* dx* dy* dxdy
and q = Q~*.

= P°sin2(a-/3),

110 TRANSFORMATION OF SURFACES BY BENDING.

</•'• f/*z

Differentiating these values of -7-5, &c., we shall obtain two values of •, . ,

d*z
and of j—-j-\, which being equated will give two equations of condition.

Now if s be measured along a curve of the first system, and R be any
function of x and y, then

dR dR dR .

-j-f = -5- cos a+ — ,- sui a.
as ax 1 1 ij

, dR dRds'

ana -,— -, = -j-> -*—, .
du as du

We may also show that T-t = - ,

ds r

da da d , ids'

, ., da . da d , ids . ,\

and that cos a -5 sin a -y- = -y- log I -7—, sin d> .

dy dx ds = \du ^

By substituting these values in the equations thus obtained, they are
reduced to the two equations given at the end of (Art. 15). This method of
investigation introduces no difficulty except that of somewhat long equations, and
is therefore satisfactory as supplementary to the geometrical method given at
length.

As an example of the method given in page (2), we may apply it to
the case of the surface whose equation is

x \ + pLY = M'.

This surface may be generated by the motion of a straight line whose
equation is of the form

i — 1, y = aemt (l + -} ,

t being the variable, by the change of which we pass from one position of the
line to another. This line always passes through the circle

z = 0, xt + y> = a\
and the straight lines z = c, x = 0,
and 2= — c, y = 0,
which may therefore be taken as the directors of the surface.

TRANSFORMATION OF SURFACES BY BENDING. Ill

Taking two consecutive positions of this line, in which the values of t
are t and t + St, we may find by the ordinary methods the equation to the
shortest line between them, its length, and the co-ordinates of the point in which
it intersects the first line.

Calling the length S£,

and the co-ordinates of the point of intersection are

x = 2a cos3 1, y = 2a sin3 1 , z= —c cos 2t.
The angle 80 between the consecutive lines is

The distance So- between consecutive shortest lines is

and the angle 8$ between these latter lines is

-ct

Hence if we suppose £, 6, cr, <j>, and t to vanish together, we shall have by
integration

^27^(1-C082<)'

(T =

= £^L(1-COS20,

2Va' + cjV
<H-7^=*.

I ft* _L y>>

By bending the surface about its generating lines we alter the value of <£
in any manner without changing £, 6, or o-. For instance, making <f> — 0, all the
generating lines become parallel to the same plane. Let this plane be that of
xy, then £ is the distance of a generating line from that plane. The projections

112 TRANSFORMATION OF SURFACES BY BENDING.

of the generating lines on the plane of xy will, by their ultimate intersections,
form a curve, the length of which is measured by <r, and the angle which its
tangent makes with the axis of x by 0, 0 and or being connected by the equation

3a' + 2c* /. 2ja* + c' \

•=—-=== 1 - cos - 0],

2>/a' + c'\ « /

which shows the curve to be an epicycloid.

The generating lines of the surface when bent into this form are therefore
tangents to a cylindrical surface on an epicycloidal base, touching that surface
along a curve which is always equally inclined to the plane of the base, the
tangents themselves being drawn parallel to the base.

We may now consider the bending of the surface of revolution

Putting r = «/x' + y1, then the equation of the generating line is

This is the well-known hypocycloid of four cusps.

Let s be the length of the curve measured from the cusp in the axis of z,
then,

wherefore, r = (§)*c~*s*.

Let 6 be the angle which the plane of any generating line makes with
that of xz, then s and 0 determine the position of any point on the surface.
The length and breadth of an element of the surface will be 8s and r86.

Now let the surface be bent in the manner formerly described, so that 0
becomes ff ', and r, r , when

& = u.d and rf=-r,
/*

then »/ = (t)lc-|/i-1*1

= (f)*c'-M,
provided c' = ^c.

The equation between r' and s being of the same form as that between
r and a shows that the surface when bent is similar to the original surface, its
dimensions being multiplied by /*'.

TRANSFORMATION OF SURFACES BY BENDING. 113

This, however, is true only for one half of the surface when bent. The
other half is precisely symmetrical, but belongs to a surface which is not con-
tinuous with the first.

The surface in its original form is divided by the plane of xy into two
parts which meet in that plane, forming a kind of cuspidal edge of a circular
form which limits the possible value of s and r.

After being bent, the surface still consists of the same two parts, but the
edge in which they meet is no longer of the cuspidal form, but has a finite

angle = 2 cos"1 - , and the two sheets of the surface become parts of two different
P

surfaces which meet but are not continuous.

NOTE.

As an example of the application of the more general theory of " lines of bending," let us
consider the problem which has been already solved by Professor Jellett.

To determine ilie conditions under which one portion of a surface may be rendered rigid, while
tiie remainder is flexible.

Suppose the lines of bending to be traced on the surface, and the corresponding poly-
hedron to be formed, as in (9) and (10), then if the angle of one of the four edges which
meet at any solid angle of the polyhedron be altered by bending, those of the other three
must be also altered. These edges terminate in other solid angles, the forms of which will
also be changed, and therefore the effect of the alteration of one angle of the polyhedron will
be communicated to every other angle within the system of lines of bending which defines
the form of the polyhedron.

If any portion of the surface remains unaltered it must lie beyond the limits of the
system of lines of bending. We must therefore investigate the conditions of such a system
being bounded.

The boundary of any system of lines on a surface is the curve formed by the ultimate inter-
section of those lines, and therefore at any given point coincides in direction with the curve of
the system which passes through that point. In this case there are two systems of lines of
bending, which are necessarily coincident in extent, and must therefore have the same boundary.
At any point of this boundary therefore the directions of the lines of bending of the first
and second systems are coincident.

But, by (7), these two directions must be "conjugate" to each other, that is, must corre-
spond to conjugate diameters of the " Conic of Contact." Now the only case in which con-

VOL. I. 15

114 TRANSFORMATION OF SURFACES BY BENDING.

jugate diameters of a conic can coincide, is when the conic is an hyperbola, and both diameters
coincide with one of the asymptotes ; therefore the boundary of the system of lines of bending
must be a curve at every point of which the conic of contact is an hyperbola, one of whose
asymptotes lies in the direction of the curve. The radius of " normal curvature " must there-
fore by (3) be infinite at every point of the curve. This is the geometrical property of
what Professor Jcllett calls a " Curve of Flexure," so that we may express the result as
follows :

If one portion of a surface be fixed, while the remainder is bent, the boundary of the fixed
portion is a curve of flexure.

This theorem includes those given at p. (92), relative to a fixed curve on a surface, for in
a surface whose curvatures are of the same sign, there can be no " curves of flexure," and
in a developable surface, they are the rectilinear sections. Although the cuspidal edge, or
arete de rebroussement, satisfies the analytical condition of a curve of flexure, yet, since its
form determines that of the whole surface, it cannot remain fixed while the form of the surface
is changed.

In concavo-convex surfaces, the curves of flexure must either have tangential curvature or
be straight lines. Now if we put ^=0 in the equations of Art. (17), we find that the
lines of bending of both systems have no tangential curvature at the point where they touch
the curve of flexure. They must therefore lie entirely on the convex side of that curve, and
therefore

If a curve of flexure be fixed, Hie surface on tiie concave side of the curve is not flexible.

I have not yet been able to determine whether the surface is inflexible on the convex side
of the curve. It certainly is so in some cases which I have been able to work out, but I
have no general proof.

When a surface has one or more rectilinear sections, the portions of the surface between
them may revolve as rigid bodies round those lines as axes in any manner, but no other motion
is possible. The case in which the rectilinear sections form an infinite series has been discussed
in Sect. (I.).

[From the Cambridge and Dublin Mathematical Journal, Vol. ix.]

V. On a particular case of the descent of a heavy body in a resisting

medium.

EVERY one must have observed that when a slip of paper falls through
the air, its motion, though undecided and wavering at first, sometimes becomes
regular. Its general path is not in the vertical direction, but inclined to it
at an angle which remains nearly constant, and its fluttering appearance will
be found to be due to a rapid rotation round a horizontal axis. The direction
of deviation from the vertical depends on the direction of rotation.

If the positive directions of an axis be toward the right hand and upwards,
and the positive angular direction opposite to the direction of motion of the
hands of a watch, then, if the rotation is in the positive direction, the hori-
zontal part of the mean motion will be positive.

These effects are commonly attributed to some accidental peculiarity in the
form of the paper, but a few experiments with a rectangular slip of paper
(about two inches long and one broad), will shew that the direction of rotation
is determined, not by the irregularities of the paper, but by the initial circum-
stances of projection, and that the symmetry of the form of the paper greatly
increases the distinctness of the phenomena. We may therefore assume that
if the form of the body were accurately that of a plane rectangle, the same
effects would be produced.

The following investigation is intended as a general explanation of the true
cause of the phenomenon.

I suppose the resistance of the air caused by the motion of the plane to
be in the direction of the normal and to vary as the square of the velocity
estimated in that direction.

Now though this may be taken as a sufficiently near approximation to the
magnitude of the resisting force on the plane taken as a whole, the pressure

15—2

116 DESCENT OF A HEAVY BODY IN A RESISTING MEDIUM.

on any given element of the surface will vary with its position so that the
resultant force will not generally pass through the centre of gravity.

It is found by experiment that the position of the centre of pressure
depends on the tangential part of the motion, that it lies on that side of the
centre of gravity towards which the tangential motion of the plane is directed,
and that its distance from that point increases as the tangential velocity in-
creases.

I am not aware of any mathematical investigation of this effect. The
explanation may be deduced from experiment.

Place a body similar in shape to the slip of paper obliquely in a current
of some visible fluid. Call the edge where the fluid first meets the plane the
first edge, and the edge where it leaves the plane, the second edge, then we
may observe that

(1) On the anterior side of the plane the velocity of the fluid increases
as it moves along the surface from the first to the second edge, and therefore
by a known law in hydrodynamics, the pressure must diminish from the first
to the second edge.

(2) The motion of the fluid behind the plane is very unsteady> but may
be observed to consist of a series of eddies diminishing in rapidity as they
pass behind the plane from the first to the second edge, and therefore relieving
the posterior pressure most at the first edge.

Both these causes tend to make the total resistance greatest at the first
edge, and therefore to bring the centre of pressure nearest to that edge.

Hence the moment of the resistance about the centre of gravity will always
tend to turn the plane towards a position perpendicular to the direction of the
current, or, in the case of the slip of paper, to the path of the body itself. It
will be shewn that it is this moment that maintains the rotatory motion of
the falling paper.

When the plane has a motion of rotation, the resistance will be modified
on account of the unequal velocities of different parts of the surface. The
magnitude of the whole resistance at any instant will not be sensibly altered
if the velocity of any point due to angular motion be small compared with that
due to the motion of the centre of gravity. But there will be an additional
moment of the resistance round the centre of gravity, which will always act in
the direction opposite to that of rotation, and will vary directly as the normal
and angular velocities together.

DESCENT OF A HEAVY BODY IN A RESISTING MEDIUM. 117

The part of the moment due to the obliquity of the motion will remain
nearly the same as before.

We are now prepared to give a general explanation of the motion of the
slip of paper after it has become regular.

Let the angular position of the paper be determined by the angle between
the normal to its surface and the axis of x, and let the angular motion be
such that the normal, at first coinciding with the axis of x, passes towards
that of y.

The motion, speaking roughly, is one of descent, that is, in the negative
direction along the axis of y.

The resolved part of the resistance in the vertical direction will always
act upwards, being greatest when the plane of the paper is horizontal, and
vanishing when it is vertical.

When the motion has become regular, the effect of this force during a
whole revolution will be equal and opposite to that of gravity during the same
time.

Since the resisting force increases while the normal is in its first and third
quadrants, and diminishes when it is in its second and fourth, the maxima of
velocity will occur when the normal is in its first and third quadrants, and
the minima when it is in the second and fourth.

The resolved part of the resistance in the horizontal direction will act in
the positive direction along the axis of x in the first and third quadrants, and
in the negative direction during the second and fourth ; but since the resistance
increases with the velocity, the whole effect during the first and third quadrants
will be greater than the whole effect during the second and fourth. Hence
the horizontal part of the resistance will act on the whole in the positive
direction, and will therefore cause the general path of the body to incline in
that direction, that is, toward the righf.

That part of the moment of the resistance about the centre of gravity
which depends on the angular velocity will vary in magnitude, but will always
act in the negative direction. The other part, which depends on the obliquity
of the plane of the paper to the direction of motion, will be positive in the
first and third quadrants and negative in the second and fourth ; but as its
magnitude increases with the velocity, the positive effect will be greater than
the negative.

When the motion has become regular, the effect of this excess in the

118 DESCENT OF A HEAVY BODY IX A RESISTING MEDIUM.

|>ositive direction will be equal and opposite to the negative effect due to the
angular velocity during a whole revolution.

The motion will then consist of a succession of equal and similar parts
j>erformed in the same manner, each part corresponding to half a revolution of
the paper.

These considerations will serve to explain the lateral motion of the paper,
and the maintenance of the rotatory motion.

Similar reasoning will shew that whatever be the initial motion of the
|>aper, it cannot remain uniform.

Any accidental oscillations will increase till their amplitude exceeds half a
revolution. The motion will then become one of rotation, and will continually
approximate to that which we have just considered.

It may be also shewn that this motion will be unstable unless it take
place about the longer axis of the rectangle.

If this axis is inclined to the horizon, or if one end of the slip of paper
be different from the other, the path will not be straight, but in the form of
a helix. There will be no other essential difference between this case and that
of the symmetrical arrangement.

Trinity College, April 5, 1853.

[From the Transactions of the Royal Scottish Society of Arts, Vol. iv. Part in.]

VI. On the Theory of Colours in relation to Colour- Blindness.
A letter to Dr G. Wilson.

DEAR SIR, — As you seemed to think that the results which I have obtained
in the theory of colours might be of service to you, I have endeavoured to
arrange them for you in a more convenient form than that in which I first
obtained them. I must premise, that the first distinct statement of the theory
of colour which I adopt, is to be found in Young's Lectures on Natural Philo-
sophy (p. 345, Kelland's Edition) ; and the most philosophical enquiry into it
which I have seen is that of Helmholtz, which may be found in the Annals of
Philosophy for 1852.

It is well known that a ray of light, from any source, may be divided by
means of a prism into a number of rays of different refrangibility, forming a
series called a spectrum. The intensity of the light is different at different
points of this spectrum ; and the law of intensity for different refrangibilities
differs according to the nature of the incident light. In Sir John F. W.
Herschel's Treatise on Light, diagrams will be found, each of which represents
completely, by means of a curve, the law of the intensity and refrangibility of
a beam of solar light after passing through various coloured media.

I have mentioned this mode of defining and registering a beam of light,
because it is the perfect expression of what a beam of light is in itself, con-
sidered with respect to all its properties as ascertained by the most refined
instruments. When a beam of light falls on the human eye, certain sensations
are produced, from which the possessor of that organ judges of the colour and
intensity of the light. Now, though every one experiences these sensations, and
though they are the foundation of all the phenomena of sight, yet, on account
of their absolute simplicity, they are incapable of analysis, and can never become
in themselves objects of thought. If we attempt to discover them, we must

120 THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS.

do so by artificial means; and our reasonings on them must be guided by some
theory.

The most general form in which the existing theory can be stated is this, —

There are certain sensations, finite in number, but infinitely variable in
degree, which may be excited by the different kinds of light. The compound
sensation resulting from all these is the object of consciousness, is a simple act
of vision.

It is easy to see that the number of these sensations corresponds to what
may be called in mathematical language the number of independent variables, of
which sensible colour is a function.

This will be readily understood by attending to the following cases : —

1. When objects are illuminated by homogeneous yellow light, the only
thing which can be distinguished by the eye is difference of intensity or
brightness.

If we take a horizontal line, and colour it black at one end, with increasing
degrees of intensity of yellow light towards the other, then every visible object
will have a brightness corresponding to some point in this line.

In this case there is nothing to prove the existence of more than one
sensation in vision.

In those photographic pictures in which there is only one tint of which
the different intensities correspond to the different degrees of illumination of the
object, we have another illustration of an optical effect depending on one variable
only.

2. Now, suppose that different kinds of light are emanating from different
sources, but that each of these sources gives out perfectly homogeneous light,
then there will be two things on which the nature of each ray will depend : —
(1) its intensity or brightness; (2) its hue, which may be estimated by its
position in the spectrum, and measured by its wave length.

If we take a rectangular plane, and illuminate it with the different kinds
of homogeneous light, the intensity at any point being proportional to its hori-
zontal distance along the plane, arid its wave length being proportional to its
height above the foot of the plane, then the plane will display every possible
variety of homogeneous light, and will furnish an instance of an optical effect
depending on two variables.

THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS.

121

3. Now, let us take the case of nature. We find that colours differ not
only in intensity and hue, but also in tint ; that is, they are more or less pure.
We might arrange the varieties of each colour along a line, which should begin
with the homogeneous colour as seen in the spectrum, and pass through all
gradations of tint, so as to become continually purer, and terminate in white.

We have, therefore, three elements in our sensation of colour, each of which
may vary independently. For distinctness sake I have spoken of intensity, hue,
and tint ; but if any other three independent qualities had been chosen, the
one set might have been expressed in terms of the other, and the results identified.

The theory which I adopt assumes the existence of three elementary sen-
sations, by the combination of which all the actual sensations of colour are
produced. It will be shewn that it is not necessary to specify any given colours
as typical of these sensations. Young has called them red, green, and violet ; but
any other three colours might have been chosen, provided ih&t white resulted
from their combination in proper proportions.

Before going farther I would observe, that the important part of the theory
is not that three elements enter into our sensation of colour, but that there are
only three. Optically, there are as many elements in the composition of a ray
of light as there are different kinds of light in its spectrum ; and, therefore,
strictly speaking, its nature depends on an infinite number of independent
variables.

I now go on to the geometrical form into which the theory may be thrown.
Let it be granted that the three pure sensations corre-
spond to the colours red, green, and violet, and that we
can estimate the intensity of each of these sensations
numerically.

Let v, r, g be the angular points of a triangle, and
conceive the three sensations as having their positions at
these points. If we find the numerical measure of the
red, green, and violet parts of the sensation of a given
colour, and then place weights proportional to these parts

at r, g, and v, and find the centre of gravity of the three weights by the
ordinary process, that point will be the position of the given colour, and the
numerical measure of its intensity will be the sum of the three primitive
sensations.

In this way, every possible colour may have its position and intensity

VOL. I. 16

THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS.

ascertained; and it is easy to see that when two compound colours are com-
bined, their centre of gravity is the position of the new colour.

The idea of this geometrical method of investigating colours is to be found
in Newton's Opticks (Book I., Part 2, Prop. 6), but I am not aware that it has
been ever employed in practice, except in the reduction of the experiments
which I have just made. The accuracy of the method depends entirely on the
truth of the theory of three sensations, and therefore its success is a testimony
in favour of that theory.

Every possible colour must be included within the triangle rgv. White
will be found at some point, u\ within the triangle. If lines be drawn through
w to any point, the colour at that point will vary in hue according to the
angular position of the line drawn to w, and the purity of the tint will depend
on the length of that line.

Though the homogeneous rays of the prismatic spectrum are absolutely pure
in themselves, yet they do hot give rise to the "pure sensations" of which we
are speaking. Every ray of the- spectrum gives rise to all three sensations,
though in different proportions ; hence the position of the colours of the spectrum
is not at the boundary of the triangle, but in some curve C R Y G B V
considerably within the triangle. The nature of this curve is not yet determined,
but may form the subject of a future investigation *.

All natural colours must be within this curve, and all ordinary pigments
do in fact lie very much within it. The experiments on the colours of the
spectrum which I have made are not brought to the same degree of accuracy as
those on coloured papers. I therefore proceed at once to describe the mode of
making those experiments which I have found most simple and convenient.

The coloured paper is cut into the form of discs, each with a small hole
in the centre, and divided along a radius, so as to admit
of several of them being placed on the same axis, so that
part of each is exposed. By slipping one disc over another,
we can expose any given portion of each colour. These
discs are placed on a little top or teetotum, consisting of
a flat disc of tin-plate and a vertical axis of ivory. This
axis passes through the centre of the discs, and the quantity of each colour exposed
is measured by a graduation on the rim of the disc, which is divided into 100 parts.

* [See the author's Memoir in the Philosophical Transactions, I860, on the Theory of Compound
Colours, and on the relations of the Colours of the Spectrum.]

THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS. 123

By spinning the top, each colour is presented to the eye for a time pro-
portional to the angle of the sector exposed, and I have found by independent
experiments, that the colour produced by fast spinning is identical with that
produced by causing the light of the different colours to fall on the retina at
once.

By properly arranging the discs, any given colour may be imitated and
afterwards registered by the graduation on the rim of the top. The principal
use of the top is to obtain colour-equations. These are got by producing, by
two different combinations of colours, the same mixed tint. For this purpose
there is another set of discs, half the diameter of the others, which lie above
them, and by which the second combination of colours is formed.

The two combinations being close together, may be accurately compared, and
when they are made sensibly identical, the proportions of the different colours
in each is registered, and the results equated.

These equations in the case of ordinary vision, are always between four
colours, not including black.

From them, by a very simple rule, the different colours and compounds have
their places assigned on the triangle of colours. The rule for finding the position
is this : — Assume any three points as the positions of your three standard colours,
whatever they are ; then form an equation between the three standard colours,
the given colour and black, by arranging these colours on the inner and outer
circles so as to produce an identity when spun. Bring the given colour to the
left-hand side of the equation, and the three standard colours to the right hand,
leaving out black, then the position of the given colour is the centre of gravity
of three masses, whose weights are as the number of degrees of each of the
standard colours, taken positive or negative, as the case may be.

In this way the triangle of colours may be constructed by scale and compass
from experiments on ordinary vision. I now proceed to state the results of
experiments on Colour-Blind vision.

If we find two combinations of colours which appear identical to a Colour-
Blind person, and mark their positions on the triangle of colours, then the
straight line passing through these points will pass through all points corre-
sponding to other colours, which, to such a person, appear identical with the first
two.

We may in the same way find other lines passing through the series of

16—2

ll-M THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS.

colours which appear alike to the Colour-Blind. All these
lines either pass through one point or are parallel, ac-
cording to the standard colours which we have assumed,
:ind the other arbitrary assumptions we may have made.
Knowing this law of Colour-Blind vision, we may predict
any number of equations which will be true for eyes
having this defect.

The mathematical expression of the difference between
Colour-Blind and ordinary vision is, that colour to the
former is a function of two independent variables, but to an ordinary eye, of
three ; and that the relation of the two kinds of vision is not arbitrary, but
indicates the absence of a determinate sensation, depending perhaps upon some
undiscovered structure or organic arrangement, which forms one-third of the
apparatus by which we receive sensations of colour.

Suppose the absent structure to be that which is brought most into play
when red light falls on our eyes, then to the Colour-Blind red light will be
visible only so far as it affects the other two sensations, say of blue and
green. It will, therefore, appear to them much less bright than to us, and will
excite a sensation not distinguishable from that of a bluish-green light.

I cannot at present recover the results of all my experiments ; but I recollect
that the neutral colours for a Colour-Blind person may be produced by com-
bining 6 degrees of ultramarine with 94 of vermilion, or 60 of emerald-green
with 40 of ultramarine. The first of these, I suppose to represent to our eyes
the kind of red which belongs to the red sensation. It excites the other two
sensations, and is, therefore, visible to the Colour-Blind, but it appears very
dark to them and of no definite colour. I therefore suspect that one of the
three sensations in perfect vision will be found to correspond to a red of the
same hue, but of much greater purity of tint. Of the nature of the other two,
I can say nothing definite, except that one must correspond to a blue, and the
other to a green, verging to yellow.

I hope that what I have written may help you in any way in your
experiments. I have put down many things simply to indicate a way of thinking
about colours which belongs to this theory of triple sensation. We are indebted
to Newton for the original design ; to Young for the suggestion of the means
of working it out ; to Prof. Forbes* for a scientific history of its application

*Phil. Mag. 1848.

THE THEORY OF COLOURS IN RELATION TO COLOUR-BLINDNESS. 125

to practice ; to Helmholtz for a rigorous examination of the facts on which it
rests; and to Prof. Grassrnan (in the Phil. Mag. for 1852), for an admirable
theoretical exposition of the subject. The colours given in Hay's Nomenclature
of Colours are illustrations of a similar theory applied to mixtures of pigments,
but the results are often different from those in which the colours are combined
by the eye alone. I hope soon to have results with pigments compared with
those given by the prismatic spectrum, and then, perhaps, some more definite
results may be obtained. Yours truly,

J. C. MAXWELL.

EDINBURGH, 4th Jan. 1855.

[From the Transactions of tiie Royal Society of Edinburgh, Vol. XXI. Part II.]

VII. Experiments on Colour, as perceived by the Eye, with remarks on Colour-
blindness. Communicated by Dr Gregory.

THE object of the following communication is to describe a method by
which every variety of visible colour may be exhibited to the eye in such a
form as to admit of accurate comparison ; to shew how experiments so made
may be registered numerically ; and to deduce from these numerical results
certain laws of vision.

The different tints are produced by means of a combination of discs of paper,
painted with the pigments commonly used in the arts, and arranged round an
axis, so that a sector of any required angular magnitude of each colour may be
exposed. When this system of discs is set in rapid rotation, the sectors of
the different colours become indistinguishable, and the whole appears of one uni-
form tint. The resultant tints of two different combinations of colours may be
compared by using a second set of discs of a smaller size, and placing these over
the centre of the first set, so as to leave the outer portion of the larger discs
exposed. The resultant tint of the first combination will then appear in a ring
round that of the second, and may be very carefully compared with it.

The form in which the experiment is most manageable is that of the com-
mon top. An axis, of which the lower extremity is conical, carries a circular
plate, which serves as a support for the discs of coloured paper. The circumfer-
ence of this plate is divided into 100 equal parts, for the purpose of ascertaining
the proportions of the different colours which form the combination. When the
discs have been properly arranged, the upper part of the axis is screwed down,
so as to prevent any alteration in the proportions of the colours.

The instrument used in the first series of experiments (at Cambridge, in
November, 1854) was constructed by myself, with coloured papers procured from

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 127

Mr D. R Hay. The experiments made in the present year were with the
improved top made by Mr J. M. Bryson, Edinburgh, and coloured papers pre-
pared by Mr T. Purdie, with the unmixed pigments used in the arts. A number
of Mr Bryson's tops, with Mr Purdie's coloured papers has been prepared, so as
to afford different observers the means of testing and comparing results inde-
pendently obtained.

The colour used for Mr Purdie's papers were —

Vermilion . . V Ultramarine . . U Emerald Green . , EG

Carmine ... 0 Prussian Blue . . PB Brunswick Green . . BG

Red Lead . . RL Verditer Blue . . VB Mixture of Ultramarine

Orange Orpiment . OO and Chrome . . UO

Orange Chrome . OC
Chrome Yellow . CY
Gamboge . . Gam

Pale Chrome . . PC

Ivory Black . . Bk

Snow White . . SW

White Paper (Pirie, Aberdeen).

The colours in the first column are reds, oranges, and yellows ; those in
the second, blues ; and those in the third, greens. Vermilion, ultramarine, and
emerald green, seem the best colours to adopt in referring the rest to a uniform
standard. They are therefore put at the head of the h'st, as types of three
convenient divisions of colour, red, blue, and green.

It may be asked, why some variety of yellow was not chosen in place of
green, which is commonly placed among the secondary colours, while yellow
ranks as a primary \ The reason for this deviation from the received system is,
that the colours on the discs do not represent primary colours at all, but are
simply specimens of different kinds of paint, and the choice of these was deter-
mined solely by the power of forming the requisite variety of combinations. Now,
if red, blue, and yellow, had been adopted, there would have been a difficulty
in forming green by any compound of blue and yellow, while the yellow formed
by vermilion and emerald green is tolerably distinct. This will be more clearly
perceived after the experiments have been discussed, by referring to the diagram.

As an example of the method of experimenting, let us endeavour to form a
neutral gray by the combination of vermilion, ultramarine, and emerald green.
The most perfect results are obtained by two persons acting in concert, when

128 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

the operator arranges the colours and spins the top, leaving the eye of the
observer free from the distracting effect of the bright colours of the papers when
at rest.

After placing discs of these three colours on the circular plate of the top,
and smaller discs of white and black above them, the operator must spin the
top, and demand the opinion of the observer respecting the relation of the
outer ring to the inner circle. He will be told that the outer circle is too
red, too blue, or too green, as the case may be, and that the inner one is too
light or too dark, as compared with the outer. The arrangement must then be
changed, so as to render the resultant tint of the outer and inner circles more
nearly alike. Sometimes the observer will see the inner circle tinted with the
complementary colour of the outer one. In this case the operator must interpret
the observation with respect to the outer circle, as the inner circle contains only
black and white.

By a little experience the operator will learn how to put his questions, and
how to interpret their answers. The observer should not look at the coloured
papers, nor be told the proportions of the colours during the experiments.
When these adjustments have been properly made, the resultant tints of the
outer and inner circles ought to be perfectly indistinguishable, when the top
has a sufficient velocity of rotation. The number of divisions occupied by the
different colours must then be read off on the edge of the plate, and registered
in the form of an equation. Thus, in the preceding experiment we have ver-
milion, ultramarine, and emerald green outside, and black and white inside. The
numbers, as given by an experiment on the Gth March 1855, in daylight without
sun, are —

•37 V + -27 U + -36 EG = '28 SW + 72 Bk (1).

The method of treating these equations will be given when we come to the
theoretical view of the subject.

In this way we have formed a neutral gray by the combination of the
three standard colours. We may also form neutral grays of different intensities
by the combination of vermilion and ultramarine with the other greens, and thus
obtain the quantities of each necessary to neutralize a given quantity of the
proposed green. By substituting for each standard colour in succession one of the
colours which stand under it, we may obtain equations, each of which contains
two standard colours, and one of the remaining colours.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 129

Thus, in the case of pale chrome, we have, from the same set of experiments,
•34 PC + -55U + -12 EG='37 SW + '63 Bk (2).

We may also make experiments in which the resulting tint is not a neutral
gray, but a decided colour. Thus we may combine ultramarine, pale chrome, and
black, so as to produce a tint identical with that of a compound of vermilion
and emerald-green. Experiments of this sort are more difficult, both from the
inability of the observer to express the difference which he detects in two tints
which have, perhaps, the same hue and intensity, but differ in purity ; and also
from the complementary colours which are produced in the eye after gazing too
long at the colours to be compared.

The best method of arriving at a result in the case before us, is to render
the hue of the red and green combination something like that of the yellow, to
reduce the purity of the yellow by the admixture of blue, ^nd to diminish its
intensity by the addition of black. These operations must be repeated and
adjusted, till the two tints are not merely varieties of the same colour, but
absolutely the same. An experiment made 5th March gives —

•39 PC + -21 U + -40 Bk = '59 V+'41 EG (3).

That these experiments are really evidence relating to the constitution of the
eye, and not mere comparisons of two things which are in themselves identical,
may be shewn by observing these resultant tints through coloured glasses, or by
using gas-light instead of day-light. The tints which before appeared identical
will now be manifestly different, and will require alteration, to reduce them to
equality.

Thus, in the case of carmine, we have by day-light,

•440 + '22 U+ -34 EG ='17 SW+'83Bk,
while by gas-light (Edinburgh)

•47 C + -08 U-K45 EG = '25 SW + 75 Bk,

which shews that the yellowing effect of the gas-light tells more on the white
than on the combination of colours. If we examine the two resulting tints
which appeared identical in experiment (3), observing the whirling discs through
a blue glass, the combination of yellow, blue, and black, appears redder than the
other, while through a yellow glass, the red and green mixture appears redder.
So also a red glass makes the first side of the equation too dark, and a green
glass makes it too light.

VOL. I. 17

130 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

The apparent identity of the tints in these experiments is therefore not real,
but a consequence of a determinate constitution of the eye, and hence arises
the importance of the results, as indicating the laws of human vision.

The first result which is worthy of notice is, that the equations, as observed
by different persons of ordinary vision, agree in a remarkable manner. If care
be taken to secure the same kind of light in all the experiments, the equations,
as determined by two independent observers, will seldom shew a difference of
more than three divisions in any part of the equation containing the bright
standard colours. As the duller colours are less active in changing the resultant
tint, their true proportions cannot be so well ascertained. The accuracy of vision
of each observer may be tested by repeating the same experiment at different
times, and comparing the equations so found.

Experiments of this kind, made at Cambridge in November 1854, shew that
of ten observers, the best were accurate to within l£ division, and agreed
within 1 division of the mean of all ; and the worst contradicted themselves to
the extent of 6 degrees, but still were never more than 4 or 5 from the mean
of all the observations.

We are thus led to conclude —

1st. That the human eye is capable of estimating the likeness of colours
with a precision which in some cases is very great.

2nd. That the judgment thus formed is determined, not by the real identity
of the colours, but by a cause residing in the eye of the observer.

3rd. That the eyes of different observers vary in accuracy, but agree with
each other so nearly as to leave no doubt that the law of colour-vision is
identical for all ordinary eyes.

Investigation of the Law of the Perception of Colour.

Before proceeding to the deduction of the elementary laws of the perception
of colour •from the numerical results previously obtained, it will be desirable
to point out some general features of the experiments which indicate the form
which these laws must assume.

Returning to experiment (1), in which a neutral gray was produced from
red, blue, and green, we may observe, that, while the adjustments were incom-

EXPERIMENTS ON COLOUK, AS PERCEIVED BY THE EYE. 131

plete, the difference of the tints could be detected only by one circle appearing
more red, more green, or more blue than the other, or by being lighter or
darker, that is, having an excess or defect of all the three colours together.
Hence it appears that the nature of a colour may be considered as dependent
on three things, as, for instance, redness, blueness, and greenness. This is con-
firmed by the fact that any tint may be imitated by mixing red, blue, and
green alone, provided that tint does not exceed a certain brilliancy.

Another way of shewing that colour depends on three things is by con-
sidering how two tints, say two lilacs, may differ. In the first place, one may
be lighter or darker than the other, that is, the tints may differ in shade.
Secondly, one may be more blue or more red than the other, that is, they may
differ in hue. Thirdly, one may be more or less decided in its colour ; it may vary
from purity on the one hand, to neutrality on the other. This is sometimes
expressed by saying that they may differ in tint.

Thus, in shade, hue, and tint, we have another mode of reducing the
elements of colour to three. It will be shewn that these two methods of con-
sidering colour may be deduced one from the other, and are capable of exact
numerical comparison.

On a Geographical Method of Exhibiting the Relations of Colours.

The method which exhibits to the eye most clearly the results of this theory
of the three elements of colour, is that which supposes each colour to be repre-
sented by a point in space, whose distances from three co-ordinate planes are
proportional to the three elements of colour. But as any method by which the
operations are confined to a plane is preferable to one requiring space of three
dimensions, we shall only consider for the present that which has been adopted
for convenience, founded on Newton's Circle of colours and Mayer and Young's
Triangle.

Vermilion, ultramarine, and emerald-green, being taken (for convenience) as-
standard colours, are conceived to be represented by three points, taken (for con-
venience) at the angles of an equilateral triangle. Any colour compounded of
these three is to be represented by a point found by conceiving masses propor-
tional to the several components of the colour placed at their respective angular
points, and taking the centre of gravity of the three masses. In this way, each

17—2

132 1 XPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

colour will indicate by its position the proportions of the elements of which it is
composed. The total intensity of the colour is to be measured by the whole
number of divisions of V, U, and EG, of which it is composed. This may be
indicated by a number or coefficient appended to the name of the colour, by
which the number of divisions it occupies must be multiplied to obtain its mass
in calculating the results of new combinations.

This will be best explained by an example on the diagram (No. 1). We
have, by experiment (l),

•37 V + -27 U + -36 EG = -28 SW + 72 Bk.

To find the position of the resultant neutral tint, we must conceive a mass
of '37 at V, of '27 at U, and of '36 at EG, and find the centre of gravity.
This may be done by taking the line UV, and dividing it in the proportion of
•37 to '27 at the point a, where

aV : aU :: '27 : '37.

Then, joining a with EG, divide the joining line in W in the proportion of '3(»
to ('37 + '27), W will be the position of the neutral tint required, which is not
white, but 0'28 of white, diluted with 0'72 of black, which has hardly any effect
whatever, except in decreasing the amount of the other colour. The total in-
tensity of our white paper will be represented by ^ = 3*57; so that, whenever
white enters into an equation, the number of divisions must be multiplied by
the coefficient 3 '5 7 before any true results can be obtained.

We may take, as the next example, the method of representing the relation
of pale chrome to the standard colours on our diagram, by making use of ex-
periment (2), in which pale chrome, ultramarine, and emerald-green, produced a
neutral gray. The resulting equation was

•33PC + -55U + -12EG = '37SW + -63Bk (2).

In order to obtain the total intensity of white, we must multiply the
number of divisions, '37, by the proper coefficient, which is 3'57. The result is
T32, which therefore measures the total intensity on both sides of the equation.

Subtracting the intensity of '55 U + '12 EG, or '67 from T32, we obtain '65
as the corrected value of '3 3 PC. It will be convenient to use these corrected
values of the different colours, taking care to distinguish them by small initials
instead of capitals.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 133

Equation (2) then becomes

•65 pc + '55 U + -12 EG= 1-32 w.
Hence pc must be situated at a point such that w is the centre of gravity

of -

To find it, we begin by determining ft the centre of gravity of -55 U + '12EG,
then, joining fiw, the point we are seeking must lie at a certain distance on
the other side of w from c. This distance may be found from the proportion,

•65 : (-55 + -12) :: /3w : w pc,

which determines the position of pc. The proper coefficient, by which the ob-
served values of PC must be corrected, is ||, or 1'97.

We have thus determined the position and coefficient of a colour by a single
experiment, in which it was made to produce a neutral tint along with two of
the standard colours. As this may be done with every possible colour, the
method is applicable wherever we can obtain a disc of the proposed colour. In
this way the diagram (No. l) has been laid down from observations made in
daylight, by a good eye of the ordinary type.

It has been observed that experiments, in which the resultant tint is neutral,
are more accurate than those in which the resulting tint has a decided colour,
as in experiment (3), owing to the effects of accidental colours produced in the
eye in the latter case. These experiments, however, may be repeated till a
very good mean result has been obtained.

But since the elements of every colour have been already fixed by our
previous observations and calculations, the agreement of these results with those
calculated from the diagram forms a test of the correctness of our method.

By experiment (No. 3), made at the same time with (l) and (2), we have

'39 PC + '21 U + '40Bk = '59 V + '41 EG (3).

Now, joining U with pc, and V with EG, the only common point is that
at which they cross, namely y.

Measuring the parts of the line V EG, we find them in the proportion of

•58V and '42 EG = 1 '00 y.

Similarly, the line U pc is divided in the proportion

78 pc and -22U=l-OOy.

134 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE,

But 78 pc must be divided by 1'97, to reduce it to PC, as was previously
explained. The result of calculation is, therefore,

•39 PC + '22 U + '39 Bk = '58 V + "42 EG,
the black being introduced simply to fill up the circle.

This result differs very little from that of experiment (3), and it must be
recollected that these are single experiments, made independently of theory, and
chosen at random.

Experiments made at Cambridge, with all the combinations of five colours,
shew that theory agrees with calculation always within 0-012 of the whole,
and sometimes within 0'002. By the repetition of these experiments at the
numerous opportunities which present themselves, the accuracy of the results
may be rendered still greater. As it is, I am not aware that the judgments
of the human eye with respect to colour have been supposed capable of so
severe a test.

Further consideration of the Diagram of Colours,

We have seen how the composition of any tint, in terms of our three
standard colours, determines its position on the diagram and its proper coefficient.
In the same way, the result of mixing any other colours, situated at other
points of the diagram, is to be found by taking the centre of gravity of their
reduced masses, as was done in the last calculation (experiment 3).

We have now to turn our attention to the general aspect of the diagram.

The standard colours, V, U, and EG, occupy the angles of an equilateral
triangle, and the rest are arranged in the order in which they participate in
red, blue, and green, the neutral tint being at the point w within the triangle.
If we now draw lines through w to the different colours ranged round it, we
shall find that, if we pass from one line to another in the order in which they
lie from red to green, and through blue back again to red, the order will be —

Coefficient. Coefficient.

Carmine ..... 0'4

Vermilion . . . . . 1 0

Red Lead 1-3

Orange Orpiment . . . 1-0

Orange Chrome . . . 1 -6

Chrome Yellow . . . 1-5

Gamboge 1-8

Pale Chrome .... 2-0

Mixed Green (U C) ... 0-4

Brunswick Green . . . 0'2

Emerald Green . . . . I/O

Verditer Blue .... 0-8

Prussian Blue . . . . " 1

Ultramarine 1 '0

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 135

It may be easily seen that this arrangement of the colours corresponds to
that of the prismatic spectrum ; the only difference being that the spectrum
is deficient in those fine purples which lie between ultramarine and vermilion,
and which are easily produced by mixture. The experiments necessary for deter-
mining the exact relation of this list to the lines in the spectrum are not yet
completed.

If we examine the colours represented by different points in one of these
lines through w, we shall find the purest and most decided colours at its outer
extremity, and the faint tints approaching to neutrality nearer to w.

If we also study the coefficients attached to each colour, we shall find that
the brighter and more luminous colours have higher numbers for their coefficients
than those which are dark.

In this way, the qualities which we have already distinguished as hue, tint,
and shade, are represented on the diagram by angular position with respect to w,
distance from w, and coefficient ; and the relation between the two methods of
reducing the elements of colour to three becomes a matter of geometry.

Theory of the Perception of Colour.

Opticians have long been divided on this point ; those who trusted to
popular notions and their own impressions adopting some theory of three primary
colours, while those who studied the phenomena of light itself proved that no
such theory could explain the constitution of the spectrum. Newton, who was
the first to demonstrate the actual existence of a series of kinds of light,
countless in number, yet all perfectly distinct, was also the first to propound
a method of calculating the effect of the mixture of various coloured light ;
and this method was substantially the same as that which we have just
verified. It is true, that the directions which he gives for the construction
of his circle of colours are somewhat arbitrary, being probably only intended
as an indication of the general nature of the method, but the method itself
is mathematically reducible to the theory of three elements of the colour-
sensation*.

* See Note III. For a confirmation of Newton's analysis of Light, see Helmholtz, Pogg. Ann.
1852; and Phil. May. 1852, Part ir.

136 ExriiuiMKXTS ox COLOUR, AS PERCEIVED BY THE EYE.

Young, who made the next great step in the establishment of the theory
of light, seems also to have been the first to follow out the necessary conse-
quences of Newton's suggestion on the mixture of colours. He saw that, since
this triplicity has no foundation in the theory of light, its cause must be looked
for in the constitution of the eye; and, by one of those bold assumptions
which sometimes express the result of speculation better than any cautious
trains of reasoning, he attributed it to the existence of three distinct modes
of sensation in the retina, each of which he supposed to be produced in different
degrees by the different rays. These three elementary effects, according to his
view, correspond to the three sensations of red, green, and violet, and would
separately convey to the sensorium the sensation of a red, a green, and a violet
picture ; so that by the superposition of these pictures, the actual variegated
world is represented*.

In order fully to understand Young's theory, the function which he
attributes to each system of nerves must be carefully borne in mind. Each nerve
acts, not, as some have thought, by conveying to the mind the knowledge of the
length of an undulation of light, or of its periodic time, but simply by being
more or less affected by the rays which fall on it. The sensation of each
elementary nerve is capable only of increase and diminution, and of no other
change. We must also observe, that the nerves corresponding to the red
sensation are affected chiefly by the red rays, but in some degree also by those
of every other part of the spectrum ; just as red glass transmits red rays freely,
but also suffers those of other colours to pass in smaller quantity.

This theory of colour may be illustrated by a supposed case taken from
the art of photography. Let it be required to ascertain the colours of a land-
scape, by means of impressions taken on a preparation equally sensitive to rays of
every colour.

Let a plate of red glass be placed before the camera, and an impression
taken. The positive of this will be transparent wherever the red light has been
abundant in the landscape, and opaque where it has been wanting. Let it now
be put in a magic lantern, along with the red glass, and a red picture will be
thrown on the screen.

Let this operation be repeated with a green and a violet glass, and, by

* Young's Lectures, p. 345, Kelland's Edition. See also Helmholtz's statement of Young's Theory,
in his Paper referred to in Note I. ; and Herechel's Light, Art 518.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 137

means of three magic lanterns, let the three images be superimposed on the
screen. The colour of any point on the screen will then depend on that of the
corresponding point of the landscape ; and, by properly adjusting the intensities
of the lights, &c., a complete copy of the landscape, as far as visible colour is
concerned, will be thrown on the screen. The only apparent difference will be,
that the copy will be more subdued, or less pure in tint, than the original.
Here, however, we have the process performed twice — first on the screen, and
then on the retina.

This illustration will shew how the functions which Young attributes to the
three systems of nerves may be imitated by optical apparatus. It is therefore
unnecessary to search for any direct connection between the lengths of the
undulations of the various rays of light and the sensations as felt by us, as
the threefold partition of the properties of light may be effected by physical
means. The remarkable correspondence between the results of experiments on
different individuals would indicate some anatomical contrivance identical in all.
As there is little hope of detecting it by dissection, we may be content at
present with any subsid'ary evidence which we may possess. Such evidence is
furnished by those individuals who have the defect of vision which was
described by Dalton, and which is a variety of that which Dr G. Wilson has
lately investigated, under the name of Colour-Blindness.

Testimony of the Colour-Blind with respect to Colour.

Dr George Wilson has described a great number of cases of colour-
blindness, some of which involve a general indistinctness in the appreciation
of colour, while in others, the errors of judgment are plainly more numerous
in those colours which approach to red and green, than among those which
approach to blue and yellow. In these more definite cases of colour-blindness,
the phenomena can be tolerably well accounted for by the hypothesis of an
insensibility to red light ; and this is, to a certain extent, confirmed by the
fact, that red objects appear to these eyes decidedly more obscure than to
ordinary eyes. But by experiments made with the pure spectrum, it appears
that though the red appears much more obscure than other colours, it is not
wholly invisible, and, what is more curious, resembles the green more than
any other colour. The spectrum to them appears faintly luminous in the red;

VOL. i. 18

138 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

bright yellow from orange to yellow, bright but not coloured from yellow-
green to blue, and then strongly coloured hi the extreme blue and violet,
after which it seems to approach the neutral obscure tint of the red. It is
not easy to see why an insensibility to red rays should deprive the green
rays, which have no optical connection with them, of their distinctive appearance.
The phenomena seem rather to lead to the conclusion that it is the red
sensation which is wanting, that is, that supposed system of nerves which is
affected in various degrees by all light, but chiefly by red. We have fortunately
the means of testing this hypothesis by numerical results.

Of the subjects of my experiments at Cambridge, four were decided cases
of colour-blindness. Of these two, namely, Mr R and Mr S., were not
sufficiently critical in their observations to afford any results consistent within
10 divisions of the colour- top. The remaining two, Mr N. and Mr X., were
as consistent in their observations as any persons of ordinary vision can be,
while the results shewed all the more clearly how completely their sensations
must differ from ours.

The method of experimenting was the same as that adopted with ordinary
eyes, except that in these cases the operator can hardly influence the result
by yielding to his own impressions, as he has no perception whatever of the
similarity of the two tints as seen by the observer. The questions which he
must ask are two, Which circle appears most blue or yellow ? Which appears
lightest and which darkest ? By means of the answers to these questions he
must adjust the resulting tints to equality in these respects as it appears to
the observer, and then ascertain that these tints now present no difference of
colour whatever to his eyes. The equations thus obtained do not require five
colours including black, but four only. For instance, the mean of several obser-
vations gives —

•19 G + -05 B+76 Bk=100R (4).

[In these experiments R, B, G, Y, stand for red, blue, green, and yellow
papers prepared by Mr D. R. Hay. I am not certain that they are identical
with his standard colours, but I believe so. Their relation to vermilion, ultra-
marine, and emerald-green is given in diagram (l). Their relations to each other
are very accurately given in diagram (2).]

It appears, then, that the dark blue-green of the left side of the equation
is equivalent to the full red of the right side.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 139

Hence, if we divide the line BG in the proportion 19 to 5 at the point /8,
and join R& the tint at /3 will differ from that at R (to the colour-blind)
only in being more brilliant in the proportion of 100 to 24, and all inter-
mediate tints on the line R/3 will appear to them of the same hue, but
of intermediate intensities.

Now, if we take a point D, so that RD is to R/3 in the proportion of
24 to 100 — 24, or 76, the tint of D, if producible, should be invisible to
the colour-blind. D, therefore, represents the pure sensation which is unknown
to the colour-blind, and the addition of this sensation to any others cannot
alter it in their estimation. It is for them equivalent to black.

Hence, if we draw lines through D in different directions, the colours
belonging to any line ought to differ only in intensity as seen by them, so
that one of them may be reduced to the other by the addition of black
only. If we draw DW and produce it, all colours on the upper side of DW
will be varieties of blue, and those on the under side varieties of yellow, so
that the line DW is a boundary line between their two kinds of colour, blue
and yellow being the names by which they call them.

The accuracy of this theory will be evident from the comparison of the
experiments which I had an opportunity of making on Mr N. and Mr X. with
each other, and with measurements taken from the diagram No. 2, which was
constructed from the observations of ordinary eyes only, the point D alone
being ascertained from a series of observations by Mr N.

Taking the point y, between R and B, it appears, by measurement of the
lines Ry and By, that y corresponds to

•07B + -93R.

By measurement of Wy and Dy, and correction by means of the coefficient
of W, and calling D black in the colour-blind language, y corresponds to

•105 W + -895 Bk.

Therefore

By measurement -93 R+ '07 B = '105 W + '895 Bk

(5).

By observation N. & X. together '94 R+-Q6 B = '10 W + '90 Bk
By X. alone '93 R + -07 B = '10 W + -90 Bk

The agreement here is as near as can be expected.

18—2

140

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

By a similar calculation with respect to the point 8, between B and G,

By measurement ............... '43 B + '57 G = '335 W + '665 Bk

Observed by N. and X ....... '41 B + -59 G = '34 W + -66 Bk ...... (6).

By X. alone ..................... '42 B+-58 G = '32 W + -68 Bk

We may also observe, that the line GD crosses RY. At the point of inter-
section we have —

By calculation ........................ -87 R+'13

Observed by N. and X .......... '86 R-K14

X .......... -84R+-16

X. . -90R + -10

= '34 G + -66 Bk 1
= '40 G + -60 Bk
= -31 G + -69 Bk
= -27 G + 73 Bk

(7).

Here observations are at variance, owing to the decided colours produced
affecting the state of the retina, but the mean agrees well with calculation.

Drawing the line BY, we find that it cuts lines through D drawn to every
colour. Hence all colours appear to the colour-blind as if composed of blue
and yellow. By measurement on the diagram, we find for red

Measured -138 Y + -123 B + 749 Bk = 100 R"

Observed by N.... -15 Y+'ll B+74 Bk = 100R[ (8).

X....-13 Y + -11 B + 76 Bk = 100R

For green we have in the same way —
Measured ...... 705 Y + -295

Observed by N.... 70 Y + -30

X.... 70 Y + -30

= '95 G + '05 Bk
= '86G + '14Bk
= -83 G + -17 Bk

.(9).

For white —

Measured ...... -407 Y + '593 B = '326 W + '674 Bk

Observed by N.... -40 Y+'60 B = '33 W+'67 Bk

X.... -44 Y+-56 B = '33 W+-G7 Bk

The accuracy of these results shews that, whether the hypothesis of the
want of one element out of three necessary to perfect vision be actually true
or not, it affords a most trustworthy foundation on which to build a theory
of colour-blindness, as it expresses completely the observed facts of the case.
They also furnish us with a datum for our theory of perfect vision, namely,
the point D, which points out the exact nature of the colour-sensation, which
must be added to the colour-blind eye to render it perfect. I am not aware

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 141

of any method of determining by a legitimate process the nature of the other
two sensations, although Young's reasons for adopting something like green and
violet appear to me worthy of attention.

The only remaining subject to which I would call the attention of the
Society is the effect of coloured glasses on the colour-blind. Although they can-
not distinguish reds and greens from varieties of gray, the transparency of red
and green glasses for those kinds of light is very different. Hence, after finding
a case such as that in equation (4), in which a red and a green appear iden-
tical, on looking through a red glass they see the red clearly and the green
obscurely, while through a green glass the red appears dark and the green light.

By furnishing Mr X. with a red and a green glass, which he could dis-
tinguish only by their shape, I enabled him to make judgments in previously
doubtful cases of colour with perfect certainty. I have since had a pair of
spectacles constructed with one eye-glass red and the other green. These Mr X.
intends to use for a length of time, and he hopes to acquire the habit of discri-
minating red from green tints by their different effects on his two eyes. Though
he can never acquire our sensation of red, he may then discern for himself what
things are red, and the mental process may become so familiar to him as to act
unconsciously like a new sense.

In one experiment, after looking at a bright light, with a red glass over one
eye and a green over the other, the two tints in experiment (4) appeared to him
altered, so that the outer circle was lighter according to one eye, and the inner
according to the other. As far as I could ascertain, it appeared as if the eye
which had used the red glass saw the red circle brightest. This result, which
seems at variance with what might be expected, I have had no opportunity of
verifying.

This paper is already longer than was originally intended. For further
information I would refer the reader to Newton's Opticks, Book i. Part n., to
Young's Lectures on Natural Philosophy, page 345, to Mr D. R Hay's works on
Colours, and to Professor Forbes on the "Classification of Colours" (Phil, Mag.,
March, 1849).

The most remarkable paper on the subject is that of M. Helmholtz, in the
Philosophical Magazine for 1852, in which he discusses the different theories of
primary colours, and describes his method of mixing the colours of the spectrum.
An examination of the results of M. Helmholtz with reference to the theory

142 KXI'I'IUMENTS ON COLOUR, AS PKKCKIVED BY THE EYE.

of three elements of colour, by Professor Grassmann, is translated in the Phil.
Mag., April, 1854.

References to authors on colour-blindness are given in Dr G. Wilson's papers
on that subject. A valuable Letter of Sir J. F. W. Herechel to Dalton on his
peculiarity of vision, is to be found in the Life of Dalton by Dr Henry.

I had intended to describe some experiments on the propriety of the method
of mixing colours by rotation, which might serve as an extension of Mr Swan's
experiments on instantaneous impressions on the eye. These, together with the
explanation of some phenomena which seem to be at variance with the theory of
vision here adopted, must be deferred for the present. On some future occasion,
I hope to be able to connect these simple experiments on the colours of pigments
with others in which the pure hues of the spectrum are used. I have already
constructed a model of apparatus for this purpose, and the results obtained are
sufficiently remarkable to encourage perseverance.

NOTE I.
On different Methods of Exhibiting the Mixtures of Colours.

(l) Mechanical Mixture of Coloured Powder*.

By grinding coloured powders together, the differently-coloured particles may
be so intermingled that the eye cannot distinguish the colours of the separate
powders, but receives the impression of a uniform tint, depending on the nature
and proportions of the pigments used. In this way, Newton mixed the powders
of orpiment, purple, bise, and viride cms, so as to form a gray, which, in sun-
light, resembled white paper in the shade. (Newton's Ojtticks, Book I. Part n.,
Exp. xv.) This method of mixture, besides being adopted by all painters, has
been employed by optical writers as a means of obtaining numerical results.
The specimens of such mixtures given by D. R. Hay in his works on Colour,
and the experiments of Professor J. D. Forbes on the same subject, shew the
importance of the method as a means of classifying colours. There are two
objections, however, to this method of exhibiting colours to the eye. When
two powders of unequal fineness are mixed, the particles of the finer powder
cover over those of the coarser, so as to produce more than their due effect
in influencing the resultant tint. For instance, a small quantity of lamp-black,

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 143

mixed with a large quantity of chalk, will produce a mixture which is nearly
black. Although the powders generally used are not so different in this respect
as lamp-black and chalk, the results of mixing given weights of any coloured
powders must be greatly modified by the mode in which these powders have
been prepared.

Again, the light which reaches the eye from the surface of the mixed pow-
ders consists partly of light which has fallen on one of the substances mixed
without being modified by the other, and partly of light which, by repeated
reflection or transmission, has been acted on by both substances. The colour of
these rays will not be a mixture of those of the substances, but will be the
result of the absorption due to both substances successively. Thus, a mixture of
yellow and blue produces a neutral tint tending towards red, but the remainder
of white light, after passing through both, is green ; and this green is generally
sufficiently powerful to overpower the reddish gray due to ^he separate colours
of the substances mixed. This curious result has been ably investigated by
Professor Helmholtz of Konigsberg, in his Memoir on the Theory of Compound
Colours, a translation of which may be found in the Annals of Philosophy for
1852, Part 2.

(2) Mixture of differently-coloured Beams of Light by Superposition

on an Opaque Screen.

When we can obtain light of sufficient intensity, this method produces the
most beautiful results. The best series of experiments of this kind are to be
found in Newton's Opticks, Book I. Part n. The different arrangements for
mixing the rays of the spectrum on a screen, as described by Newton, form
a very complete system of combinations of lenses and prisms, by which almost
every possible modification of coloured light may be produced. The principal
objections to the use of thia method are — (1) The difficulty of obtaining a con-
stant supply of uniformly intense light ; (2) The uncertainty of the effect of
the position of the screen with respect to the incident beams and the eye of
the observer ; (3) The possible change in the colour of the incident light due
to the fluorescence of the substance of the screen. Professor Stokes has found
that many substances, when illuminated by homogeneous light of one refrangi-
bility, become themselves luminous, so as to emit light of lower refrangibility.
This phenomenon must be carefully attended to when screens are used to exhibit
light.

144 EXPERIMENTS OX COLOUR, AS PERCEIVED BY THE EYK.

(3) Union of Coloured Beams by a Prism so as to form one Beam.

The mode of viewing the beam of light directly, without first throwing it
on a screen, was not much used by the older experimenters, but it possesses
the advantage of saving much light, and admits of examining the rays before
they have been stopped in aiiy way. In Newton's llth proposition of the 2nd
Book, an experiment is described, in which a beam is analysed by a prism,
concentrated by a lens, and recombined by another prism, so as to form a beam
of white light similar to the incident beam. By stopping the coloured rays at
the lens, any proposed combination may be made to pass into the emergent
beam, where it may be received directly by the eye, or on a screen, at pleasure.

The experiments of Helmholtz on the colours of the spectrum were made
with the ordinary apparatus for directly viewing the pure spectrum, two oblique
slits crossing one another being employed to admit the light instead of one
vertical slit. Two pure spectra were then seen crossing each other, and so
exhibiting at once a large number of combinations. The proportions of these
combinations were altered by varying the inclination of the slits to the plane of
refraction, and in this way a number of very remarkable results were obtained, —
for which see his Memoir, before referred to.

In experiments of the same kind made by myself in August 1852 (inde-
pendently of M. Helmholtz), I used a combination of three moveable vertical
slits to admit the light, instead of two cross slits, and observed the compound
ray through a slit made in a screen on which the pure spectrum is formed.
In this way a considerable field of view was filled with the mixed light, and
might be compared with another part of the field illuminated by light proceeding
from a second system of slits, placed below the first set. The general character
of the results agreed with those of M. Helmholtz. The chief difficulties seemed
to arise from the defects of the optical apparatus of my own eye, which ren-
dered apparent the compound nature of the light, by analysing it as a prism
or an ordinary lens would do, whenever the lights mixed differed much in
refrangibility.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 145

(4) Union of two beams by means of a transparent surface, which reflects

the first and transmits the second.

The simplest experiment of this kind is described by M. Helmholtz. He
places two coloured wafers on a table, and then, taking a piece of transparent
glass, he places it between them, so that the reflected image of one apparently
coincides with the other as seen through the glass. The colours are thus mixed,
and, by varying the angle of reflection, the relative intensities of the reflected
and transmitted beams may be varied at pleasure.

In an instrument constructed by myself for photometrical purposes two re-
flecting plates were used. They were placed in a square tube, so as to polarize
the incident light, which entered through holes in the sides of the tubes, and
was reflected in the direction of the axis. In this way two beams oppositely
polarized were mixed, either of which could be coloured in any way by coloured
glasses placed over the holes in the tube. By means of a Nicol's prism placed
at the end of the tube, the relative intensities of the two colours as they
entered the eye could be altered at pleasure.

(5) Union of two coloured beams by means of a doubly-refracting Prism.

I am not aware that this method has been tried, although the opposite
polarization of the emergent rays is favourable to the variation of the experiment.

(6) Successive presentation of the different Colours to the Retina.

It has long been known, that light does not produce its full effect on the
eye at once, and that the effect, when produced, remains visible for some time
after the light has ceased to act. In the case of the rotating disc, the various
colours become indistinguishable, and the disc appears of a uniform tint, which
is in some sense the resultant of the colours so blended. This method of com-
bining colours has been used since the time of Newton, to exhibit the results
of theory. The experiments of Professor J. D. Forbes, which I witnessed in
1849, first encouraged me to think that the laws of this kind of mixture might
be discovered by special experiments. After repeating the well-known experiment
in which a series of colours representing those of the spectrum are combined

VOL. i. 19

146 EXPERIMENTS ON COLOUR, AB PERCEIVED BY THE EYE.

to form gray, Professor Forbes endeavoured to form a neutral tint, by the
combination of three colours only. For this purpose, he combined the three
so-called primary colours, red, blue, and yellow, but the resulting tint could
not be rendered neutral by any combination of these colours; and the reason
was found to be, that blue and yellow do not make green, but a pinkish tint,
when neither prevails in the combination. It was plain, that no addition of
red to this, could produce a neutral tint.

This result of mixing blue and yellow was, I believe, not previously known.
It directly contradicted the received theory of colours, and seemed to be at
variance with the fact, that the same blue and yellow paint, when ground
together, do make green. Several experiments were proposed by Professor Forbes,
in order to eliminate the effect of motion, but he was not then able to under-
take them. One of these consisted in viewing alternate stripes of blue and
yellow, with a telescope out of focus. I have tried this, and find the resultant
tint pink as before*. I also found that the beams of light coloured by trans-
mission through blue and yellow glasses appeared pink, when mixed on a screen,
while a beam of light, after passing through both glasses, appeared green. By
the help of the theory of absorption, given by Herschelt, I made out the
complete explanation of this phenomenon. Those of pigments were, I think, first
explained by Helmholtz in the manner above referred toj.

It may still be asked, whether the effect of successive presentation to the
eye is identical with that of simultaneous presentation, for if there is any action
of the one kind of light on the other, it can take place only in the case of
simultaneous presentation. An experiment tending to settle this point is recorded
by Newton (Book I. Part n., Exp. 10). He used a comb with large teeth to
intercept various rays of the spectrum. When it was moved slowly, the various
colours could be perceived, but when the speed was increased the result was
jierfect whiteness. For another form of this experiment, see Newton's Sixth
Letter to Oldenburg (Horsley's Edition, Vol. IV., page 335).

In order more fully to satisfy myself on this subject, I took a disc in
which were cut a number of slits, so as to divide it into spokes. In a plane,
nearly passing through the axis of this disc, I placed a blue glass, so that one

* See however Encyc. Metropolilana, Art. "Light," section 502. t Ib. sect. 516.

t I have lately seen a passage in Moigno's Cosmos, stating that M. Plateau, in 1819, had obtain^!
gray by whirling together gamboge and Prussian blue. Corresjxmdance Jfatft. et Phys. de M. Quetelet,
Vol. v., p. 221.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 147

half of the disc might be seen by transmitted light — blue, and the other by
reflected light — white. In the course of the reflected light I placed a yellow
glass, and in this way I had two nearly coincident images of the slits, one
yellow and one blue. By turning the disc slowly, I observed that in some
parts the yellow slits and the blue slits appeared to pass over the field alter-
nately, while in others they appeared superimposed, so as to produce alternately
tfoeir mixture, which was pale pink, and complete darkness. As long as the
disc moved slowly I could perceive this, but when the speed became great, the
whole field appeared uniformly coloured pink, so that those parts in which the
colours were seen successively were indistinguishable from those in which they
were presented together to the eye.

Another form in which the experiment may be tried requires only the
colour-top above described. The disc should be covered with alternate sectors
of any two colours, say red and green, disposed alternately in four quadrants.
By placing a piece of glass above the top, in the plane of the axis, we make
the image of one half seen by reflection coincide with that of the other seen
by transmission. It will then be seen that, in the diameters of the top which
are parallel and perpendicular to the plane of reflection, the transmitted green
coincides with the reflected green, and the transmitted red with the reflected
red, so that the result is always either pure red or pure green. But in the
diameters intermediate to these, the transmitted red coincides with the reflected
green, and vice versa, so that the pure colours are never seen, but only their
mixtures. As long as the top is spun slowly, these parts of the disc will
appear more steady in colour than those in which the greatest alternations
take place ; but when the speed is sufficiently increased, the disc appears per-
fectly uniform in colour. From these experiments it appears, that the apparent
mixture of colours is not due to a mechanical superposition of vibrations, or
to any mutual action of the mixed rays, but to some cause residing in the
constitution of the apparatus of vision.

(7) Presentation of the Colours to be mixed one to each Eye.

This method is said not to succeed with some people ; but I have always
found that the mixture of colours was perfect, although it was difficult to con-
ceive the objects seen by the two eyes as identical. In using the spectacles,

19—2

148 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

of which one eye is green and the other red, I have found, when looking at
an arrangement of green and red papers, that some looked metallic and others
transparent. This arises from the very different relations of brightness of the
two colours as seen by each eye through the spectacles, which suggests the false
conclusion, that these differences are the result of reflection from a polished
surface, or of light transmitted through a clear one.

NOTE IT.
liesults of Experiments with Mr Hay's Papers at Cambridge, November, 1854.

The mean of ten observations made by six observers gave

•449R+-299 G + -252 B='224 W+ 776 Bk (1).

•696 R+-304 G = '181 B+-327 Y + '492 Bk (2).

These two equations served to determine the positions of white and yellow
in diagram No. 2. The coefficient of W is 4 '447, and that of yellow 2'506.

From these data we may deduce three other equations, either by calcu-
lation, or by measurement on the diagram (No. 2).

Eliminating green from the equations, we find

•565 B + -435 Y = -307 R+'304 W + -389 Bk (3).

The mean of three observations by three different observers gives

•573 B + -477 Y='313 R + -297 W+'390Bk.
Errors of calculation - -008 B + '008 Y - '006 R + '007 W - '001 Bk.

The point on the diagram to which this equation corresponds is the intersec-
tion of the lines BY and RW, and the resultant tint is a pinkish -gray.

Eliminating red from the equations, we obtain

Calculation -533 B + '150 G + -317 Y = '337 W + -663 Bk

By 10 observations '537 B+'146 G+'317 Y='337 W + -663 Bk
Errors -'004 +'004

K 1 in i i nating blue '660 R + -340 G = '218 Y + '108 W + -682 Bk"

' (5).

By 5 observations '672 R + -328 G = '224 Y + '094 W + '672 Bk
Errors -'012 +'012 -'006 +'014 +'008

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 149

NOTE III.
On the Theory of Compound Colours.

Newton's theorem on the mixture of colours is to be found in his Opticks,
Book i., Part IL, Prop. vi.

In a mixture of primary colours, the quantity and quality of each, being
given, to know the colour of the compound.

He divides the circumference of a circle into parts proportional to the seven
musical intervals, in accordance with his opinion of the divisions of the spectrum.
He then conceives the colours of the spectrum arranged round the circle, and at
the centre of gravity of each of the seven arcs he places a little circle, the
area of which represents the number of rays of the corresponding colour which
enter into the given mixture. He takes the centre of gravity of all these circles
to represent the colour formed by the mixture. The hue is determined by
drawing a line through the centre of the circle and this point to the circum-
ference. The position of this line points out the colour of the spectrum which
the mixture most resembles, and the distance of the resultant tint from the
centre determines the fulness of its colour.

Newton, by this construction (for which he gives no reasons), plainly shews
that he considered it possible to find a place within his circle for every possible
colour, and that the entire nature of any compound colour may be known from
its place in the circle. It will be seen that the same colour may be compounded
from the colours of the spectrum in an infinite variety of ways. The apparent
identity of all these mixtures, which are optically different, as may be shewn by
the prism, implies some law of vision not explicitly stated by Newton. This
law, if Newton's method be true, must be that which I have endeavoured to
establish, namely, the threefold nature of sensible colour.

With respect to Newton's construction, we now know that the proportions
of the colours of the spectrum vary with the nature of the refracting medium.
The only absolute index of the kind of light is the time of its vibration. The
length of its vibration depends on the medium in which it is ; and if any pro-
portions are to be sought among the wave-lengths of the colours, they must
be determined for those tissues of the eye in which their physical effects are

150 \PERIMENTS ON COLOUR, AS PERCEIVED BY THE I:

supposed to terminate. It may be remarked, that the apparent colour of the
spectrum changes most rapidly at three points, which lie respectively in the
yellow, between blue and green, and between violet and blue. The wave-lengths
of the corresponding rays in »W<r are in the proportions of three geometric
means between 1 and 2 very nearly. This result, however, is not to be con-
sidered established, unless confirmed by better observations than mine.

The only safe method of completing Newton's construction is by an exami-
nation of the colours of the spectrum and their mixtures, and subsequent
calculation by the method used in the experiments with coloured papers. In
this way I hope to determine the relative positions in the colour-diagram of
every ray of the spectrum, and its relative intensity in the solar light. The
spectrum will then form a curve not necessarily circular or even re-entrant, and
its peculiarities so ascertained may form the foundation of a more complete
theory of the colour-sensation.

On the relation of tlte pure rays of the Spectrum to the three assumed Elementnrij

Sensations.

If we place the three elementary colour-sensations (which we may call, after
Young, red, green, and violet) at the angles of a triangle, all colours which
the eye can possibly perceive (whether by the action of light, or by pressure,
disease, or imagination) must be somewhere within this triangle, those which lie
farthest from the centre being the fullest and purest colours. Hence the colours
which lie at the middle of the sides are the purest of their kind which the
eye can see, although not so pure as the elementary sensations.

It is natural to suppose that the pure red, green, and violet rays of the
spectrum produce the sensations which bear their names in the highest purity.
But from this supposition it would follow that the yellow, composed of the red
and green of the spectrum, would be the most intense yellow possible, while
it is the result of experiment, that the yellow of the spectrum itself is much
more full in colour. Hence the sensations produced by the pure red and green
rays of the spectrum are not the pure sensations of our theory. Newton has
remarked, that nb two colours of the spectrum produce, when mixed, a colour
equal in fulness to the intermediate colour. The colours of the spectrum are
all more intense than any compound ones. Purple is the only colour which

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE. 151

must be produced by combination. The experiments of Helmholtz lead to the
same conclusion ; and hence it would appear that we can find no part of the
spectrum which produces a pure sensation.

An additional, though less satisfactory evidence of this, is supplied by the
observation of the colours of the spectrum when excessively bright. They then
appear to lose their peculiar colour, and to merge into pure whiteness. This
is probably due to the want of capacity of the organ to take in so strong an
impression ; one sensation becomes first saturated, and the other two speedily
follow it, the final effect being simple brightness.

From these facts I would conclude, that every ray of the spectrum is capable
of producing all three pure sensations, though in different degrees. The curve,
therefore, which we have supposed to represent the spectrum will be quite within
the triangle of colour. All natural or artificial colours, being compounded of
the colours of the spectrum, must lie within this curve, and, therefore, the colours
corresponding to those parts of the triangle beyond this curve must be for ever
unknown to us. The determination of the exact nature of the pure sensations,
or of their relation to ordinary colours, is therefore impossible, unless we can
prevent them from interfering with each other as they do. It may be possible
to experience sensations more pure than those directly produced by the spec-
trum, by first exhausting the sensibility to one colour by protracted gazing, and
then suddenly turning to its opposite. But if, as I suspect, colour-blindness be
due to the absence of one of these sensations, then the point D in diagram (2),
which indicates their absent sensation, indicates also our pure sensation, which
we may call red, but which we can never experience, because all kinds of
light excite the other sensations.

Newton has stated one objection to his theory, as follows: — "Also, if only
two of the primary colours, which in the circle are opposite to one another, be
mixed in an equal proportion, the point Z" (the resultant tint) "shall fall upon
the centre 0 " (neutral tint) ; " and yet the colour compounded of these two shall
not be perfectly white, but some faint anonymous colour. For I could never yet, by
mixing only two primary colours, produce a perfect white." This is confirmed by
the experiments of Helmholtz ; who, however, has succeeded better with some
pairs of colours than with others.

In my experiments on the spectrum, I came to the same result ; but it
appeared to me that the very peculiar appearance of the neutral tints produced

152 EXPERIMENTS ON COLOUR, AB PERCEIVED BY THE EYE.

was owing to some optical effect taking place in the transparent part of the
eye on the mixture of two rays of very different refrangibility. Most eyes are
by no means achromatic, so that the images of objects illuminated with mixed
light of this kind appear divided into two different colours ; and even when
there is no distinct object, the mixtures become in some degree analysed, so as
to present a very strange, and certainly "anonymous" appearance.

Additional Note on the more recent experiments of M. Helmholtz*.

In his former memoir on the Theory of Compound Colours f, M. Helmholtz
arrived at the conclusion that only one pair of homogeneous colours, orange-
yellow and indigo-blue, were strictly complementary. This result was shewn by
Professor GrassmannJ to be at variance with Newton's theory of compound
colours ; and although the reasoning was founded on intuitive rather than
t.-.\l>erimental truths, it pointed out the tests by which Newton's theory must
be verified or overthrown. In applying these tests, M. Helmholtz made use of
an apparatus similar to that described by M. Foucault§, by which a screen of
white paper is illuminated by the mixed light. The field of mixed colour is
much larger than in M. Helmholtz's former experiments, and the facility of
forming combinations is much increased. In this memoir the mathematical theory
of Newton's circle, and of the curve formed by the spectrum, with its possible
transformations, is completely stated, and the form of this curve is in some
degree indicated, as far as the determination of the colours which lie on oppo-
site sides of white, and of those which lie opposite the part of the curve which
is wanting. The colours between red and yellow-green are complementary to
colours between blue-green and violet, and those between yellow-green and blue-
green have no homogeneous complementaries, but must be neutralized by various
hues of purple, i.e., mixtures of red and violet. The names of the complementary
colours, with their wave-lengths in air, as deduced from Fraunhofer's measure-
ments, are given in the following table : —

* PoggendorfFa Annalen, Bd. XCIV. (I am indebted for the perusal of this Memoir to Professor
Stokes.)

t 76. Bd. LXXXVII. Annals of Philosophy, 1852, Part n.

* 76. Bd. LXXXIX. Ann. Phil., 1854, April.

S Ib. Bd. LXXXVIII. Moigno, Cosmos, 1853, Tom. u., p. 232.

EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EYE.

153

Colour

Wave-length

Complementary
Colour

Wave-length

1 tut io of
wave-lengths

Red ...

2425

Green-blue .

1818

1-334

Orange .

2244

Blue . . .

1809

1-240

Gold-yellow

2162

Blue . . .

1793

1-206

Gold yellow

2120

Blue . . .

1781

1-190

Yellow . .

2095

Indigo-blue

1716

1-221

Yellow . .

2085

Indigo-blue

1706

1-222

Green-yellow

2082

Violet . .

1600-

1-301

(The wave-lengths are expressed in milliontha of a Paris inch.)

(In order to reduce these wave-lengths to their actual lengths in the eye,
each must be divided by the index of refraction for that kind of light in the
medium in which the physical effect of the vibrations is supposed to take place.)

Although these experiments are not in themselves sufficient to give the com-
plete theory of the curve of homogeneous colours, they determine the most
important element of that theory in a way which seems very accurate, and I
cannot doubt that when a philosopher who has so fully pointed out the im-
portance of general theories in physics turns his attention to the theory of
sensation, he will at least establish the principle that the laws of sensation can
be successfully investigated only after the corresponding physical laws have been
ascertained, and that the connection of these two kinds of laws can be appre-
hended only when the distinction between them is fully recognised.

NOTE IV.
Description of the Figures. Plate I.

No. 1. is the colour-diagram already referred to, representing, on Newton's principle, the relations of
different coloured papers to the three standard colours — vermilion, emerald-green, and ultra-
marine. The initials denoting the colours are explained ill the list at page 276, and the
numbers belonging to them are their coefficients of intensity, the use of which has been
explained. The initials H.R., H.B., and H.G., represent the red, blue and green papers
of Mr HAY, and serve to connect this diagram with No. (2), which takes these colours for
its standards.
VOL. I. 20

154 EXPERIMENTS ON COLOUR, AS PERCEIVED BY THE EVI

No. 2. represents the relations of Mr HAY'S red, blue, groen, white, and yellow papers, as deter-
mined by a large number of experiments at Cambridge. — (See Note II.). The use of the
IK >int D, in calculating the results of colour-blindness, is explained in the Pa|»T.

Fig. 3. represents a disc of the larger size, with its slit

Fig. 4. shows the mode of combining two discs of the smaller size.

Fig. 5. shows the combination of discs, as placed on the top, in the first experiment described
in the Paper.

Fig. G. represents the method of spinning the top, when speed is required.

The last four figures are half the actual size.

Colour-tops of the kind used in these experiments, with paper discs of the colours whose relations
are represented in No. 1, are to be had of Mr J. M. BBYSON, Optician, Edinburgh.

VOL. I. PLATE L

P1G.3

-ff-g
Ban-,

BBO6* \

FIG. 6

FIG.5

llu-.l-O

To /'if p<ir>? ]f>4

Cambridge. Uniivi'tity Pratt

[From the Transactions of the Cambridge Philosophical Society, Vol. X. Part I.]

VIII. On Faraday's Lines of Force.
[Read Dec. 10, 1855, and Feb. 11, 1856.]

THE present state of electrical science seems peculiarly unfavourable to specu-
lation. The laws of the distribution of electricity on the surface of conductors
have been analytically deduced from experiment ; some parts of the mathematical
theory of magnetism are established, while in other parts the experimental data
are wanting ; the theory of the conduction of galvanism and that of the mutual
attraction of conductors have been reduced to mathematical formulae, but have
not fallen into relation with the other parts of the science. No electrical theory
can now be put forth, unless it shews the connexion not only between electricity
at rest and current electricity, but between the attractions and inductive effects
of electricity in both states. Such a theory must accurately satisfy those laws,
the mathematical form of which is known, and must afford the means of calcu-
lating the effects in the limiting cases where the known formulae are inapplicable.
In order therefore to appreciate the requirements of the science, the student
must make himself familiar with a considerable body of most intricate mathe-
matics, the mere retention of which in the memory materially interferes with
further progress. The first process therefore in the effectual study of the science,
must be one of simplification and reduction of the results of previous investiga-
tion to a form in which the mind can grasp them. The results of this simplifi-
cation may take the form of a purely mathematical formula or of a physical
hypothesis. In the first case we entirely lose sight of the phenomena to be
explained ; and though we may trace out the consequences of given laws, we
can never obtain more extended views of the connexions of the subject. If,
on the other hand, we adopt a physical hypothesis, we see the phenomena only
through a medium, and are liable to that blindness to facts and rashness in

20—2

156 ON FARADAY'S LINES OF FORCE.

assumption which a partial explanation encourages. We must therefore discover
some method of investigation which allows the mind at every step to lay hold
of a clear physical conception, without being committed to any theory founded
on the physical science from which that conception is borrowed, so that it is
neither drawn aside from the subject in pursuit of analytical subtleties, nor carried
beyond the truth by a favourite hypothesis.

In order to obtain physical ideas without adopting a physical theory we must
make ourselves familiar with the existence of physical analogies. By a physical
analogy I mean that partial similarity between the laws of one science and those
of another which makes each of them illustrate the other. Thus all the mathe-
matical sciences are founded on relations between physical laws and laws of
numbers, so that the aim of exact science is to reduce the problems of nature
to the determination of quantities by operations with numbers. Passing from
the most universal of all analogies to a very partial one, we find the same
resemblance in mathematical form between two different phenomena giving rise
to a physical theory of light.

The changes of direction which light undergoes in passing from one medium
to another, are identical with the deviations of the patli of a particle in moving
through a narrow space in which intense forces act. This analogy, which extends
only to the direction, and not to the velocity of motion, was long believed to
be the true explanation of the refraction of light ; and we still find it useful
in the solution of certain problems, in which we employ it without danger, as
an artificial method. The other analogy, between light and the vibrations of an
elastic medium, extends much farther, but, though its importance and fruitfulness
cannot be over-estimated, we must recollect that it is founded only on a resem-
blance in form between the laws of light and those of vibrations. By stripping
it of its physical dress and reducing it to a theory of " transverse alternations,"
we might obtain a system of truth strictly founded on observation, but probably
deficient both in the vividness of its conceptions and the fertility of its method.
I have said thus much on the disputed questions of Optics, as a preparation
for the discussion of the almost universally admitted theory of attraction at a
distance.

We have all acquired the mathematical conception of these attractions. We
can reason about them and determine their appropriate forms or formulae. These
formulae have a distinct mathematical significance, and their results are found
to be in accordance with natural phenomena. There is no formula in applied

ON FARADAY'S LINES OF FORCE. 157

mathematics more consistent with nature than the formula of attractions, and no
theory better established in the minds of men than that of the action of bodies
on one another at a distance. The laws of the conduction of heat in uniform
media appear at first sight among the most different in their physical relations
from those relating to attractions. The quantities which enter into them are
temperature, Jloio of heat, conductivity. The word force is foreign to the subject.
Yet we find that the mathematical laws of the uniform motion of heat in
homogeneous media are identical in form with those of attractions varying in-
versely as the square of the distance. We have only to substitute source of
heat for centre of attraction, flow of heat for accelerating effect of attraction at
any point, and temperature for potential, and the solution of a problem in
attractions is transformed into that of a problem in heat.

This analogy between the formulae of heat and attraction was, I believe,
first pointed out by Professor William Thomson in the Camb. Math. Journal,
Vol. III.

Now the conduction of heat is supposed to proceed by an action between
contiguous parts of a medium, while the force of attraction is a relation be-
tween distant bodies, and yet, if we knew nothing more than is expressed in
the mathematical formulae, there would be nothing to distinguish between the
one set of phenomena and the other.

It is true, that if we introduce other considerations and observe additional
facts, the two subjects will assume very different aspects, but the mathematical
resemblance of some of their laws will remain, and may still be made useful
in exciting appropriate mathematical ideas.

It is by the use of analogies of this kind that I have attempted to bring
before the mind, in a convenient and manageable form, those mathematical ideas
which are necessary to the study of the phenomena of electricity. The methods
are generally those suggested by the processes of reasoning which are found in
the researches of Faraday*, and which, though they have been interpreted
mathematically by Prof. Thomson and others, are very generally supposed to be
of an indefinite and unmathematical character, when compared with those em-
ployed by the professed mathematicians. By the method which I adopt, I hope
to render it evident that I am not attempting to establish any physical theory
of a science in which I have hardly made a single experiment, and that the
limit of my design is to shew how, by a strict application of the ideas and

* See especially Series xxxvm. of the Experimental Researches, and Phil. Mag. 1852.

158 ON FARADAY'S LINES OF FORCE.

methods of Faraday, the connexion of the very different orders of phenomena
which he has discovered may be clearly placed before the mathematical mind.
I shall therefore avoid as much as I can the introduction of anything which
does not serve as a direct illustration of Faraday's methods, or of the mathe-
matical deductions which may be made from them. In treating the simpler
parts of the subject I shall use Faraday's mathematical methods as well as
his ideas. When the complexity of the subject requires it, I shall use analytical
notation, still confining myself to the development of ideas originated by the
same philosopher.

I have in the first place to explain and illustrate the idea of " lines of
force."

When a body is electrified in any manner, a small body charged with posi-
tive electricity, and placed in any given position, will experience a force urging
it in a certain direction. If the small body be now negatively electrified, it will
be urged by an equal force in a direction exactly opposite.

The same relations hold between a magnetic body and the north or south
poles of a small magnet. If the north pole is urged in one direction, the south
pole is urged in the opposite direction.

In this way we might find a line passing through any point of space, such
that it represents the direction of the force acting on a positively electrified
particle, or on an elementary north pole, and the reverse direction of the force
on a negatively electrified particle or an elementary south pole. Since at every
point of space such a direction may be found, if we commeilce at any point
and draw a line so that, as we go along it, its direction at any point shall
always coincide with that of the resultant force at that point, this curve will
indicate the direction of that force for every point through which it passes, and
might be called on that account a line of force. We might in the same way
draw other lines of force, till we had filled all space with curves indicating by
their direction that of the force at any assigned point.

We should thus obtain a geometrical model of the physical phenomena,
which would tell us the direction of the force, but we should still require some
method of indicating the intensity of the force at any point. If we consider
these curves not as mere lines, but as fine tubes of variable section carrying
an incompressible fluid, then, since the velocity of the fluid is inversely as the
section of the tube, we may make the velocity vary according to any given law,
\>y regulating the section of the tube, and in this way we might represent the

ON FAEADAYS LINES OF FORCE. 159

intensity of the force as well as its direction by the motion of the fluid in
these tubes. This method of representing the intensity of a force by the velocity
of an imaginary fluid in a tube is applicable to any conceivable system of forces,
but it is capable of great simplification in the case in which the forces are such
as can be explained by the hypothesis of attractions varying inversely as the
square of the distance, such as those observed in electrical and magnetic pheno-
mena. In the case of a perfectly arbitrary system of forces, there will generally
be interstices between the tubes ; but in the case of electric and magnetic forces
it is possible to arrange the tubes so as to leave no interstices. The tubes will
then be mere surfaces, directing the motion of a fluid filling up the whole space.
It has been usual to commence the investigation of the laws of these forces by
at once assuming that the phenomena are due to attractive or repulsive forces
acting between certain points. We may however obtain a different view of the
subject, and one more suited to our more difficult inquiries, /by adopting for the
definition of the forces of which we treat, that they may be represented in
magnitude and direction by the uniform motion of an incompressible fluid.

I propose, then, first to describe a method by which the motion of such a
fluid can be clearly conceived ; secondly to trace the consequences of assuming
certain conditions of motion, and to point out the application of the method to
some of the less complicated phenomena of electricity, magnetism, and galvanism ;
and lastly to shew how by an extension of these methods, and the introduction
of another idea due to Faraday, the laws of the attractions and inductive actions
of magnets and currents may be clearly conceived, without making any assump-
tions as to the physical nature of electricity, or adding anything to that which
has been already proved by experiment.

By referring everything to the purely geometrical idea of the motion of an
imaginary fluid, I hope to attain generality and precision, and to avoid the
dangers arising from a premature theory professing to explain the cause of the
phenomena. If the results of mere speculation which I have collected are found
to be of any use to experimental philosophers, in arranging and interpreting
their results, they will have served their purpose, and a mature theory, in which
physical facts will be physically explained, will be formed by those who by
interrogating Nature herself can obtain the only true solution of the questions
which the mathematical theory suggests.

160 ON FARADAY'S LINES OP FORCE.

I. Theory of the Motion of an incompressible FluiJ.

(1) The substance here treated of must not be assumed to possess any of
the properties of ordinary fluids except those of freedom of motion and resistance
to compression. It is not even a hypothetical fluid which is introduced to
explain actual phenomena. It is merely a collection of imaginary properties
which may be employed for establishing certain theorems in pure mathematics in
a way more intelligible to many minds and more applicable to physical problems
than that in which algebraic symbols alone are used. The use of the word
" Fluid " will not lead us into error, if we remember that it denotes a purely
imaginary substance with the following property :

T7ie portion of fluid which at any instant occupied a given volume, will at
any succeeding instant occupy an equal volume.

This law expresses the incompressibility of the fluid, and furnishes us with
a convenient measure of its quantity, namely its volume. The unit of quantity
of the fluid will therefore be the unit of volume.

(2) The direction of motion of the fluid will in general be different at
different points of the space which it occupies, but since the direction is deter-
minate for every such point, we may conceive a line to begin at any point and
to be continued so that every element of the line indicates by its direction the
direction of motion at that point of space. Lines drawn in such a manner that
their direction always indicates the direction of fluid motion are called lines of
fluid motion.

If the motion of the fluid be what is called steady motion, that is, if the
direction and velocity of the motion at any fixed point be independent of the
tune, these curves will represent the paths of individual particles of the fluid,
but if the motion be variable this will not generally be the case. The cases
of motion which will come under our notice will be those of steady motion.

(3) If upon any surface which cuts the lines of fluid motion we draw a
closed curve, and if from every point of this curve we draw a line of motion,
these lines of motion will generate a tubular surface which we may call a tube
of fluid motion. Since this surface is generated by lines in the direction of fluid

ON FARADAY'S LINES OF FORCE. 161

motion no part of the fluid can flow across it, so that this imaginary surface
is as impermeable to the fluid as a real tube.

(4) The quantity of fluid which in unit of time crosses any fixed section
of the tube is the same at whatever part of the tube the section be taken.
For the fluid is incompressible, and no part runs through the sides of the tube,
therefore the quantity which escapes from the second section is equal to that
which enters through the first.

If the tube be such that unit of volume passes through any section in
unit of time it is called a unit tube of fluid motion.

(5) In what follows, various units will be referred to, and a finite number
of lines or surfaces will be drawn, representing in terms of those units the
motion of the fluid. Now in order to define the motion in every part of the
fluid, an infinite number of lines would have to be drawn 'at indefinitely small
intervals ; but since the description of such a system of lines would involve
continual reference to the theory of limits, it has been thought better to suppose
the lines drawn at intervals depending on the assumed unit, and afterwards to
assume the unit as small as we please by taking a small submultiple of the
standard unit.

(6) To define the motion of the whole fluid by means of a system of unit
tubes.

Take any fixed surface which cuts all the lines of fluid motion, and draw
upon it any system of curves not intersecting one another. On the same surface
draw a second system of curves intersecting the first system, and so arranged
that the quantity of fluid which crosses the surface within each of the quadri-
laterals formed by the intersection of the two systems of curves shall be unity
in unit of time. From every point in a curve of the first system let a line
of fluid motion be drawn. These lines will form a surface through which no
fluid passes. Similar impermeable surfaces may be drawn for all the curves of
the first system. The curves of the second system will give rise to a second
system of impermeable surfaces, which, by their intersection with the first system,
will form quadrilateral tubes, which will be tubes of fluid motion. Since each
quadrilateral of the cutting surface transmits unity of fluid in unity of time,
every tube in the system will transmit unity of fluid through any of its sections
in unit of time. The motion of the fluid at every part of the space it occupies

VOL. i. 21

162 ON FARADAY'S LINES OF FORCE.

is determined by this system of unit tubes; for the direction of motion is that
of the tube through the point in question, and the velocity is the reciprocal
of the area of the section of the unit tube at that point.

(7) We have now obtained a geometrical construction which completely
defines the motion of the fluid by dividing the space it occupies into a system
of unit tubes. We have next to shew how by means of these tubes we may
ascertain various points relating to the motion of the fluid.

A unit tube may either return into itself, or may begin and end at differ-
ent points, and these may be either in the boundary of the space in which we
investigate the motion, or within that space. In the first case there is a con-
tinual circulation of fluid in the tube, in the second the fluid enters at one end
and flows out at the other. If the extremities of the tube are in the bound-
ing surface, the fluid may be supposed to be continually supplied from without
from an unknown source, and to flow out at the other into an unknown reser-
voir ; but if the origin of the tube or its termination be within the space under
consideration, then we must conceive the fluid to be supplied by a source within
that space, capable of creating and emitting unity of fluid in unity of time, and
to be afterwards swallowed ,up by a sink capable of receiving and destroying
the same amount continually.

There is nothing self-contradictory in the conception of these sources where
the fluid is created, and sinks where it is annihilated. The properties of the
fluid are at our disposal, we have made it incompressible, and now we suppose
it produced from nothing at certain points and reduced to nothing at others.
The places of production will be called sources, and their numerical value will be
the number of units of fluid which they produce in unit of time. The places
of reduction will, for want of a better name, be called sinks, and will be esti-
mated by the number of units of fluid absorbed in unit of time. Both places
will sometimes be called sources, a source being understood to be a sink when
its sign is negative.

(8) It is evident that the amount of fluid which passes any fixed surface
is measured by the number of unit tubes which cut it, and the direction in
which the fluid passes is determined by that of its motion in the tubes. If
the surface be a closed one, then any tube whose terminations lie on the same
side of the surface must cross the surface as many times in the one direction
as in the other, and therefore must carry as much fluid out of the surface as

ON FAEADAYS LINES OF FORCE. 163

it carries in. A tube which begins within the surface and ends without it
will carry out unity of fluid ; and one which enters the surface and terminates
within it will carry in the same quantity. In order therefore to estimate the
amount of fluid which flows out of the closed surface, we must subtract the
number of tubes which end within the surface from the number of tubes which
begin there. If the result is negative the fluid will on the whole flow inwards.

If we call the beginning of a unit tube a unit source, and its termination
a unit sink, then the quantity of fluid produced within the surface is estimated
by the number of unit sources minus the number of unit sinks, and this must
flow out of the surface on account of the incompressibility of the fluid.

In speaking of these unit tubes, sources and sinks, we must remember what
was stated in (5) as to the magnitude of the unit, and how by diminishing
their size and increasing their number we may distribute them according to any

law however complicated.

/•

(9) If we know the direction and velocity of the fluid at any point in

two different cases, and if we conceive a third case in which the direction and
velocity of the fluid at any point is the resultant of the velocities in the two
former cases at corresponding points, then the amount of fluid which passes a
given fixed surface in the third case will be the algebraic sum of the quantities
which pass the same surface in the two former cases. For the rate at which
the fluid crosses any surface is the resolved part of the velocity normal to the
surface, and the resolved part of the resultant is equal to the sum of the
resolved parts of the components.

Hence the number of unit tubes which cross the surface outwards in the
third case must be the algebraical sum of the numbers which cross it in the
two former cases, and the number of sources within any closed surface will be
the sum of the numbers in the two former cases. Since the closed surface may
be taken as small as we please, it is evident that the distribution of sources
and sinks in the third case arises from the simple superposition of the distri-
butions in the two former cases.

II. Theory of the uniform motion of an imponderable incompressible fluid

through a resisting medium.

(10) The fluid is here supposed to have no inertia, and its motion is opposed
by the action of a force which we may conceive to be due to the resistance of a

21—2

104 ON FARADAY'S LINES OF FORCE.

medium through which the fluid is supposed to flow. This resistance depends on
the nature of the medium, and will in general depend on the direction in which
the fluid moves, as well as on its velocity. For the present we may restrict
ourselves to the case of a uniform medium, whose resistance is the same in all
directions. The law which we assume is as follows.

Any portion of the fluid moving through tJie resisting medium is directly
opposed by a retarding force proportional to its velocity.

If the velocity be represented by v, then the resistance will be a force equal
to kv acting on unit of volume of the fluid in a direction contrary to that of
motion. In order, therefore, that the velocity may be kept up, there must be a
greater pressure behind any portion of the fluid than there is in front of it, so
that the difference of pressures may neutralise the effect of the resistance. Con-
ceive a cubical unit of fluid (which we may make as small as we please, by (5)),
and let it move in a direction perpendicular to two of its faces. Then the resist-
ance will be kv, and therefore the difference of pressures on the first and second
faces is kv, so that the pressure diminishes in the direction of motion at the rate
of kv for every unit of length measured along the line of motion ; so that if we
measure a length equal to h units, the difference of pressure at its extremities
will be kvh.

(11) Since the pressure is supposed to vary continuously in the fluid, all
the points at which the pressure is equal to a given pressure p will lie on a
certain surface which we may call the surface (p) of equal pressure. If a series
of these surfaces be constructed in the fluid corresponding to the pressures 0, 1,
2, 3 &c., then the number of the surface will indicate the pressure belonging to
it, and the surface may be referred to as the surface 0, 1, 2 or 3. The unit of
pressure is that pressure which is produced by unit of force acting on unit of
surface. In order therefore to diminish the unit of pressure as in (5) we must
diminish the unit of force in the same proportion.

(12) It is easy to see that these surfaces of equal pressure must be perpen-
dicular to the lines of fluid motion ; for if the fluid were to move in any other
direction, there would be a resistance to its motion which could not be balanced
by any difleTence of pressures. (We must remember that the fluid here con-
sidered has no inertia or mass, and that its properties are those only which are
formally assigned to it, so that the resistances and pressures are the only things

ON FARAD AYS LINES OF FORCE. 165

to be considered.) There are therefore two sets of surfaces which by their inter-
section form the system of unit tubes, and the system of surfaces of equal pres-
sure cuts both the others at right angles. Let h be the distance between two
consecutive surfaces of equal pressure measured along a line of motion, then since
the difference of pressures = 1,

kvh=l,

which determines the relation of v to h, so that one can be found when the
other is known. Let s be the sectional area of a unit tube measured on a
surface of equal pressure, then since by the definition of a unit tube

vs = l,

we find by the last equation

s = kh.

(13) The surfaces of equal pressure cut the unit tubes /into portions whose
length is h and section s. These elementary portions of unit tubes will be called
unit cells. In each of them unity of volume of fluid passes from a pressure p to
a pressure (p — 1) in unit of time, and therefore overcomes unity of resistance in
that time. The work spent in overcoming resistance is therefore unity in every
cell in every unit of time.

(14) If the surfaces of equal pressure are known, the direction and magni-
tude of the velocity of the fluid at any point may be found, after which the
complete system of unit tubes may be constructed, and the beginnings and end-
ings of these tubes ascertained and marked out as the sources whence the fluid
is derived, and the sinks where it disappears. In order to prove the converse of \
this, that if the distribution of sources be given, the pressure at every point may
be found, we must lay down certain preliminary propositions.

(15) If we know the pressures at every point in the fluid in two different
cases, and if we take a third case in which the pressure at any point is the
sum of the pressures at corresponding points in the two former cases, then the
velocity at any point in the third case is the resultant of the velocities in the
other two, and the distribution of sources is that due to the simple superposition
of the sources in the two former cases.

For the velocity in any direction is proportional to the rate of decrease of
the pressure in that direction ; so that if two systems of pressures be added

166 ox FARADAY'S LINES OF FORCE.

together, since the rate of decrease of pressure along any line will be the sum
of the combined rates, the velocity in the new system resolved in the same
direction will be the sum of the resolved parta in the two original systems.
The velocity in the new system will therefore be the resultant of the velocities
at corresponding points in the two former systems.

It follows from this, by (9), that the quantity of fluid which crosses any
fixed surface is, in the new system, the sum of the corresponding quantities in
the old ones, and that the sources of the two original systems are simply
combined to form the third.

It is evident that in the system in which the pressure is the difference
of pressure in the two given systems the distribution of sources will be got
by changing the sign of all the sources in the second system and adding them
to those in the first.

(16) If the pressure at every point of a closed surface be the same and
equal to p, and if there be no sources or sinks within the surface, then there
will be no motion of the fluid within the surface, and the pressure within it
will be uniform and equal to p.

For if there be motion of the fluid within the surface there will be tubes
of fluid motion, and these tubes must either return into themselves or be
terminated either within the surface or at its boundary. Now since the fluid
always flows from places of greater pressure to places of less pressure, it
cannot flow in a re-entering curve; since there are no sources or sinks within
the surface, the tubes cannot begin or end except on the surface ; and since
the pressure at all points of the surface is the same, there can be no motion
in tubes having both extremities on the surface. Hence there is no motion
within the surface, and therefore no difference of pressure which would cause
motion, and since the pressure at the bounding surface is p, the pressure at
any point within it is also p.

(17) If the pressure at every point of a given closed surface be known,
and the distribution of sources within the surface be also known, then only
one distribution of pressures can exist within the surface.

For if two different distributions of pressures satisfying these conditions
could be found, a third distribution could be formed in which the pressure at
any point should be the difference of the pressures in the two former distri-
butions. In this case, since the pressures at the surface and the sources within

ox FARADAY'S LINES OF FORCE. 167

it are the same in both distributions, the pressure at the surface in the third
distribution would be zero, and all the sources within the surface would
vanish, by (15).

Then by (16) the pressure at every point in the third distribution must
be zero ; but this is the difference of the pressures in the two former cases,
and therefore these cases are the same, and there is only one distribution of
pressure possible.

(18) Let us next determine the pressure at any point of an infinite body
of fluid in the centre of which a unit source is placed, the pressure at an
infinite distance from the source being supposed to be zero.

The fluid will flow out from the centre symmetrically, and since unity of
volume flows out of every spherical surface surrounding the point in unit of
time, the velocity at a distance r from the source will be

1

v =

47JT3'

Tc

The rate of decrease of pressure is therefore kv or —3, and since the

pressure = 0 when r is infinite, the actual pressure at any point will be

ft

The pressure is therefore inversely proportional to the distance from the
source.

It is evident that the pressure due to a unit sink will be negative and

equal to — — - .

47JT

If we have a source formed by the coalition of S unit sources, then the

resulting pressure will be p = - - , so that the pressure at a given distance
varies as the resistance and number of sources conjointly.

(19) If a number of sources and sinks coexist in the fluid, then in order
to determine the resultant pressure we have only to add the pressures which
each source or sink produces. For by (15) this will be a solution of the
problem, and by (17) it will be the only one. By this method we can
determine the pressures due to any distribution of sources, as by the method

168 ON FARADAY'S LIMES OF FORCE.

of (14) we can determine the distribution of sources to which a given distri-
bution of pressures is due.

(20) We have next to shew that if we conceive any imaginary surface
as fixed in space and intersecting the lines of motion of the fluid, we may
substitute for the fluid on one side of this surface a distribution of sources
upon the surface itself without altering in any way the motion of the fluid
on the other side of the surface.

For if we describe the system of unit tubes which defines the motion of
the fluid, and wherever a tube enters through the surface place a unit source,
and wherever a tube goes out through the surface place a unit sink, and at the
same time render the surface impermeable to the fluid, the motion of the fluid
in the tubes will go on as before.

(21) If the system of pressures and the distribution of sources which pro-
duce them be known in a medium whose resistance is measured by k, then in
order to produce the same system of pressures in a medium whose resistance
is unity, the rate of production at each source must be multiplied by k. For
the pressure at any point due to a given source varies as the rate of produc-
tion and the resistance conjointly ; therefore if the pressure be constant, the
rate of production must vary inversely as the resistance.

(22) On the conditions to be fulfilled at a surface which separates two media
whose coefficients of resistance are k and k'.

These are found from the consideration, that the quantity of fluid which
flows out of the one medium at any point flows into the other, and that the
pressure varies continuously from one medium to the other. The velocity normal
to the surface is the same in both media, and therefore the rate of diminution
of pressure is proportional to the resistance. The direction of the tubes of
motion and the surfaces of equal pressure will be altered after passing through
the surface, and the law of this refraction will be, that it takes place in the
plane passing through the direction of incidence and the normal to the surface,
and that the tangent of the angle of incidence is to the tangent of the angle
of refraction as k' is to k.

(23) Let the space within a given closed surface be filled with a medium
different from that exterior to it, and let the pressures at any point of this
compound system due to a given distribution of sources within and without

ON FARADAY'S LINES OF FOKCE. 169

the surface be given ; it is required to determine a distribution of sources which
would produce the same system of pressures in a medium whose coefficient of
resistance is unity.

Construct the tubes of fluid motion, and wherever a unit tube enters either
medium place a unit source, and wherever it leaves it place a unit sink. Then
if we make the surface impermeable all will go on as before.

Let the resistance of the exterior medium be measured by k, and that of
the interior by k'. Then if we multiply the rate of production of all the sources
in the exterior medium (including those in the surface), by k, and make the
coefficient of resistance unity, the pressures will remain as before, and the same
will be true of the interior medium if we multiply all the sources in it by k',
including those in the surface, and make its resistance unity.

Since the pressures on both sides of the surface are now equal, we may
suppose it permeable if we please. /

We have now the original system of pressures produced in a uniform medium
by a combination of three systems of sources. The first of these is the given
external system multiplied by k, the second is the given internal system multi-
plied by k', and the third is the system of sources and sinks on the surface
itself. In the original case every source in the external medium had an equal
sink in the internal medium on the other side of the surface, but now the
source is multiplied by k and the sink by k', so that the result is for every
external unit source on the surface, a source = (k — k'). By means of these three
systems of sources the original system of pressures may be produced in a medium
for which k=l.

(24) Let there be no resistance in the medium within the closed surface,
that is, let k' = 0, then the pressure within the closed surface is uniform and
equal to p, and the pressure at the surface itself is also p. If by assuming
any distribution of pairs of sources and sinks within the surface in addition to
the given external and internal sources, and by supposing the medium the same
within and without the surface, we can render the pressure at the surface uni-
form, the pressures so found for the external medium, together with the uniform
pressure p in the internal medium, will be the true and only distribution of
pressures which is possible.

For if two such distributions could be found by taking different imaginary
distributions of pairs of sources and sinks within the medium, then by taking
VOL. I. 22

170 ON FARADAY'S LINES OF FORCE.

the difference of the two for a third distribution, we should have the pressure
of the bounding surface constant in the new system and as many sources as
sinks within it, and therefore whatever fluid flows in at any point of the surface,
an equal quantity must flow out at some other point.

In the external medium all the sources destroy one another, and we have
an infinite medium without sources surrounding the internal medium. The pres-
sure at infinity is zero, that at the surface is constant. If the pressure at the
surface is positive, the motion of the fluid must be outwards from every point
of the surface ; if it be negative, it must flow inwards towards the surface. But
it has been shewn that neither of these cases is possible, because if any fluid
enters the surface an equal quantity must escape, and therefore the pressure at
the surface is zero in the third system.

The pressure at all points in the boundary of the internal medium in the
third case is therefore zero, and there are no sources, and therefore the pressure
is everywhere zero, by (16).

The pressure in the bounding surface of the internal medium is also zero,
and there is no resistance, therefore it is zero throughout ; but the pressure in
the third case is the difference of pressures in the two given cases, therefore
these are equal, and there is only one distribution of pressure which is possible,
namely, that due to the imaginary distribution of sources and sinks.

(25) When the resistance is infinite in the internal medium, there can be
no passage of fluid through it or into it. The bounding surface may therefore
be considered as impermeable to the fluid, and the tubes of fluid motion will
run along it without cutting it.

If by assuming any arbitrary distribution of sources within the surface in
addition to the given sources in the outer medium, and by calculating the
resulting pressures and velocities as in the case of a uniform medium, we can
fulfil the condition of there being no velocity across the surface, the system of
pressures in the outer medium will be the true one. For since no fluid passes
through the surface, the tubes in the interior are independent of those outside,
and may be taken away without altering the external motion.

(26) If the extent of the internal medium be small, and if the difference
of resistance in the two media be also small, then the position of the unit tubes
will not be much altered from what it would be if the external medium filled
the whole space.

ON FARADAY'S LINES OF FORCE. 171

On this supposition we can easily calculate the kind of alteration -which
the introduction of the internal medium will produce ; for wherever a unit tube

k' — k
enters the surface we must conceive a source producing fluid at a rate —j — ,

K

and wherever a tube leaves it we must place a sink annihilating fluid at the

k' — k
rate — j— , then calculating pressures on the supposition that the resistance in

1C

both media is k, the same as in the external medium, we shall obtain the true
distribution of pressures very approximately, and we may get a better result
by repeating the process on the system of pressures thus obtained.

(27) If instead of an abrupt change from one coefficient of resistance to
another we take a case in which the resistance varies continuously from point
to point, we may treat the medium as if it were composed of thin shells each
of which has uniform resistance. By properly assuming a distribution of sources
over the surfaces of separation of the shells, we may treat the case as if the
resistance were equal to unity throughout, as in (23). The sources will then
be distributed continuously throughout the whole medium, and will be positive
whenever the motion is from places of less to places of greater resistance, and
negative when in the contrary direction.

(28) Hitherto we have supposed the resistance at a given point of the
medium to be the same in whatever direction the motion of the fluid takes
place ; but we may conceive a case in which the resistance is different in
different directions. In such cases the lines of motion will not in general be
perpendicular to the surfaces of equal pressure. If a, b, c be the components
of the velocity at any point, and a, ft, y the components of the resistance at
the same point, these quantities will be connected by the following system of
linear equations, which may be called "equations of conduction," and will be
referred to by that name.

In these equations there are nine independent coefficients of conductivity. In
order to simplify the equations, let us put

............ &c ............. &c.

22—2

172 ON FARADAY'S LIKES or FORCE.

where

and I, m. n are direction-cosines of a certain fixed line in space.

The equations then become

^p -my) T,
- no) T,
-l T.

By the ordinary transformation of co-ordinates we may get rid of the
coefficients marked S. The equations then become

where I', m', n' are the direction-cosines of the fixed line with reference to the
new axes. If we make

dp 0 dp dp

a= - ™*=

the equation of continuity

da db dc _
dx dy dz ~

becomes

and if we make x

d*P . d*p , d'p

then :j£+T7^+ jw=°'

dj? drf d£

the ordinary equation of conduction.

It appears therefore that the distribution of pressures is not altered by
the existence of the coefficient T. Professor Thomson has shewn how to
conceive a substance in which this coefficient determines a property having
reference to an axis, which unlike the axes of Plt Pt, Pt is dipolar.

For further information on the equations of conduction, see Professor
Stokes On the Conduction of Heat in Crystals (Cambridge and Dublin Math.
Journ.), and Professor Thomson On the Dynamical Theory of Heat, Part v.
(Transactions of Royal Society of Edinburgh, VoL xxi. Part i.).

ON FARADAY'S LINES OF FORCE. 173

It is evident that all that has been proved in (14), (15), (16), (17), with
respect to the superposition of different distributions of pressure, and there being
only one distribution of pressures corresponding to a given distribution of sources,
will be true also in the case in which the resistance varies from point to point,
and the resistance at the same point is different in different directions. For
if we examine the proof we shall find it applicable to such cases as well as to
that of a uniform medium.

(29) We now are prepared to prove certain general propositions which are
true in the most general case of a medium whose resistance is different in
different directions and varies from point to point.

We may by the method of (28), when the distribution of pressures is
known, construct the surfaces of equal pressure, the tubes of fluid motion, and
the sources and sinks. It is evident that since in each cell into which a unit
tube is divided by the surfaces of equal pressure unity of fluid passes from
pressure p to pressure (p — 1) in unit of time, unity of work is done by the
fluid in each cell in overcoming resistance.

The number of cells in each unit tube is determined by the number of
surfaces of equal pressure through which it passes. If the pressure at the
beginning of the tube be p and at the end p', then the number of cells in
it will be p—p'. Now if the tube had extended from the source to a place
where the pressure is zero, the number of cells would have been p, and if
the tube had come from the sink to zero, the number would have been p',
and the true number is the difference of these.

Therefore if we find the pressure at a source S from which S tubes
proceed to be p, Sp is the number of cells due to the source S ; but if S' of
the tubes terminate in a sink at a pressure p', then we must cut off S'p' cells
from the number previously obtained. Now if we denote the source of S
tubes by S, the sink of & tubes may be written — £?, sinks always being
reckoned negative, and the general expression for the number of cells in the
system will be 2 (Sp).

(30) The same conclusion may be arrived at by observing that unity of
work is done on each cell. Now in each source S, S units of fluid are
expelled against a pressure p, so that the work done by the fluid in over-
coming resistance is Sp. At each sink in which S' tubes terminate, S' units
of fluid sink into nothing under pressure p' ; the work done upon the fluid by

174 ON FARADAY'S LIMES OF FORCE.

the pressure is therefore S'p'. The whole work done by the fluid may there-
fore be expressed by

or more concisely, considering sinks as negative sources,

(31) Let S represent the rate of production of a source in any medium,
and let p be the pressure at any given point due to that source. Then if we
superpose on this another equal source, every pressure will be doubled, and
thus by successive superposition we find that a source nS would produce a
pressure np, or more generally the pressure at any point due to a given
source varies as the rate of production of the source. This may be expressed
by the equation

where R is a coefficient depending on the nature of the medium and on the
positions of the source and the given point. In a uniform medium whose
resistance is measured by k,

&? /?_A
P~±vr' ~47rr'

R may be called the coefficient of resistance of the medium between the source
and the given point. By combining any number of sources we have generally

(32) In a uniform medium the pressure due to a source S

A^
P~47r r'

At another source S" at a distance r we shall have

k

if p' be the pressure at S due to &. If therefore there be two systems of
sources 2(<S') and S(6"), and if the pressures due to the first be p and to the
second p', then

For every term &p has a term Sp' equal to it.

ON FARADAY'S LINES OF FORCE. 175

(33) Suppose that in a uniform medium the motion of the fluid is every-
where parallel to one plane, then the surfaces of equal pressure will be
perpendicular to this plane. If we take two parallel planes at a distance equal
to k from each other, we can divide the space between these planes into unit
tubes by means of cylindric surfaces perpendicular to the planes, and these
together with the surfaces of equal pressure will divide the space into cells of
which the length is equal to the breadth. For if h be the distance between
consecutive surfaces of equal pressure and s the section of the unit tube, we
have by (13) s = kh.

But s is the product of the breadth and depth ; but the depth is k,
therefore the breadth is h and equal to the length.

If two systems of plane curves cut each other at right angles so as to
divide the plane into little areas of which the length and breadth are equal,
then by taking another plane at distance k from the first and erecting
cylindric surfaces on the plane curves as bases, a system of cells will be
formed which will satisfy the conditions whether we suppose the fluid to run
along the first set of cutting lines or the second*.

Application of the Idea of Lines of Force.

I have now to shew how the idea of lines of fluid motion as described
above may be modified so as to be applicable to the sciences of statical elec-
tricity, permanent magnetism, magnetism of induction, and uniform galvanic
currents, reserving the laws of electro-magnetism for special consideration.

I shall assume that the phenomena of statical electricity have been already
explained by the mutual action of two opposite kinds of matter. If we consider
one of these as positive electricity and the other as negative, then any two
particles of electricity repel one another with a force which is measured by the
product of the masses of the particles divided by the square of their distance.

Now we found in (18) that the velocity of our imaginary fluid due to a
source S at a distance r varies inversely as r*. Let us see what will be the
effect of substituting such a source for every particle of positive electricity. The
velocity due to each source would be proportional to the attraction due to the .
corresponding particle, and the resultant velocity due to all the sources would

* See Cambridge and Dublin Mathematical Journal, Vol. HI. p. 286.

176 ON FARADAY'S LINES OF FORCE.

be proportional to the resultant attraction of all the particles. Now we may find
the resultant pressure at any point by adding the pressures due to the given
sources, and therefore we may find the resultant velocity in a given direction
from the rate of decrease of pressure in that direction, and this will be
proportional to the resultant attraction of the particles resolved in that direction.

Since the resultant attraction in the electrical problem is proportional to
the decrease of pressure in the imaginary problem, and since we may select
any values for the constants in the imaginary problem, we may assume that the
resultant attraction in any direction is numerically equal to the decrease of
pressure in that direction, or

dp
A- ~dx-

By this assumption we find that if V be the potential,

d V= Xdx + Ydy + Zdz = - dp,

or since at an infinite distance F=0 and p = Q, V= —p.
In the electrical problem we have

In the fluid p = Z ;

^j-

tC

If k be supposed very great, the amount of fluid produced by each source
in order to keep up the pressures will be very small.

The potential of any system of electricity on itself will be

, .

If £ (dm), 2 (dm') be two systems of electrical particles and p, p' the potentials
due to them respectively, then by (32)

or the potential of the first system on the second is equal to that of the second
system on the first.

ON FAHADAY'S LINES OF FORCE. 177

So that in the ordinary electrical problems the analogy in fluid motion is
of this kind :

V=-p,

X= — T-=ku,
ax

dm = — S,
4ir

k
whole potential of a system = —"S,Vdm =—- W, where W is the work done by

the fluid in overcoming resistance.

The lines of forces are the unit tubes of fluid motion, and they may be
estimated numerically by those tubes.

Theory of Dielectrics.

The electrical induction exercised on a body at a distance depends not
only on the distribution of electricity in the inductric, and the form and posi-
tion of the inducteous body, but on the nature of the interposed medium, or
dielectric. Faraday* expresses this by the conception of one substance having
a greater inductive capacity, or conducting the lines of inductive action more
freely than another. If we suppose that in our analogy of a fluid in a resisting
medium the resistance is different in different media, then by making the
resistance less we obtain the analogue to a dielectric which more easily conducts
Faraday's lines.

It is evident from (23) that in this case there will always be an apparent
distribution of electricity on the surface of the dielectric, there being negative
electricity where the lines enter and positive electricity where they emerge. In
the case of the fluid there are no real sources on the surface, but we use
them merely for purposes of calculation. In the dielectric there may be no
real charge of electricity, but only an apparent electric action due to the surface.

If the dielectric had been of less conductivity than the surrounding medium,
we should have had precisely opposite effects, namely, positive electricity where
lines enter, and negative where they emerge.

* Series zi.
VOL. I. 23

178 ON FARADAY'S LINES OP FORCE.

If the conduction of the dielectric is perfect or nearly so for the small
quantities of electricity with which we have to do, then we have the case of
(.24). The dielectric is then considered as a conductor, its surface is a surface
of equal potential, and the resultant attraction near the surface itself is per-
pendicular to it

TJieory of Permanent Magnets.

A magnet is conceived to be made up of elementary magnetized particles,
each of which has its own north and south poles, the action of which upon
other north and south poles is governed by laws mathematically identical with
those of electricity. Hence the same application of the idea of lines of force
can be made to this subject, and the same analogy of fluid motion can be
employed to illustrate it.

But it may be useful to examine the way in which the polarity of the
elements of a magnet may be represented by the unit cells in fluid motion.
In each unit cell unity of fluid enters by one face and flows out by the opposite
face, so that the first face becomes a unit sink and the second a unit source
with respect to the rest of the fluid. It may therefore be compared to an
elementary magnet, having an equal qxiantity of north and south magnetic
matter distributed over two of its faces. If we now consider the cell as forming
part of a system, the fluid flowing out of one cell will flow into the next, and
so on, so that the source will be transferred from the end of the cell to the
end of the unit tube. If all the unit tubes begin and end on the bounding
surface, the sources and sinks will be distributed entirely on that surface, and in
the case of a magnet which has what has been called a solenoidal or tubular
distribution of magnetism, all the imaginary magnetic matter will be on the
surface*.

Ttieory of Paramagnetic and Diamagnetic Induction.

Faraday t has shewn that the effects of paramagnetic and diamagnetic bodies
in the magnetic field may be explained by supposing paramagnetic bodies to

* See Professor Thomson On the Mathematical Theory of Magnetism, Chapters in. and v. Phil.
Trant. 1851.

t Experimental ReteanJut (3292).

ON FARADAY'S LINES OF FORCE. 179

conduct the lines of force better, and diamagnetic bodies worse, than the
surrounding medium. By referring to (23) and (26), and supposing sources to
represent north magnetic matter, and sinks south magnetic matter, then if a
paramagnetic body be in the neighbourhood of a north pole, the lines of force
on entering it will produce south magnetic matter, and on leaving it they will
produce an equal amount of north magnetic matter. Since the quantities of
magnetic matter on the whole are equal, but the southern matter is nearest
to the north pole, the result will be attraction. If on the other hand the body
be diamagnetic, or a worse conductor of lines of force than the surrounding
medium, there will be an imaginary distribution of northern magnetic matter
where the lines pass into the worse conductor, and of southern where they pass
out, so that on the whole there will be repulsion.

We may obtain a more general law from the consideration that the poten-
tial of the whole system is proportional to the amount of work done by the
fluid in overcoming resistance. The introduction of a second medium increases
or diminishes the work done according as the resistance is greater or less than
that of the first medium. The amount of this increase or diminution will vary
as the square of the velocity of the fluid.

Now, by the theory of potentials, the moving force in any direction is
measured by the rate of decrease of the potential of the system in passing along
that direction, therefore when k', the resistance within the second medium, is
greater than k, the resistance in the surrounding medium, there is a force tend-
ing from places where the resultant force v is greater to where it is less, so
that a diamagnetic body moves from greater to less values of the resultant
force *.

In paramagnetic bodies k' is less than k, so that the force is now from
points of less to points of greater resultant magnetic force. Since these results
depend only on the relative values of k and k', it is evident that by changing
the surrounding medium, the behaviour of a body may be changed from para-
magnetic to diamagnetic at pleasure.

It is evident that we should obtain the same mathematical results if we
had supposed that the magnetic force had a power of exciting a polarity in
bodies which is in the same direction as the lines in paramagnetic bodies, and

* Experimental Kesearches (2797), (2798). See Thomson, Cambridge and Dublin Matiiemntical
Journal, May, 1847.

23—2

L00 ON FARADAY'S LINES OF FORCE.

in the reverse direction in diamagnetic bodies*. In fact we have not as yet
come to any facts which would lead us to choose any one out of these three
theories, that of lines of force, that of imaginary magnetic matter, and that of
induced polarity. As the theory of lines of force admits of the most precise,
and at the same time least theoretic statement, we shall allow it to stand for
the present.

Theory of Magnecrystallic Induction.

The theory of Faradayt with respect to the behaviour of crystals in the
magnetic field may be thus stated. In certain crystals and other substances the
lines of magnetic force are conducted with different facility in different directions.
The body when suspended in a uniform magnetic field will turn or tend to turn
into such a position that the lines of force shall pass through it with least resist-
ance. It is not difficult by means of the principles in (28) to express the laws
of this kind of action, and even to reduce them in certain cases to numerical
formulae. The principles of induced polarity and of imaginary magnetic matter
are here of little use ; but the theory of lines of force is capable of the most
perfect adaptation to this class of phenomena.

Theory of the Conduction of Current Electricity.

It is in the calculation of the laws of constant electric currents that the
theory of fluid motion which we have laid down admits of the most direct appli-
cation. In addition to the researches of Ohm on this subject, we have those
of M. Kirchhoff, Ann. de Chim. XLI. 496, and of M. Quincke, XLvn. 203, on the
Conduction of Electric Currents in Plates. According to the received opinions
we have here a current of fluid moving uniformly in conducting circuits, which
oppose a resistance to the current which has to be overcome by the application
of an electro-motive force at some part of the circuit. On account of this
resistance to the motion of the fluid the pressure must be different at different
points in the circuit. This pressure, which is commonly called electrical tension,

* Exp. Re$. (2429), (3320). See Weber, Poggendorff, LXXXVII. p. 145. Pro£ TyndaU, Phil.
Trans. 1856, p. 237.

t Exp. Ret. (2836), <kc.

ON FARADAY'S LINES OF FORCE. 181

is found to be physically identical with the potential in statical electricity, and
thus we have the means of connecting the two sets of phenomena. If we knew
what amount of electricity, measured statically, passes along that current which
we assume as our unit of current, then the connexion of electricity of tension
with current electricity would be completed*. This has as yet been done only
approximately, but we know enough to be certain that the conducting powers of
different substances differ only in degree, and that the difference between glass
and metal is, that the resistance is a great but finite quantity in glass, and a
small but finite quantity in metal. Thus the analogy between statical electricity
and fluid motion turns out more perfect than we might have supposed, for there
the induction goes on by conduction just as in current electricity, but the quan-
tity conducted is insensible owing to the great resistance of the dielectrics t.

On Electro-motive Forces.

When a uniform current exists in a closed circuit it is evident that some
other forces must act on the fluid besides the pressures. For if the current
were due to difference of pressures, then 'it would flow from the point of
greatest pressure in both directions to the point of least pressure, whereas in
reality it circulates in one direction constantly. We must therefore admit the
existence of certain forces capable of keeping up a constant current in a closed
circuit. Of these the most remarkable is that which is produced by chemical
action. A cell of a voltaic battery, or rather the surface of separation of the
fluid of the cell and the zinc, is the seat of an electro-motive force which
can maintain a current in opposition to the resistance of the circuit. If we
adopt the usual convention in speaking of electric currents, the positive current
is from the fluid through the platinum, the conducting circuit, and the zinc,
back to the fluid again. If the electro-motive force act only in the surface of
separation of the fluid and zinc, then the tension of electricity in the fluid
must exceed that in the zinc by a quantity depending on the nature and
length of the circuit and on the strength of the current in the conductor.
In order to keep up this difference of pressure there must be an electro-motive
force whose intensity is measured by that difference of pressure. If F be the
electro-motive force, / the quantity of the current or the number of electrical

* See Exp. Bet. (371). t Exp. Res. VoL ra. p. 513.

182 ON FARADAY'S LINES OF FORCE.

units delivered in unit of time, and K a quantity depending on the length
and resistance of the conducting circuit, then

F=IK=p-p',
where p is the electric tension in the fluid and p' in the zinc.

If the circuit be broken at any point, then since there is no current the
tension of the part which remains attached to the platinum will be p, and
that of the other will be p', p—p' or F affords a measure of the intensity
of the current. This distinction of quantity and intensity is very useful *,
but must be distinctly understood to mean nothing more than this : — The
quantity of a current is the amount of electricity which it transmits in unit
of time, and is measured by / the number of unit currents which it contains.
The intensity of a current is its power of overcoming resistance, and is
measured by F or IK, where K is the resistance of the whole circuit.

The same idea of quantity and intensity may be applied to the case of
magnetism f. The quantity of magnetization in any section of a magnetic
body is measured by the number of lines of magnetic force which pass through
it. The intensity of magnetization in the section depends on the resisting
power of the section, as well as on the number of lines which pass through
it. If k be the resisting power of the material, and S the area of the section,
and / the number of lines of force which pass through it, then the whole
intensity throughout the section

F Tk
= f=IS'

When magnetization is produced by the influence of other magnets only,
we may put p for the magnetic tension at any point, then for the whole
magnetic solenoid

When a solenoidal magnetized circuit returns into itself, the magnetization
does not depend on difference of tensions only, but on some magnetizing force
of which the intensity is F.

If i be the quantity of the magnetization at any point, or the number of
lines of force passing through unit of area in the section of the solenoid, then

* Exp. Het. Vol. in. p. 519. t Exp. Res. (2870), (3203).

ON FARADAY'S LINES OP FORCE. 183

the total quantity of magnetization in the circuit is the number of lines which
pass through any section, I=*S,idydz, where dydz is the element of the section,
and the summation is performed over the whole section.

The intensity of magnetization at any point, or the force required to
keep up the magnetization, is measured by ki—f, and the total intensity of
magnetization in the circuit is measured by the sum of the local intensities all
round the circuit,

where dx is the element of length in the circuit, and the summation is extended
round the entire circuit.

In the same circuit we have always F=IK, where K is the total resistance
of the circuit, and depends on its form and the matter of which it is
composed.

On the Action of closed Currents at a Distance.

The mathematical laws of the attractions and repulsions of conductors have
been most ably investigated by Ampere, and his results have stood the test of
subsequent experiments.

From the single assumption, that the action of an element of one current
upon an element of another current is an attractive or repulsive force acting
in the direction of the line joining the two elements, he has determined by
the simplest experiments the mathematical form of the law of attraction, and
has put this law into several most elegant and useful forms. We must
recollect however that no experiments have been made on these elements of
currents except under the form of closed currents either in rigid conductors
or in fluids, and that the laws of closed currents can only be deduced from
such experiments. Hence if Ampere's formulae applied to closed currents give
true results, their truth is not proved for elements of currents unless we
assume that the action between two such elements must be along the line which
joins them. Although this assumption is most warrantable and philosophical in
the present state of science, it will be more conducive to freedom of investi-
gation if we endeavour to do without it, and to assume the laws of closed currents
as the ultimate datum of experiment.

184 ON FARADAY'S LINES OF FORCE.

Ampere has shewn that when currents are combined according to the law
of the parallelogram of forces, the force due to the resultant current is the
resultant of the forces due to the component currents, and that equal and
opposite currents generate equal and opposite forces, and when combined
neutralize each other.

He has also shewn that a closed circuit of any form has no tendency to
turn a moveable circular conductor about a fixed axis through the centre of
the circle perpendicular to its plane, and that therefore the forces in the case
of a closed circuit render Xdx+ Ydy + Zdz a complete differential.

Finally, he has shewn that if there be two systems of circuits similar
and similarly situated, the quantity of electrical current in corresponding
conductors being the same, the resultant forces are equal, whatever be the
absolute dimensions of the systems, which proves that the forces are, cceteris
paribus, inversely as the square of the distance.

From these results it follows that the mutual action of two closed currents
whose areas are very small is the same as that of two elementary magnetic
bars magnetized perpendicularly to the plane of the currents.

The direction of magnetization of the equivalent magnet may be pre-
dicted by remembering that a current travelling round the earth from east
to west as the sun appears to do, would be equivalent to that magnetization
which the earth actually possesses, and therefore in the reverse direction to
that of a magnetic needle when pointing freely.

If a number of closed unit currents in contact exist on a surface, then at
all points in which two currents are in contact there will be two equal and
opposite currents which will produce no effect, but all round the boundary of the
surface occupied by the currents there will be a residual current not neutralized
by any other ; and therefore the result will be the same as that of a single
unit current round the boundary of all the currents.

From this it appears that the external attractions of a shell uniformly
magnetized perpendicular to its surface are the same as those due to a current
round its edge, for each of the elementary currents in the former case has
the same effect as an element of the magnetic shell.

If we examine the lines of magnetic force produced by a closed current,
we shall find that they form closed curves passing round the current and
embracing it, and that the total intensity of the magnetizing force all along
the closed line of force depends on the quantity of the electric current only.

ON FARADAY'S LINES OF FOKCE. 185

The number of unit lines* of magnetic force due to a closed current depends
on the form as well as the quantity of the current, but the number of unit
cells')- in each complete line of force is measured simply by the number of unit
currents which embrace it. The unit cells in this case are portions of space in
which unit of magnetic quantity is produced by unity of magnetizing force.
The length of a cell is therefore inversely as the intensity of the magnetizing
force, and its section inversely as the quantity of magnetic induction at that
point.

The whole number of cells due to a given current is therefore proportional
to the strength of the current multiplied by the number of lines of force
which pass through it. If by any change of the form of the conductors the
number of cells can be increased, there will be a force tending to produce that
change, so that there is always a force urging a conductor transverse to the
lines of magnetic force, so as to cause more lines of force ^o pass through the
closed circuit of which the conductor forms a part.

The number of cells due to two given currents is got by multiplying
the number of lines of inductive magnetic action which pass through each by
the quantity of the currents respectively. Now by (9) the number of lines
which pass through the first current is the sum of its own lines and those
of the second current which would pass through the first if the second current
alone were in action. Hence the whole number of cells will be increased by
any motion which causes more lines of force to pass through either circuit,
and therefore the resultant force will tend to produce such a motion, and the
work done by this force during the motion will be measured by the number
of new cells produced. All the actions of closed conductors on each other may
be deduced from this principle.

On Electric Currents produced by Induction.

Faraday has shewnj that when a conductor moves transversely to the lines
of magnetic force, an electro-motive force arises in the conductor, tending to
produce a current in it. If the conductor is closed, there is a continuous
current, if open, tension is the result. If a closed conductor move transversely
to the lines of magnetic induction, then, if the number of lines which pass

* Exp. Res. (3122). See Art. (6) of this paper. t Art. (13).

J Exp. Res. (3077), &c.
VOL. I. 24

186 ON FARADAY'S LINES OF FORCE.

through it does not change during the motion, the electro-motive forces in the
circuit will be in equilibrium, and there will be no current. Hence the electro-
motive forces depend on the number of lines which are cut by the conductor
during the motion. If the motion be such that a greater number of lines pass
through the circuit formed by the conductor after than before the motion,
then the electro-motive force will be measured by the increase of the number
of lines, and will generate a current the reverse of that which would have
produced the additional lines. When the number of lines of inductive magnetic
action through the circuit is increased, the induced current will tend to diminish
the number of lines, and when the number is diminished the induced current
will tend to increase them.

That this is the true expression for the law of induced currents is shewn
from the fact that, in whatever way the number of lines of magnetic induction
passing through the circuit be increased, the electro-motive effect is the same,
whether the increase take place by the motion of the conductor itself, or of other
conductors, or of magnets, or by the change of intensity of other currents, or
by the magnetization or demagnetization of neighbouring magnetic bodies, or
lastly by the change of intensity of the current itself.

In all these cases the electro-motive force depends on the cliange in the
number of lines of inductive magnetic action which pass through the circuit*.

* The electro-magnetic forces, which tend to produce motion of the material conductor, must be
carefully distinguished from the electro-motive forces, which tend to produce electric currents.

Let an electric current be passed through a maas of metal of any form. The distribution of
the currents within the metal will be determined by the laws of conduction. Now let a constant
electric current be passed through another conductor near the first If the two currents are in the
same direction the two conductors will be attracted towards each other, and would come nearer if
not held in their positions. But though the material conductors are attracted, the currents (which
are free to choose any course within the metal) will not alter their original distribution, or incline
towards each other. For, since no change takes place in the system, there will be no electro-motive
forces to modify the original distribution of currents.

In this case we hare electro-magnetic forces acting on the material conductor, without any
electro-motive forces tending to modify the current which it carries.

Let us take as another example the case of ft linear conductor, not forming a closed circuit,
and let it be made to traverse the lines of magnetic force, either by its own motion, or by changes
in the magnetic field. An electro-motive force will act in the direction of the conductor, and, as it
cannot produce a current, because there is no circuit, it will produce electric tension at the extremi-
ties. There will be no electro-magnetic attraction on the material conductor, for this attraction
depends on the existence of the current within it, and this is prevented by the circuit not being closed.

Here then we have the opposite case of an electro-motive force acting on the electricity in the
conductor, but no attraction on its material particles.

ON FARADAY'S LINES OF FORCE. 187

It is natural to suppose that a force of this kind, which depends on a
change in the number of lines, is due to a change of state which is measured
by the number of these lines. A closed conductor in a magnetic field may
be supposed to be in a certain state arising from the magnetic action.
As long as this state remains unchanged no effect takes place, but, when the
state changes, electro-motive forces arise, depending as to their intensity and
direction on this change of state. I cannot do better here than quote a
passage from the first series of Faraday's Experimental Researches, Art. (60).

" While the wire is subject to either volta-electric or magno-electric
induction it appears to be in a peculiar state, for it resists the formation of
an electrical current in it ; whereas, if in its common condition, such a current
would be produced; and when left uninfluenced it has the power of originating a
current, a power which the wire does not possess under ordinary circumstances.
This electrical condition of matter has not hitherto been / recognised, but it
probably exerts a very important influence in many if not most of the phe-
nomena produced by currents of electricity. For reasons which will immediately
appear (7) I have, after advising with several learned friends, ventured to
designate it as the electro-tonic state." Finding that all the phenomena could
be otherwise explained without reference to the electro-tonic state, Faraday in
his second series rejected it as not necessary ; but in his recent researches *
he seems still to think that there may be some physical truth in his
conjecture about this new state of bodies.

The conjecture of a philosopher so familiar with nature may sometimes be
more pregnant with truth than the best established experimental law disco-
vered by empirical inquirers, and though not bound to admit it as a physical
truth, we may accept it as a new idea by which our mathematical conceptions
may be rendered clearer.

In this outline of Faraday's electrical theories, as they appear from a
mathematical point of view, I can do no more than simply state the mathe-
matical methods by which I believe that electrical phenomena can be best
comprehended and reduced to calculation, and my aim has been to present the
mathematical ideas to the mind in an embodied form, as systems of lines or
surfaces, and not as mere symbols, which neither convey the same ideas, nor
readily adapt themselves to the phenomena to be explained. The idea of the
electro-tonic state, however, has not yet presented itself to my mind in such a

* (3172) (3269).

24—2

188 ON FARADAY'S LINES OF FORCE.

form that its nature and properties may be clearly explained without reference
to mere symbols, and therefore I propose in the following investigation to use
symbols freely, and to take for granted the ordinary mathematical operations.
By a careful study of the laws of elastic solids and of the motions of viscous
fluids, I hope to discover a method of forming a mechanical conception of this
electro-tonic state adapted to general reasoning*.

PART II.
On Faraday's "Electro-tonic State."

When a conductor moves in the neighbourhood of a current of electricity,
or of a magnet, or when a current or magnet near the conductor is moved, or
altered in intensity, then a force acts on the conductor and produces electric
tension, or a continuous current, according as the circuit is open or closed. This
current is produced only by changes of the electric or magnetic phenomena sur-
rounding the conductor, and as long as these are constant there is no observed
effect on the conductor. Still the conductor is in different states when near a
current or magnet, and when away from its influence, since the removal or
destruction of the current or magnet occasions a current, which would not have
existed if the magnet or current had not been previously in action.

Considerations of this kind led Professor Faraday to connect with his
discovery of the induction of electric currents the conception of a state into
which all bodies are thrown by the presence of magnets and currents. This
state does not manifest itself by any known phenomena as long as it is undis-
turbed, but any change in this state is indicated by a current or tendency
towards a current. To this state he gave the name of the " Electro-tonic
State," and although he afterwards succeeded in explaining the phenomena
which suggested it by means of less hypothetical conceptions, he has on several
occasions hinted at the probability that some phenomena might be discovered
which would render the electro-tonic state an object of legitimate induction.
These speculations, into which Faraday had been led by the study of laws
which he has well established, and which he abandoned only for want of experi-

* See Prot W. Thomson On a Mechanical Representation of Electric, Magnetic and Galvanic
Forces. Camb. and Dub. Math. Jour. Jan. 1847.

ON FARADAY'S LINES OF FORCE. 189

mental data for the direct proof of the unknown state, have not, I think, been
made the subject of mathematical investigation. Perhaps it may be thought
that the quantitative determinations of the various phenomena are not suffi-
ciently rigorous to be made the basis of a mathematical theory ; Faraday,
however, has not contented himself with simply stating the numerical results of
his experiments and leaving the law to be discovered by calculation. Where
he has perceived a law he has at once stated it, in terms as. unambiguous as
those of pure mathematics ; and if the mathematician, receiving this as a physical
truth, deduces from it other laws capable of being tested by experiment, he
has merely assisted the physicist in arranging his own ideas, which is con-
fessedly a necessary step in scientific induction.

In the following investigation, therefore, the laws established by Faraday
will be assumed as true, and it will be shewn that by following out his
speculations other and more general laws can be deduced, from them. If it
should then appear that these laws, originally devised to include one set of
phenomena, may be generalized so as to extend to phenomena of a different
class, these mathematical connexions may suggest to physicists the means of
establishing physical connexions ; and thus mere speculation may be turned to
account in experimental science.

On Quantity and Intensity as Properties of Electric Currents.

It is found that certain effects of an electric current are equal at what-
ever part of the circuit they are estimated. The quantities of water or of
any other electrolyte decomposed at two different sections of the same circuit,
are always found to be equal or equivalent, however different the material and
form of the circuit may be at the two sections. The magnetic effect of a
conducting wire is also found to be independent of the form or material of
the wire in the same circuit. There is therefore an electrical effect which is
equal at every section of the circuit. If we conceive of the conductor as the
channel along which a fluid is constrained to move, then the quantity of fluid
transmitted by each section will be the same, and we may define the quantity
of an electric current to be the quantity of electricity which passes across a
complete section of the current in unit of time. We may for the present
measure quantity of electricity by the quantity of water which it would decom-
pose in unit of time.

190 ON FARADAY'S LINES OF FORCE.

In order to express mathematically the electrical currents in any conductor,
we must have a definition, not only of the entire flow across a complete section,
but also of the flow at a given point in a given direction.

DEF. The quantity of a current at a given point and in a given direction
is measured, when uniform, by the quantity of electricity which flows across
unit of area taken at that point perpendicular to the given direction, and when
variable by the quantity which would flow across this area, supposing the flow
uniformly the same as at the given point.

In the following investigation, the quantity of electric current at the point
(xyz) estimated in the directions of the axes x, y, z respectively will be denoted
by a,, 6,, ct.

The quantity of electricity which flows in unit of time through the ele-
mentary area dS

= dS (lat + mbt + nct),
where I, m, n are the direction-cosines of the normal to dS.

This flow of electricity at any point of a conductor is due to the electro-
motive forces which act at that point. These may be either external or internal.

External electro- motive forces arise either from the relative motion of currents
and magnets, or from changes in their intensity, or from other causes acting
at a distance.

Internal electro-motive forces arise principally from difference of electric
tension at points of the conductor in the immediate neighbourhood of the point
in question. The other causes are variations of chemical composition or of tem-
perature in contiguous parts of the conductor.

Let pt represent the electric tension at any point, and Xv Yt, Zt the sums
of the parts of all the electro-motive forces arising from other causes resolved
parallel to the co-ordinate axes, then if a,, /3,, ya be the effective electro-motive
forces

dx

(A).

ON FARADAY'S LINES OF FORCE. 191

Now the quantity of the current depends on the electro-motive force and
on the resistance of the medium. If the resistance of the medium be uniform
in all directions and equal to &„

a, = ^aa, & = £A, y2 = &A .......................... (B),

but if the resistance be different in different directions, the law will be more
complicated.

These quantities Oj, &, y, may be considered as representing the intensity
of the electric action in the directions of x, y, z.

The intensity measured along an element da- of a curve is given by

c = la + mfi + ny,
where Z, m, n are the direction-cosines of the tangent.

The integral \tda- taken with respect to a given portion of a curve line,
represents the total intensity along that line. If the curve is a closed one, it
represents the total intensity of the electro-motive force in the closed curve.

Substituting the values of a, $, y from equations (A)
\tdo- = \(Xdx + Ydy + Zdz) -p + C.

If therefore (Xdx+ Ydy + Zdz) is a complete differential, the value of jecZo- for
a closed curve will vanish, and in all closed curves

JecZo- = \(Xdx + Ydy + Zdz),

the integration being effected along the curve, so that in a closed curve the
total intensity of the effective electro-motive force is equal to the total intensity
of the impressed electro-motive force.

The total quantity of conduction through any surface is expressed by

where

e = la + mb + nc,

I, TO, n being the direction-cosines of the normal,

. '. \edS = \\adydz + \\bdzdx + \\cdxdy >

the integrations being effected over the given surface. When the surface is a
closed one, then we may find by integration by parts

192 ON FARADAY'S LINES OF FORCE.

If we make

da db dc

\edS=lir\\\pdxdydz,

where the integration on the right side of the equation is effected over every
part of space within the surface. In a large class of phenomena, including all
cases of uniform currents, the quantity p disappears.

Magnetic Quantity and Intensity.

From his study of the lines of magnetic force, Faraday has been led to
the conclusion that in the tubular surface* formed by a system of such lines,
the quantity of magnetic induction across any section of the tube is constant,
and that the alteration of the character of these lines in passing from one
substance to another, is to be explained by a difference of inductive capacity
in the two substances, which is analogous to conductive power in the theory
of electric currents.

In the following investigation we shall have occasion to treat of magnetic
quantity and intensity in connection with electric. In such cases the magnetic
symbols will be distinguished by the suffix 1, and the electric by the suffix 2.
The equations connecting a, b, c, k, a, ft, y, p, and p, are the same in form as
those which we have just given, a, b, c are the symbols of magnetic induction
with respect to quantity; k denotes the resistance to magnetic induction, and
may be different in different directions ; a, ft, y, are the effective magnetizing
forces, connected with a, b, c, by equations (B) ; p is the magnetic tension or
potential which will be afterwards explained ; p denotes the density of real
magnetic matter and is connected with a, b, c by equations (C). As all the
details of magnetic calculations will be more intelligible after the exposition of the
connexion of magnetism with electricity, it will be sufficient here to say that
all the definitions of total quantity, with respect to a surface, the total intensity
to a curve, apply to the case of magnetism as well as to that of electricity.

* Exp. Res. 3271, definition of " Sphondyloid."

ON FARAD AYS LINES OF FORCE. 193

Electro-magnetism.

Ampere has proved the following laws of the attractions and repulsions of
electric currents :

I. Equal and opposite currents generate equal and opposite forces.

II. A crooked current is equivalent to a straight one, provided the two
currents nearly coincide throughout their whole length.

III. Equal currents traversing similar and similarly situated closed curves
act with equal forces, whatever be the linear dimensions of the circuits.

IV. A closed current exerts no force tending to turn a circular conductor
about its centre.

It is to be observed, that the currents with which Ampere worked were constant
and therefore re-entering. All his results are therefore deduced from experiments
on closed currents, and his expressions for the mutual action of the elements
of a current involve the assumption that this action is exerted in the direction
of the line joining those elements. This assumption is no doubt warranted by the
universal consent of men of science in treating of attractive forces considered
as due to the mutual action of particles ; but at present we are proceeding
on a different principle, and searching for the explanation of the phenomena,
not in the currents alone, but also in the surrounding medium.

The first and second laws shew that currents are to be combined like
velocities or forces.

The third law is the expression of a property of all attractions which may
be conceived of as depending on the inverse square of the distance from a fixed
system of points ; and the fourth shews that the electro-magnetic forces may
always be reduced to the attractions and repulsions of imaginary matter properly
distributed.

In fact, the action of a very small electric circuit on a point in its neigh-
bourhood is identical with that of a small magnetic element on a point outside
it. If we divide any given portion of a surface into elementary areas, and
cause equal currents to flow in the same direction round all these little areas,
the effect on a point not in the surface will be the same as that of a shell
coinciding with the surface, and uniformly magnetized normal to its surface.
But by the first law all the currents forming the little circuits will destroy

VOL. I. 25

104 ON FARADAY'S LINKS OF FORCE.

one another, and leave a single current running round the bounding line. So
that the magnetic effect of a uniformly magnetized shell is equivalent to that
of an electric current round the edge of the shell. If the direction of the current
coincide with that of the apparent motion of the sun, then the direction of
magnetization of the imaginary shell will be the same as that of the real mag-
netization of the earth*.

The total intensity of magnetizing force in a closed curve passing through
and embracing the closed current is constant, and may therefore be made a
measure of the quantity of the current. As this intensity is independent of the
form of the closed curve and depends only on the quantity of the current which
passes through it, we may consider the elementajry case of the current which
flows through the elementary area dydz.

Let the axis of x point towards the west, z towards the south, and y
upwards. Let x, y, z be the coordinates of a point in the middle of the area

, then the total intensity measured round the four sides of the element is

ffe

-

Total intensity = - dy dz.

The quantity of electricity conducted through the elementary area dydz is
".'/'/</:, and therefore if we define the measure of an electric current to be the
total intensity of magnetizing force in a closed curve embracing it, we shall have

_dfr_dyi
^~ dz dy '
, dy, da,

=~

c=_
dy dx '

* See Experimental Researches (3205) for the relations between the electrical and magnetic circuit.
considered as mutually embracing curves.

ON FARADAY'S LINES OF FORCE. 195

These equations enable us to deduce the distribution of the currents of
electricity whenever we know the values of a, ft, y, the magnetic intensities.
If a, /8, y be exact differentials of a function of x, y, z with respect to x, y
and z respectively, then the values of a2, 63> ca disappear ; and we know that the
magnetism is not produced by electric currents in that part of the field which
we are investigating. It is due either to the presence of permanent magnetism
within the field, or to magnetizing forces due to external causes.

We may observe that the above equations give by differentiation

=
dx dy dz ~

which is the equation of continuity for closed currents. Our investigations are
therefore for the present limited to closed currents ; and we know little of the
magnetic effects of any currents which are not closed.

Before entering on the calculation of these electric and magnetic states it
may be advantageous to state certain general theorems, the truth of which may
be established analytically.

THEOREM I.
The equation

<FV d'V d'V

(where V and p are functions of x, y, z never infinite, and vanishing for all points
at an infinite distance), can be satisfied by one, and only one, value of V. See
Art. (17) above.

THEOREM II.

The value of V which will satisfy the above conditions is found by inte-
grating the expression

pdxdydz

f/f_

JJJ(x-x

where the limits of x, y, z are such as to include every point of space where p
is finite.

25—2

196 ON FARADAY'S LINES OF FORCE.

The proofs of these theorems may be found in any work on attractions or
electricity, and in particular in Green's Essay on the Application of Mathematics
to Electricity. See Arts. 18, 19 of this paper. See also Gauss, on Attractions,
translated in Taylor's Scientific Memoirs.

THEOREM III.
Let U and V be two functions of x, y, z, then

ePF\ , . , dUdV dUdV dU d

where the integrations are supposed to extend over all the space in which U
and V have values differing from 0. — (Green, p. 10.)

This theorem shews that if there be two attracting systems the actions
between them are equal and opposite. And by making U= V we find that
the potential of a system on itself is proportional to the integral of the square
of the resultant attraction through all space ; a result deducible from Art. (30),
since the volume of each cell is inversely as the square of the velocity (Arts.
12, 13), and therefore the number of cells in a given space is directly as the
square of the velocity.

THEOREM IV.

Let a, /8, y, p be quantities finite through a certain space and vanishing
in the space beyond, and let k be given for all parts of space as a continuous
or discontinuous function of x, y, z, then the equation in p

d 1 / dp\ , d 1 (a dp\ , d 1 / dp

a - -f- +-j- T p - -f- ) + -r T \y-~r

dx k \ dx] dy k \ dy) dz k \r dz

J_

+ 47rP

has one, and only one solution, in which p is always finite and vanishes at
an infinite distance.

The proof of this theorem, by Prof. W. Thomson, may be found in the
Cambridge and Dublin Mathematical Journal, Jan. 1848.

ON FARADAY'S LINES OF FORCE. 197

If a, /?, y be the electro-motive forces, p the electric tension, and k the
coefficient of resistance, then the above equation is identical with the equation
of continuity

da, ,dbtdc,

j -- r -j — r -y- + 4;rp = 0 ;
ax ay dz

and the theorem shews that when the electro-motive forces and the rate of
production of electricity at every part of space are given, the value of the
electric tension is determinate.

Since the mathematical laws of magnetism are identical with those of elec-
tricity, as far as we now consider them, we may regard a, /?, y as magnetizing
forces, p as magnetic tension, and p as real magnetic density, k being the
coefficient of resistance to magnetic induction.

The proof of this theorem rests on the determination of/ the minimum value
of

where V is got from the equation

dfV

and p has to be determined.

The meaning of this integral in electrical language may be thus brought
out. If the presence of the media in which k has various values did not
affect the distribution of forces, then the "quantity" resolved in x would be

dV dV

simply -j— and the intensity k -y— . But the actual quantity and intensity are

y- (a.— -/-} and a— -/-, and the parts due to the distribution of media alone
K \ ax/ ax

are therefore

k \ dx) dx ' dx dx '

Now the product of these represents the work done on account of this
distribution of media, the distribution of sources being determined, and taking
in the terms in y and z we get the expression Q for the total work done

198 ON FARADAY'S LINES OF FORCE.

by that part of the whole effect at any point which is due to the distribution
of conducting media, and not directly to the presence of the sources.

This quantity Q is rendered a minimum by one and only one value of p,
namely, that which satisfies the original equation.

THEOREM V.

If a, b, c be three functions of x, y, z satisfying the equation

da db dc _
dx dy dz~

it is always possible to find three functions a, ft, y which shall satisfy the equa-

tions

dfl _dy =
dz dy

dy-^ = b

dx dz

_
dy dx

Let A =• \cdy, where the integration is to be performed upon c considered
as a function of y, treating x and z as constants. Let B*=\adz, C = \bdx,
A' = $bdz, R = \cdx, C'*=\ady, integrated in the same way.

Then

will satisfy the given equations ; for

dB dy [da , (dc , (db , (da
- = dz-dx-dx+

and 0 i

dB dy (da , (da , (da
•'• ~j — ~T~ = I ~:r. dx + I -j- dy + | -y-

= 0.

ON FARADAY'S LINES OF FORCE.

199

In the same way it may be shewn that the values of a, ft, y satisfy
the other given equations. The function \j/ may be considered at present as
perfectly indeterminate.

The method here given is taken from Prof. W. Thomson's memoir on
Magnetism (Phil. Trans. 1851, p. 283).

As we cannot perform the required integrations when a, b, c are discon-
tinuous functions of x, y, z, the following method, which is perfectly general
though more complicated, may indicate more clearly the truth of the proposition.

Let A, B, C be determined from the equations

d'A d*A d*A

d'B d'B d*B

d'C d'C d'C

T-j + -r-r + -TT+c = 0,

aar dtf dz

by the methods of Theorems I. and II., so that A, B, C are never infinite,
and vanish when x, y, or z is infinite.

Also let

_ dB dC dty

~ dz dy dx '

n_dC dA d\ji
P=dx~~dz+dy'

dA dB cfyr

<y — ^^ I *

dy dx dz '

then

+a.

If we find similar equations in y and z, and differentiate the first by x,
the second by y, and the third by z, remembering the equation between
a, b, c, we shall have

cP <P\ IdA dB d(

d/3 dy _ d IdA dB dC\ (d*A d'A d*A\
dz dy~ dx\dx dy dz) \dx* dy3 dz' /

d_ fdA_ dB d(

200 ON FARADAY'S LINES OP FORCE.

and since A, B, C are always finite and vanish at an infinite distance, the
only solution of this equation is

dA dB dC

and we have finally

dz cty

with two similar equations, shewing that a, ft, y have been rightly determined.
The function ^ is to be determined from the condition
da dft dy = (d* d?_ f*\ ..

if the left-hand side of this equation be always zero, i/» must be zero also.

THEOREM VI.

Let a, b, c be any three functions of x, y, z, it is possible to find three
functions a, /J, y and a fourth V, so that

dx dy dz ~

d0 dy dV
and a = ^ _ ^ +

dz ay dx

j _dy da dV
~ dx dz dy '

C:=_ +
dy dx dz

Let

da db dc

and let V be found from the equation

d'V

ON FARAD AYS LINES OF FORCE.

201

then

a =a —

dV
dx'

satisfy the condition

V 7, dV
' = b — -j— i

dy

,_ dV
C~°~ dz'

da' db' dc' A

-T- + -J-+ j- = 0;
ax dy dz

and therefore we can find three functions A, B, C, and from these a, ft, y, so as
to satisfy the given equations.

THEOREM VII.
The integral throughout infinity

Q = /// («i«i + &.A + c.yi) dxdydz,
where 0,6,0,, oj8,y, are any functions whatsoever, is capable of transformation into

Q = + JJJ{4"7>/>, - (o,aa + &b, + y0ca)} dxdydz,
in which the quantities are found from the equations

da, db, dc,

! + + y1
dx dy dz

afe determined from a,6,c, by the last theorem, so that

dfr_dK + dV.

dz dy dx '

a,6sCj are found from oj8,y, by the equations

and p is found from the equation

' ' '

VOL. I.

26

202 ON FARADAY'S LINES OF FORCE.

For, if we put a, in the form

dz dy dx '

and treat 6, and c, similarly, then we have by integration by parts through
infinity, remembering that all the functions vanish at the limits,

+»$-£)}***•

or Q=

and by Theorem III.

/// Vp' dxdydz = \\\ppdxdydz,
so that finally

Q = Hl^irpp - (a^a, + /S.6, + y0c,)} dxdydz.

If a,6,c, represent the components of magnetic quantity, and aj8,y, those
of magnetic intensity, then p will represent the real magnetic density, and p
the magnetic potential or tension. a,6,c, will be the components of quantity
of electric currents, and cg80y0 will be three functions deduced from ef,6,c,,
which will be found to be the mathematical expression for Faraday's Electro-
tonic state.

Let us now consider the bearing of these analytical theorems on the
theory of magnetism. Whenever we deal with quantities relating to magnetism,
we shall distinguish them by the suffix (,). Thus a,?>,c, are the components
resolved in the directions of x, y, z of the quantity of magnetic induction acting
through a given point, and Oj&y, are the resolved intensities of magnetization
at the same point, or, what is the same thing, the components of the force
which would be exerted on -a unit south pole of a magnet placed at that
point without disturbing the distribution of magnetism.

The electric currents are found from the magnetic intensities by the equations

When there are no electric currents, then

ON FARADAY'S LINES OF FORCE. 203

a perfect differential of a function of x, y, z. On the principle of analogy we
may call pl the magnetic tension.

The forces which act on a mass m of south magnetism at any point are

dpl dpl dpl

— m -P . — m -f- , and — m -4- .
ax ay dz

in the direction of the axes, and therefore the whole work done during any
displacement of a magnetic system is equal to the decrement of the integral

throughout the system.

Let us now call Q the total potential of the system on itself. The increase
or decrease of Q will measure the work lost or gained by any displacement
of any part of the system, and will therefore enable us to determine the
forces acting on that part of the system.

By Theorem III. Q may be put under the form

= + ITT 1 1 I (aA

in which Oj&yj are the differential coefficients of pt with respect to x, y, z
respectively.

If we now assume that this expression for Q is true whatever be the
values of Oj, /?„ ylt we pass from the consideration of the magnetism of permanent
magnets to that of the magnetic effects of electric currents, and we have then
by Theorem VII.

Q = 1 1 I Wi - -^ (o,«s + £A + r0cs) j- dxdydz.

So that in the case of electric currents, the components of the currents have
to be multiplied by the functions a,, ft,, y0 respectively, and the summations of
all such products throughout the system gives us the part of Q due to those
currents.

We have now obtained in the functions a,, /?„ •/„ the means of avoiding
the consideration of the quantity of magnetic induction which passes through
the circuit. Instead of this artificial method we have the natural one of con-
sidering the current with reference to quantities existing in the same space
with the current itself. To these I give the name of Electro-tonic functions, or
components of the Electro-tonic intensity.

2G— 2

204 ON FARADAY'S LINES OF FORCE.

Let us now consider the conditions of the conduction of the electric
currents within the medium during changes in the electro-tonic state. The
method which we shall adopt is an application of that given by Helmholtz in
his memoir on the Conservation of Force*.

Let there be some external source of electric currents which would generate
in the conducting mass currents whose quantity is measured by a,, 6,, c, and
their intensity by a,, /J,, yr

Then the amount of work due to this cause in the time dt is

^ /J/(aJa> + && + c»7») dxdydz
in the form of resistance overcome, and

tor dt I 1 1

in the form of work done mechanically by the electro-magnetic action of these
currents. If there be no external cause producing currents, then the quantity
representing the whole work done by the external cause must vanish, and we
have

dt 1 1 I (a,a, + 6^8, + c,y,) dxdydz + — ^ I J I (a^ + b& + c,y0) dxdydz,

where the integrals are taken through any arbitrary space. We must therefore
have

aao, + b& + c,y, = — jt (a.0,, + &,& + c,y.)

for every point of space ; and it must be remembered that the variation of
Q is supposed due to variations of a,, $,, y,, and not of a,, 6,, c,. We must
therefore treat a*, 6,, c, as constants, and the equation becomes

In order that this equation may be independent of the values of a,, 6,, c,,
each of these coefficients must = 0 ; and therefore we have the following
expressions for the electro-motive forces due to the action of magnets and
currents at a distance in terms of the electro-tonic functions,

1 rfo, 1 dfr 1 dy0

"tor'dt' "tor dt' ~ 47T dt'

* Translated in Taylor's New Scientific Memoirs, Part n.

ON FARAD AYS LINES OF FORCE. 205

It appears from experiment that the expression - refers to the change

of electro-tonic state of a given particle of the conductor, whether due to
change in the electro-tonic functions themselves or to the motion of the particle.

If a, be expressed as a function of x, y, z and t, and if x, y, z be the
co-ordinates of a moving particle, then the electro-motive force measured in the
direction of a; is

1 /daa dx dc^ dy da^ dz .,
°~- ~ ~ 4rr \dx ~dt dy ~di ~dz dt ~

The expressions for the electro-motive forces in y and z are similar. The
distribution of currents due to these forces depends on the form and arrange-
ment of the conducting media and on the resultant electric tension at any
point.

The discussion of these functions would involve us in mathematical formulae,
of which this paper is already too full. It is only on account of their physical
importance as the mathematical expression of one of Faraday's conjectures that I
have been induced to exhibit them at all in their present form. By a more
patient consideration of their relations, and with the help of those who are
engaged in physical inquiries both in this subject and in others not obviously
connected with it, I hope to exhibit the theory of the electro-tonic state in a
form in which all its relations may be distinctly conceived without reference to
analytical calculationa

Summary of the Theory of tlie Electro-tonic State.

We may conceive of the electro-tonic state at any point of space as a
quantity determinate in magnitude and direction, and we may represent the
electro-tonic condition of a portion of space by any mechanical system which
has at every point some quantity, which may be a velocity, a displacement, or
a force, whose direction and magnitude correspond to those of the supposed
electro-tonic state. This representation involves no physical theory, it is only
a kind of artificial notation. In analytical investigations we make use of the
three components of the electro-tonic state, and call them electro-tonic functions.
We take the resolved part of the electro-tonic intensity at every point of a

206 ON FARADAY'S LINES OF FORCE.

closed curve, and find by integration what we may call the entire electro-tin, n
intensity round the curve.

PROP. I. If on any surface a closed curve be drawn, and if the surface
within it be divided into small areas, then the entire intensity round the closed
cm-re is equal to tlie sum of the intensities round each of the small areas, all
••xtnnated in the same direction.

For, in going round the small areas, every boundary line between two of
them is passed along twice in opposite directions, and the intensity gained in
the one case is lost in the other. Every effect of passing along the interior
divisions is therefore neutralized, and the whole effect is that due to the
exterior closed curve.

LAW I. The entire electro-tonic intensity round the boundary of an element of
surface measures the quantity of magnetic induction which passes through that
surface, or, in other words, the number of lines of magnetic force which pass
through that surface.

By PROP. I. it appears that what is true of elementary surfaces is true also
of surfaces of finite magnitude, and therefore any two surfaces which are
bounded by the same closed curve will have the same quantity of magnetic
induction through them.

LAW II. The magnetic intensity at any point is connected with the quantity
of magnetic induction by a set of linear equations, called the equations of con-
duction*.

LAW III. The entire magnetic intensity round the boundary of any surface
measures the quantity of electric current which passes through that surface.

LAW IV. The quantity and intensity of electric currents are connected by a
."intern of equations of conduction.

By these four laws the magnetic and electric quantity and intensity may be
deduced from the values of the electro-tonic functions. I have not discussed
the values of the units, as that will be better done with reference to actual
experiments. We come next to the attraction of conductors of currents, and to
ithe induction of currents within conductors.

• See Art (28).

ox FARADAY'S LINES OF FORCE. 207

LAW V. TJie total electro-magnetic potential of a closed current is measured
by the product of the quantity of the current multiplied by the entire electro-tonic
intensity estimated in the same direction round the circuit.

Any displacement of the conductors which would cause an increase in the
potential will be assisted by a force measured by the rate of increase of the
potential, so that the mechanical work done during the displacement will be
measured by the increase of potential.

Although in certain cases a displacement in direction or alteration of inten-
sity of the current might increase the potential, such an alteration would not
itself produce work, and there will be no tendency towards this displacement,
for alterations in the current are due to electro-motive force, not to electro-
magnetic attractions, which can only act on the conductor.

LAW VI. The electro-motive force on any element of a conductor is measured
by the instantaneous rate of change of the electro-tonic intensity on that element,
whether in magnitude or direction.

The electro-motive force in a closed conductor is measured by the rate of
change of the entire electro-tonic intensity round the circuit referred to unit
of time. It is independent of the nature of the conductor, though the current
produced varies inversely as the resistance ; and it is the same in whatever
way the change of electro-tonic intensity has been produced, whether by motion
of the conductor or by alterations in the external circumstances.

Tn these six laws I have endeavoured to express the idea which I believe to
be the mathematical foundation of the modes of thought indicated in the Ex-
perimental Researches. I do not think that it contains even the shadow of a
true physical theory ; in fact, its chief merit as a temporary instrument of
research is that it does not, even in appearance, account for anything.

There exists however a professedly physical theory of electro-dynamics, which
is so elegant, so mathematical, and so entirely different from anything in this
paper, that I must state its axioms, at the risk of repeating what ought to
be well known. It is contained in M. W. Weber's Electro-dynamic Measure-
ments, and may be found in the Transactions of the Leibnitz Society, and of the
Royal Society of Sciences of Saxony*. The assumptions are,

* When this was written, I was not aware that part of M. Weber's Memoir is translated in
Taylor's Scientific Memoirs, VoL v. Art. xiv. The value of his researches, both experimental and
theoretical, renders the study of his theory necessary to every electrician.

208 ON FARADAY'S LINES OF FORCE.

(1) That two particles of electricity when in motion do not repel each other
with the same force as when at rest, but that the force is altered by a quantity
depending on the relative motion of the two particles, so that the expression for
the repulsion at distance r is

(2) That when electricity is moving in a conductor, the velocity of the
positive fluid relatively to the matter of the conductor is equal and opposite to
that of the negative fluid.

(3) The total action of one conducting element on another is the resultant
of the mutual actions of the masses of electricity of both kinds which are
in each.

(4) The electro-motive force at any point is the difference of the forces
acting on the positive and negative fluids.

From these axioms are deducible Ampere's laws of the attraction of
conductors, and those of Neumann and others, for the induction of currents.
Here then is a really physical theory, satisfying the required conditions better
perhaps than any yet invented, and put forth by a philosopher whose experi-
mental researches form an ample foundation for his mathematical investigations.
What is the use then of imagining an electro-tonic state of which we have
no distinctly physical conception, instead of a formula of attraction which we
can readily understand 1 I would answer, that it is a good thing to have
two ways of looking at a subject, and to admit that there are two ways of
looking at it. Besides, I do not think that we have any right at present to
understand the action of electricity, and I hold that the chief merit of a
temporary theory is, that it shall guide experiment, without impeding the
progress of the true theory when it appears. There are also objections to
making any ultimate forces in nature depend on the velocity of the bodies
between which they act. If the forces in nature are to be reduced to forces
acting between particles, the principle of the Conservation of Force requires
that these forces should be in the line joining the particles and functions of
the distance only. The experiments of M. Weber on the reverse polarity of
diamagnetics, which have been recently repeated by Professor Tyndall, establish
a fact which is equally a consequence of M. Weber's theory of electricity and
of the theory of lines of force.

ON FARADAY'S LINES OF FORCE. 209

With respect to the history of the present theory, I may state that the
recognition of certain mathematical functions as expressing the "electro-tonic
state " of Faraday, and the use of them in determining electro-dynamic
potentials and electro-motive forces is, as far as I am aware, original ; but the
distinct conception of the possibility of the mathematical expressions arose in
my mind from the perusal of Prof. W. Thomson's papers " On a Mechanical
Representation of Electric, Magnetic and Galvanic Forces," Cambridge and
Dublin Mathematical Journal, January, 1847, and his "Mathematical Theory of
Magnetism," Philosophical Transactions, Part I. 1851, Art. 78, &c. As an
instance of the help which may be derived from other physical investigations,
I may state that after I had investigated the Theorems of this paper
Professor Stokes pointed out to me the use which he had made of similar
expressions in his "Dynamical Theory of Diffraction," Section 1, Cambridge
Transactions, Vol. ix. Part 1. Whether the theory of these functions, consi-
dered with reference to electricity, may lead to new mathematical ideas to be
employed in physical research, remains to be seen. I propose in the rest of
this paper to discuss a few electrical and magnetic problems with reference to
spheres. These are intended merely as concrete examples of the methods of
which the theory has been given ; I reserve the detailed investigation of cases
chosen with special reference to experiment till I have the means of testing
their results.

EXAMPLES.
I. Theory of Electrical Images.

The method of Electrical Images, due to Prof. W. Thomson*, by which
the theory of spherical conductors has been reduced to great geometrical sim-
plicity, becomes even more simple when we see its connexion with the methods
of this paper. We have seen that the pressure at any point in a uniform
medium, due to a spherical shell (radius = a) giving out fluid at the rate of

a*
4*JV units in unit of tune, is kP outside the shell, and kPa inside it,

where r is the distance of the point from the centre of the shell.

* See a series of papers "On the Mathematical Theory of Electricity," in the Cambridge and
Dublin Math. Jour., beginning March, 1848.

VOL I. 27

ON FARADAYS LINES OF FORCE.

If there be two shells, one giving out fluid at a rate 4irPa', and the
other absorbing at the rate of 4irP'a't, then the expression for the pressure will
be, outside the shells,

D a* ,y a?

p = 4nP 47r/x — > .

r r

where /• and r are the distances from the centres of the two shells. Equating
this expression to zero we have, as the surface of no pressure, that for which

r' _ PV
r " TV'

Now the surface, for which the distances to two fixed points have a given
i-atio, is a sphere of which the centre O is in the line joining the centres of
the shells CC* produced, so that

and its radius

Pa*

'

If at the centre of this sphere we place another source of the fluid, then
the pressure due to this source must be added to that due to the other two ;
and since this additional pressure depends only on the distance from the centre,
it will be constant at the surface of the sphere, where the pressure due to
the two other sources is zero.

We have now the means of arranging a system of sources within a given
sphere, so that when combined with a given system of sources outside the
.sphere, they shall produce a given constant pressure at the surface of the sphere.

Let a be the radius of the sphere, and p the given pressure, and let the
given sources be at distances 6,, 63, &c. from the centre, and let their rates of
production be 47r/J,, 47rPs, &c.

a* a*
Then if at distances .- , , , &c. (measured in the same direction as 6,, bt, &c.

from the centre) we place negative sources whose rates are

, - , <fec.,

ON FARADAY'S LINES OF FORCE. 211

the pressure at the surface r = a will be reduced to zero. Now placing a source

T?Ct

47T^r- at the centre, the pressure at the surface will be uniform and equal to p.

The whole amount of fluid emitted by the surface r = a may be fo^nd by
adding the rates of production of the sources within it. The result is

fp Pt P, , }
\ f - -j- • - ~ --&c. [ .
(k bt b,

To apply this result to the case of a conducting sphere, let us suppose
the external sources 4irP,, 4irP, to be small electrified bodies, containing eu e,
of positive electricity. Let us also suppose that the whole charge of the con-
ducting sphere is —E previous to the action of the external points. Then all
that is required for the complete solution of the problem is, that the surface
of the sphere shall be a surface of equal potential, and that the total charge
of the surface shall be E.

If by any distribution of imaginary sources within the spherical surface we
can effect this, the value of the corresponding potential outside the sphere is
the true and only one. The potential inside the sphere must really be constant
and equal to that at the surface.

We must therefore find the images of the external electrified points, that
is, for every point at distance 6 from the centre we must find a point on the

same radius at a distance j- , and at that point we must place a quantity
= — e , - of imaginary electricity.

At the centre we must put a quantity E' such that

E' = E + e1 ^ + e^ + &c. ;

then if R be the distance from the centre, rt, rv &c. the distances from the
electrified points, and r\, r\, &c. the distances from their images at any point
outside the sphere, the potential at that point will be

& , i1 a 1\ il a !\ ,
P = TT + e, --- r —}+eA -- r — +&c.

R V\r, 6, rj °\r, bt rj

27—2

212 ON FARADAY'S LINES OF FORCE.

This is the value of the potential outside the sphere. At the surface we

have

n J &t a 6. O

R = a and — = -r, - = — , &c. ,

'

so that at the. surface

^ a 6, 6,
and this must also be the value of p for any point within the sphere.

For the application of the principle of electrical images the reader is referred
to Prof. Thomson's papers in the Cambridge and Dublin Mathematical Journal.

The only case which we shall consider is that in which j^ = I, and 6j is infi-
nitely distant along the axis of x, and E = 0.

The value p outside the sphere becomes then

and inside p = 0.

II. On the effect of a paramagnetic or diamagnetic sphei'e in a uniform, field of

magnetic force*.

The expression for the potential of a small magnet placed at the origin of
co-ordinates in the direction of the axis of x is

, d /m
' dx

(m\ , x

= — Im -j .
\r) r1

The effect of the sphere in disturbing the lines of force may be supposed
as a first hypothesis to be similar to that of a small magnet at the origin,
whose strength is to be determined. (We shall find this to be accurately true.)

* See Prof. Thomson, on the Theory of Magnetic Induction, Phil. Mag. March, 1851. The induc-
tive capacity of the sphere, according to that paper, is the ratio of the quantity of magnetic induction

1 V Sk'

(not the inlenrity) within the sphere to that without It is therefore equal to jB-^= -^7 — r/ accord-
ing to our notation.

ON FARADAY'S LINES OF FORCE. 213

Let the value of the potential undisturbed by the presence of the sphere be

p = Ix.
Let the sphere produce an additional potential, which for external points is

and let the potential within the sphere be

p1 = Ex.

Let k' be the coefficient of resistance outside, and k inside the sphere, then
the conditions to be fulfilled are, that the interior and exterior potentials should
coincide at the surface, and that the induction through the surface should be the
same whether deduced from the external or the internal potential. Putting
x = r cos 6, we have for the external potential ,

and for the internal

p1 = Br cos 6,

and these must be identical when r = a, or

= B.

The induction through the surface in the external medium is

and that through the interior surface is

and .-. (/-24) =

k — ¥ "\Tf

A - T P — r

• ~j

These equations give

The effect outside the sphere is equal to that of a little magnet whose
length is I and moment ml, provided

214

ON FARADAY 8 LINES OF FORCE.

Suppose this uniform field to be that due to terrestrial magnetism, then,
if k is less than k' as in paramagnetic bodies, the marked end of the equi-
valent magnet will be turned to the north. If k is greater than k' as in
diamagnetic bodies, the unmarked end of the equivalent magnet would be turned
to the north.

III. Magnetic Jield of variable Intensity.

Now suppose the intensity in the undisturbed magnetic field to vary in
magnitude and direction from one point to another, and that its components
in x, y, z are represented by a, /3, y, then, if as a first approximation we re-
gard the intensity within the sphere as sensibly equal to that at the centre,
the change of potential outside the sphere arising from the presence of the
sphere, disturbing the lines of force, will be the same as that due to three
small magnets at the centre, with their axes parallel to x, y, and z, and their
moments equal to

7, _ /•' l- _ z-' /- _ /-'

A/ A/ . 1/1 A/ Ay

2k + k'aa' ZT+Z"* 2k + k'

The actual distribution of potential within and without the sphere may be
conceived as the result of a distribution of imaginary magnetic matter on the
surface of the sphere ; but since the external effect of this superficial magnetism
is exactly the same as that of the three small magnets at the centre, the
mechanical effect of external attractions will be the same as if the three magnets
really existed.

Now let three small magnets whose lengths are llt I,, /,, and strengths
TO,, 7H,, TO,, exist at the point x, y, z with their axes parallel to the axes of
x, y, z ; then resolving the forces on the three magnets in the direction of X, we
have

f/a. /, "

da lt'

da lt

f dx2

dy 2

a + J~ o

dz 2

da Z,

• + TO,-

da /j

•+m,-

da lt

'a + dx 2.

— B T ~T~

L ay 2.

~a + dz 2.

da

(fa

da

ON FARAD AYS LINES OF FORCE. 215

Substituting the values of the moments of the imaginary magnets

v k — k' , I da 0dB dy\ k^k' a3 d ,
— A = „, . 7, a ( a — + /3 — + y-r-] = ^, . ,, — -5- (<

— ' ' .''

The force impelling the sphere in the direction of x is therefore dependent
on the variation of the square of the intensity or (a2 + /S2 + y2), as we move along
the direction of x, and the same is true for y and z, so that the law is, that
the force acting on diamagnetic spheres is from places of greater to places of
less intensity of magnetic force, and that in similar distributions of magnetic
force it varies as the mass of the sphere and the square of the intensity.

It is easy by means of Laplace's Coefficients to extend the approximation
to the value of the potential as far as we please, and to calculate the attew>
tion. For instance, if a north or south magnetic pole whose strength is M, be
placed at a distance b from a diamagnetic sphere, radius a, the repulsion will be

a3 / 2 . 1 3.2 a2 4.3 a4

,,, ,, 7/N
= M> (k-k )

When Y is small, the first term gives a sufficient approximation. The repul-
sion is then as the square of the strength of the pole, and the mass of the
sphere directly and the fifth power of the distance inversely, considering the
pole as a point.

IV. Two Spheres in uniform Jield.

Let two spheres of radius a be connected together so that their centres are
kept at a distance 6, and let them be suspended in a uniform magnetic field,
then, although each sphere by itself would have been in equilibrium at any part
of the field, the disturbance of the field will produce forces tending to make the
balls set in a particular direction.

Let the centre of one of the spheres be taken as origin, then the undis-
turbed potential is

p = Ir cos 6,

and the potential due to the sphere is

T k-k' a'

216 ON FARADAY'S LINES OF FORCE.

The whole potential is therefore equal to

k-k'

' ..«

This is the value of the square of the intensity at any point. The moment
of the couple tending to turn the combination of balls in the direction of the
original force

, d ^/k-k' A ,

L=*del (zkTk'") when r=b>

k-k'

k-k'

This expression, which must be positive, since 6 is greater than a, gives the
moment of a force tending to turn the line joining the centres of the spheres
towards the original lines of force.

Whether the spheres are magnetic or diamagnetic they tend to set in the
axial direction, and that without distinction of north and south. If, however,
one sphere be magnetic and the other diamagnetic, the line of centres will set
equatoreally. The magnitude of the force depends on the square of (k — k'), and
is therefore quite insensible except in iron*.

V. Two Spheres between the poles of a Magnet.

Let us next take the case of the same balls placed not in a uniform field
but between a north and a south pole, ±M, distant 2c from each other in the
direction of x.

* See Prof. Thomson in Phil Mag. March, 1851.

ON FARAD AYS LINES OF FORCE. 217

The expression for the potential, the middle of the line joining the poles
being the origin, is

We* + 1* - 2cr cos 0 Vc2 + r1 + 2cr cos W
From this we find as the value of P,

and the moment to turn a pair of spheres (radius a, distance 26) in the
direction in which 6 is increased is

Ic-lc J/Vfc3 .

^~ O O _ » i / ;: Sill Zi \J •

This force, which tends to turn the line of centres equatoreally for diamagnetic
and axially for magnetic spheres, varies directly as the square of the strength of
the magnet, the cube of the radius of the spheres and the square of the dis-
tance of their centres, and inversely as the sixth power of the distance of the
poles of the magnet, considered as points. As long as these poles are near each
other this action of the poles will be much stronger than the mutual action of
the spheres, so that as a general rule we may say that elongated bodies set
axially or equatoreally between the poles of a magnet according as they are mag-
netic or diamagnetic. If, instead of being placed between two poles very near
to each other, they had been placed in a uniform field such as that of terrestrial
magnetism or that produced by a spherical electro-magnet (see Ex. vin.), an
elongated body would set axially whether magnetic or diamagnetic.

In all these cases the phenomena depend on k — k', so that the sphere con-
ducts itself magnetically or diamagnetically according as it is more or less
magnetic, or less or more diamagnetic than the medium in which it is placed.

VI. On the Magnetic Phenomena of a Sphere cut from a substance whose
coefficient of resistance is different in different directions.

Let the axes of magnetic resistance be parallel throughout the sphere, and

let them be taken for the axes of x, y, z. Let klt kt, k,, be the coefficients of

resistance in these three directions, and let k' be that of the external medium,

VOL. I. 28

218 ON FARADAY'S LINES OF FORCE.

and a the radius of the sphere. Let / be the undisturbed magnetic intensity
of the field into which the sphere is introduced, and let its direction-cosines
be /, m, ».

Let us now take the case of a homogeneous sphere whose coefficient is £,
placed in a uniform magnetic field whose intensity is II in the direction of x.
The resultant potential outside the sphere would be

and for internal points

So that in the interior of the sphere the magnetization is entirely in the direc-
tion of x. It is therefore quite independent of the coefficients of resistance in
the directions of x and y, which may be changed from &, into kt and k, with-
out disturbing this distribution of magnetism. We may therefore treat the sphere
as homogeneous for each of the three components of /, but we must use a
different coefficient for each. We find for external points

k,-k'

and for internal points

3k,

The external effect is the same as that which would have been produced
if the small magnet whose moments are

' mlat>

had been placed at the origin with their directions coinciding with the axes of
x, y, z. The effect of the original force / in turning the sphere about the axis
of x may be found by taking the moments of the components of that force
on these equivalent magnets. The moment of the force in the direction of y
acting on the third magnet is

and that of the force in z on the second magnet is

ON FAKADAYS LINES OF FORCE. 219

The whole couple about the axis of x is therefore

OK («;3 — Kaj ~-a 3

(2ks + k')(2kt + k')U

tending to turn the sphere round from the axis of y towards that of z. Sup-
pose the sphere to be suspended so that the axis of x is vertical, and let /
be horizontal, then if 6 be the angle which the axis of y makes with the
direction of /, m = cos 6, n = — sin 0, and the expression for the moment becomes

k'(k,-k3) , . -
' 'J-

tending to increase 0. The axis of least resistance therefore sets axially, but
with either end indifferently towards the north.

Since in all bodies, except iron, the values of k are nearly the same as in
a vacuum, the coefficient of this quantity can be but little .altered by changing
the value of k' to k, the value in space. The expression then becomes

£ — T— * 7V sin 20,
independent of the external medium*.

VII. Permanent magnetism in a spherical shell.

The case of a homogeneous shell of a diamagnetic or paramagnetic substance
presents no difficulty. The intensity within the shell is less than what it would
have been if the shell were away, whether the substance of the shell be dia-
magnetic or paramagnetic. When the resistance of the shell is infinite, and when
it vanishes, the intensity within the shell is zero.

In the case of no resistance the entire effect of the shell on any point,
internal or external, may be represented by supposing a superficial stratum of

* Taking the more general case of magnetic induction referred to in Art. (28), we find, in the
expression for the moment of the magnetic forces, a constant term depending on T, besides those
terms which depend on sines and cosines of 0. The result is, that in every complete revolution in
the negative direction round the axis of T, a certain positive amount of work is gained; but, since
no inexhaustible source of work can exist in nature, we must admit that T=0 in all substances,
'with respect to magnetic induction. This argument does not hold in the case of electric conduction,
or in the case of a body through which heat or electricity is passing, for such states are main-
tained by the continual expenditure of work. See Prof. Thomson, Phil. Mag. March, 1851, p. 186.

28—2

ON FARADAY'S LINES OF FORCE.

magnetic matter spread over the outer surface, the density being given by the
equation

p = 37 cos 0.

Suppose the shell now to be converted into a permanent magnet, so that the
distribution of imaginary magnetic matter is invariable, then the external poten-
tial due to the shell will be

/= -/£cos0,

and the internal potential ^,= —Ir cos 6.

Now let us investigate the effect of filling up the shell with some substance
of which the resistance is k, the resistance in the external medium being k'.
The thickness of the magnetized shell may be neglected. Let the magnetic
moment of the permanent magnetism be /a', and that of the imaginary super-

ficial distribution due to the medium k = Aa*. Then the potentials are

t
external p' = (1+ A) -, cos 0, internal pt = (1+ A] r cos 0.

The distribution of real magnetism is the same before and after the introduc-
tion of the medium k, so that

< A* ~~" A/ f

or A = , ,, 1.

2k + k

The external effect of the magnetized shell is increased or diminished according
as k is greater or less than k'. It is therefore increased by filling up the shell
with diamagnetic matter, and diminished by filling it with paramagnetic matter,
such as iron.

VIII. Electro-magnetic spheincal shell,

Let us take as an example of the magnetic effects of electric currents,
an electro-magnet in the form of a thin spherical shell. Let its radius be «,
and its thickness t, and let its external effect be that of a magnet whose
moment is /a*. Both within and without the shell the magnetic effect may be
represented by a potential, but within the substance of the shell, where there

ON FARADAY'S LINES OP FORCE. 221

are electric currents, the magnetic effects cannot be represented by a potential.
Let p', pl be the external and internal potentials,

p' = / -5 cos 0, pl = Ar cos #,
and since there is no permanent magnetism, -4- = -~ , when r = a,

If we draw any closed curve cutting the shell at the equator, and at some
other point for which 0 is known, then the total magnetic intensity round this
curve will be 3/a cos 0, and as this is a measure of the total electric current which
flows through it, the quantity of the current at any point may be found by
differentiation. The quantity which flows through the element tdd is — 3/a sin 6dd,
so that the quantity of the current referred to unit of area of section is

- 31 ? sin 0.

t

If the shell be composed of a wire coiled round the sphere so that the number
of coils to the inch varies as the sine of 6, then the external effect will be
nearly the same as if the shell had been made of a uniform conducting sub-
stance, and the currents had been distributed according to the law we have just
given.

If a wire conducting a current of strength /, be wound round a sphere
of radius a so that the distance between successive coils measured along the

2(t
axis of x is - - , then there will be n coils altogether, and the value of II for

7*

the resulting electro-magnet will be

I--I
/l 0a *

The potentials, external and internal, will be

, T n a* nTn r a

P =1, Q ^ cos0, p,= -2/3 g -cos 0.

The interior of the shell is therefore a uniform magnetic field.

ON FARADAY'S LINES OF FORCE.

IX. Effect of the core of the electro-magnet.

Now let us suppose a sphere of diamagnetic or paramagnetic matter intro-
duced into the electro-magnetic coil. The result may be obtained as in the
last case, and the potentials become

n 3k r

The external effect is greater or less than before, according as k' is greater
or less than k, that is, according as the interior of the sphere is magnetic or
diamagnetic with respect to the external medium, and the internal effect is
altered in the opposite direction, being greatest for a diamagnetic medium.

This investigation explains the effect of introducing an iron core into an
electro-magnet. If the value of k for the core were to vanish altogether, the
effect of the electro-magnet would be three times that which it has without
the core. As k has always a finite value, the effect of the core is less than this.

In the interior of the electro-magnet we have a uniform field of magnetic
force, the intensity of which may be increased by surrounding the coil with a
shell of iron. If k' = 0, and the shell infinitely thick, the effect on internal points
would be tripled.

The effect of the core is greater in the case of a cylindric magnet, and
greatest of all when the core is a ring of soft iron.

X. Electro-tonic Junctions in spherical electro-magnet.

Let us now find the electro-tonic functions due to this electro-magnet.
They will be of the form

«»=o, A>=<uz, %= -<*y>

where o> is some function of r. Where there are no electric currents, we must
have a,, 6,, ct each = 0, and this implies

d
Tr

the solution of which is

ON FARADAY'S LINES OF FORCE. 223

Within the shell <u cannot become infinite ; therefore a> = C1 is the solution,
and outside a must vanish at an infinite distance, so that

..^ " ""V" ''

is the solution outside. The magnetic quantity within the shell is found by last
article to be

o T n 3 • _ ^& dy,

* 6a 2k+k' dz dy

therefore within the sphere

7.n 1

Q) = — - — — - — — .

Outside the sphere we must determine <a so as to coincide at the surface
with the internal value. The external value is therefore

= _
~

__
3k+k' r

where the shell containing the currents is made up of n coils of wire, con-
ducting a current of total quantity 7r

Let another wire be coiled round the shell according to the same law, and
let the total number of coils be n' ; then the total electro-tonic intensity
round the second coil is found by integrating

EIt = I <ua sin 6ds,
Jo

along the whole length of the wire. The equation of the wire is

ntr
where n' is a large number; and therefore

ds = a sin 0d<f>,

r,T 4ir , , 2ir ,T
.'. EIt= — wa'n = — -annl

E may be called the electro-tonic coefficient for the particular wire.

224 ON FARADAY'S LINKS OF FORCE.

XI. Spherical electro-magnetic Coil-Machine.

We have now obtained the electro-tonic function which defines the action
of the one coil on the other. The action of each coil on itself is found by
putting n1 or n* for nn'. Let the first coil be connected with an apparatus
producing a variable electro-motive force F. Let us find the effects on both
wires, supposing their total resistances to be R and R, and the quantity of
the currents / and /'.

. < _ ,

Let N etand for j^r — r>» • then the electro-motive force of the first

3

wire on the second is

-Nnn'^.
at

That of the second on itself is

The equation of the current in the second wire is therefore

-Nnn'~-Nn^=Rr ......................... (1).

at at

The equation of the current in the first wire is

(2).

Eliminating the differential coefficients, we get

R j R r _F

J. ™~ /" -i ^~ •

n n n

,, n" dl F ,, n* dF

and

from which to find I and F. For this purpose we require to know the value
of F in terms of t.

Let us first take the case in which F is constant and / and f initially = 0.
This is the case of an electro-magnetic coil-machine at the moment when the
connexion is made with the galvanic trough.

ON FARADAY'S LINES OF FORCE. 225

Putting £T for JV (# + 1^) we &&&

y a
I=7t(l~* T)'

/ Of

r— F n ~»
#/i€

The primary current increases very rapidly from 0 to -^ , and the secondary
commences at — -f» - • an(^ speedily vanishes, owing to the value of r being

.ft 71

generally very smalL

The whole work done by either current in heating the wire or in any other
kind of action is found from the expression

fOO

PRdt.
Jo

The total quantity of current is

" Idt.

1 o
For the secondary current we find

i1 4' J, Rn 2*

The work done and the quantity of the current are therefore the same as
if a current of quantity F = — jrr- had passed through the wire for a time T, where

'_ v i 71

T =

This method of considering a variable current of short duration is due to
Weber, whose experimental methods render the determination of the equivalent
current a matter of great precision.

Now let the electro-motive force F suddenly cease while the current in the
primary wire is /„ and in the secondary = 0. Then we shall have for the subse-
quent time

a r /?„' a

/-TV7 /•-£• _e~T

' "~7P „ e

VOL. I. 29

226 ON FARADAY'S LINES OF FORCE.

n '

The equivalent currents are $1, and £/, -g> — , and their duration is r.

/i 71

When the communication with the source of the current is cut off, there
will be a change of R, This will produce a change in the value of r, so that
if R be suddenly increased, the strength of the secondary current will be increased,
and its duration diminished. This is the case in the ordinary coil-machines. The
quantity N depends on the form of the machine, and may be determined by
experiment for a machine of any shape.

XII. Spherical shell revolving in magnetic field.

Let us next take the case of a revolving shell of conducting matter under
the influence of a uniform field of magnetic force. The phenomena are explained
by Faraday in his Experimental Researches, Series n., and references are there
given to previous experiments.

Let the axis of z be the axis of revolution, and let the angular velocity
be a*. Let the magnetism of the field be represented in quantity by /, inclined
at an angle 0 to the direction of z, in the plane of zx.

Let R be the radius of the spherical shell, and T the thickness. Let the
quantities a,, /Je, y«,.be the electro- tonic functions at any point of space; a,, 6,, c,,
a,, ft, y, symbols of magnetic quantity and intensity; a,, b,, c,, a,, &, y, of
electric quantity and intensity. Let pt be the electric tension at any point,

<2);

ON FARADAY'S LINES OF FORCE.

227

The expressions for a,,, $,, % due to the magnetism of the field are

a,=A<> + -ycoB0>

P0 = B0 + -(zBin.d — x cos 6],

7o= <?<,-- y sin 0,

A0, B0, C0 being constants ; and the velocities of the particles of the revolving
sphere are

dx dy dz

—j~ = — WW, — Y~ = OKT, -j- = 0.

dt dt dt

We have therefore for the electro-motive forces

1 da, II
°*=-^-di= -&2C°80(aX>

B= ._1.^= ^1

I dy. 1 / .

y2 = — -j -- f- = — - sin
47r rft 4ir 2

Returning to equations (1), we get

(^L _ ^£1} - *?A _ ^y.= 0

dc, dat\ dy, da, I I
-~----l = -p--^ = - -
dx dz 4?r2

db,\ da, d&

J - I = -= --- } -

rfx/ d dx

From which with equation (2) we find

1 1 /

1 1 /

p,= — r^- Io {(x1 4- y1) coB0 — xz sin 6}.

29—2

228 ON FARADAY'S LINES OF FORCE.

These expressions would determine completely the motion of electricity in
a revolving sphere if we neglect the action of these currents on themselves.
They express a system of circular currents about the axis of y, the quantity
of current at any point being proportional to the distance from that axis.
The external magnetic effect will be that of a small magnet whose moment

TIP
is TQ-]. «/sin 6, with its direction along the axis of y, so that the magnetism of

the field would tend to turn it back to the axis of z*.

The existence of these currents will of course alter the distribution of
the electro-tonic functions, and so they will react on themselves. Let the
final result of this action be a system of currents about an axis in the plane
of xy inclined to the axis of x at an angle <f> and producing an external effect
equal to that of a magnet whose moment is TIP.

The magnetic inductive components within the shell are

7, sin 6 — 2f cos $ in x,
— 21' sin <f> in y,
/! cos 0 in z.

Each of these would produce its own system of currents when the sphere
is in motion, and these would give rise to new distributions of magnetism,
which, when the velocity is uniform, must be the same as the original distri-
bution,

T

(I, sin 6 — 2/' cos <f>) in x produces 2 j^— r a (7, sin 6 — 2/' cos <f>) in y,

T
(-2F sin <f>) in y produces 2 J-T a> (2/' sin <f>) in x ;

7, eos 6 in z produces no currents.

We must therefore have the following equations, since the state of the shell
is the same at every instant,

T

T

- 2f sin $ = —- T <a (/, sin 6- 2f cos <£),

* The expression for />, indicates a variable electric tension in the shell, so that currents might
be collected by wires touching it at the equator and poles.

ON FARADAY'S LINES OF FORCE. 229

J7

whence cot <£ = - ^ *>, I = £ *** /.sing.

To understand the meaning of these expressions let us take a particular case.

Let the axis of the revolving shell be vertical, and let the revolution be
from north to west. Let / be the total intensity of the terrestrial magnetism,
and let the dip be 0, then /cos 6 is the horizontal component in the direction
of magnetic north.

The result of the rotation is to produce currents in the shell about an

T

axis inclined at a small angle = tan"1 —-y w to the south of magnetic west, and

the external effect of these currents is the same as that of a magnet whose
moment is

% -j __ Ta If I cos 0.

V2Wf-j.1V

The moment of the couple due to terrestrial magnetism tending to stop the
rotation is

j»n ,a

JtT cos" 0,

and the loss of work due to this in unit of time is

247T& T«t

cos' 0.

This loss of work is made up by an evolution of heat in the substance of
the shell, as is proved by a recent experiment of M. Foucault (see Comptes
Rendus, XLI. p. 450).

[From the Transactions of the Royal Scottish Society of Arts, VoL IV. Part IV.]

IX. Description of a New Form of the Platometer, an Instrument for
measuring the Areas of Plane Figures drawn on Paper*.

1. THE measurement of the area of a plane figure on a map or plan is an
operation so frequently occurring in practice, that any method by which it may
be easily and quickly performed is deserving of attention. A very able expo-
sition of the principle of such instruments will be found in the article on
Planimeters in the Reports of the Juries of the Great Exhibition, 1851.

2. In considering the principle of instruments of this kind, it will be most
convenient to suppose the area of the figure measured by an imaginary straight
line, which, by moving parallel to itself, and at the same

time altering in length to suit the form of the area,
accurately sweeps it out.

Let AZ be a fixed vertical line, APQZ the boundary
of the area, and let a variable horizontal line move
parallel to itself from A to Z, so as to have its extremi-
ties, P, M, in the curve and in the fixed straight line.
Now, suppose the horizontal line (which we shall call the
generating line) to move from the position PM to QNt
MN being some small quantity, say one inch for distinct-
During this movement, the generating line will

ness.

have swept out the narrow strip of the surface, PMNQ,
which exceeds the portion PMNp by the small triangle PQp.

But since MN, the breadth of the strip, is one inch, the strip will contain
as many square inches as PM is inches long; so that, when the generating

* Read to the Society, 22nd Jan. 1855.

ON A NEW FORM OF THE PLATOMETER.

231

line descends one inch, it sweeps out a number of square inches equal to the
number of linear inches in its length.

Therefore, if we have a machine with an index of any kind, which, while
the generating line moves one inch downwards, moves forward as many degrees
as the generating line is inches long, and if the generating line be alternately
moved an inch and altered in length, the index will mark
the number of square inches swept over during the whole
operation. By the ordinary method of limits, it may be
shown that, if these changes be made continuous instead
of sudden, the index will still measure the area of the
curve traced by the extremity of the generating line.

3. When the area is bounded by a closed curve, as
ABDC, then to determine the area we must carry the tra-
cing point from some point A of the curve, completely round
the circumference to A again. Then, while the tracing point
moves from A to C, the index will go forward and mea-
sure the number of square inches in ACRP, and, while it
moves from C to D, the index will measure backwards the
square inches in CRPD, so that it will now indicate the

square inches in ACD. Similarly, during the other part of the motion from
D to B, and from B to D, the part DBA will be measured ; so that when
the tracing point returns to D, the instrument will have measured the area
ACDB. It is evident that the whole area will appear positive or negative
according as the tracing point is carried round in the direction ACDB or ABDC.

4. We have next to consider the various methods of communicating the
required motion to the index. The first is by means of two discs, the first
having a flat horizontal rough surface, turning on a vertical

axis, OQ, and the second vertical, with its circumference rest-
ing on the flat surface of the first at P, so as to be driven
round by the motion of the first disc. The velocity of the
second disc will depend on OP, the distance of the point of
contact from the centre of the first disc; so that if OP be
made always equal to the generating line, the conditions of the instrument will
be fulfilled.

This is accomplished by causing the index-disc to slip along the radius of

232 ON A NEW FORM OF THE PLATOMETER.

the horizontal disc ; so that in working the instrument, the motion of the index-
disc is compounded of a rolling motion due to the rotation of the first disc,
and a slipping motion due to the variation of the generating line.

5. In the instrument presented by Mr Sang to the Society, the first disc is
replaced by a cone, and the action of the instrument corresponds to a mathe-
matical valuation of the area by the use of oblique co-ordinates. As he has
himself explained it very completely, it will be enough here to say, that the
index-wheel has still a motion of slipping as well as of rolling.

6. Now, suppose a wheel rolling on a surface, and pressing on it with a
weight of a pound; then suppose the coefficient of friction to be |, it will
require a force of 2 oz. at least to produce slipping at all, so that even if the
resistance of the axis, &c., amounted to 1 oz., the rolling would be perfect. But
if the wheel were forcibly pulled sideways, so as to slide along in the direction
of the axis, then, if the friction of the axis, &c., opposed no resistance to the
turning of the wheel, the rotation would still be that due to the forward motion ;
but if there were any resistance, however small, it would produce its effect in
diminishing the amount of rotation.

The case is that of a mass resting on a rough surface, which requires a
great force to produce the slightest motion ; but when some other force acts
on it and keeps it in motion, the very smallest force is sufficient to alter that
motion in direction.

7. This effect of the combination of slipping and rolling has not escaped
the observation of Mr Sang, who has both measured its amount, and shown how
to eliminate its effect. In the improved instrument as constructed by him, I
believe that the greatest error introduced in this way does not equal the ordi-
nary errors of measurement by the old process of triangulation. This accuracy,
however, is a proof of the excellence of the workmanship, and the smoothness
of the action of the instrument ; for if any considerable resistance had to be
overcome, it would display itself in the results.

8. Having seen and admired these instruments at the Great Exhibition in
1851, and being convinced that the combination of slipping and rolling was a
drawback on the perfection of the instrument, I began to search for some
arrangement by which the motion should be that of perfect rolling in every

ON A XEW FOEM OF THE PLATOMETER. 233

motion of which the instrument is capable. The forms of the rolling parts which
I considered were —

1. Two equal spheres.

2. Two spheres, the diameters being as 1 to 2.

3. A cone and cylinder, axes at right angles.

Of these, the first combination only suited my purpose. I devised several modes
of mounting the spheres so as to make the principle available. That which I
adopted is borrowed, as to many details, from the instruments already con-
structed, so that the originality of the device may be reduced to this principle —
The abolition of slipping by the use of two equal spheres.

9. The instrument (Fig. 1) is mounted on a frame, which rolls on the two
connected wheels, MM, and is thus constrained to travel up and down the
paper, moving parallel to itself.

CH is a horizontal axis, passing through two supports attached to the
frame, and carrying the wheel K and the hemisphere LAP. The wheel K rolls
on the plane on which the instrument travels, and communicates its motion to
the hemisphere, which therefore revolves about the axis AH with a velocity
proportional to that with which the instrument moves backwards or forwards.

FCO is a framework (better seen in the other figures) capable of revolving
about a vertical axis, Cc, being joined at C and c to the frame of the instru-
ment. The parts CF and CO are at right angles to each other and horizontal.
The part CO carries with it a ring, SOS, which turns about a vertical axis Oo.
This ring supports the index -sphere Bb by the extremities of its axis Ss, just
as the meridian circle carries a terrestrial globe. By this arrangement, it will
be seen that the axis of the sphere is kept always horizontal, while its centre
moves so as to be always at a constant distance from that of the hemisphere.
This distance must be adjusted so that the spheres may always remain in con-
tact, and the pressure at the point of contact may be regulated by means of
springs or compresses at O and o acting in the direction OC, oc. In this way
the rotation of the hemisphere is made to drive the index-sphere.

10. Now, let us consider the working of the instrument. Suppose the arm
CE placed so as to coincide with CD, then O, the centre of the index-sphere
will be in the prolongation of the axis HA. Suppose also that, when in this
position, the equator bB of the index-sphere is in contact with the pole A of
the hemisphere. Now, let the arch be turned into the position CE as in the

VOL. i. 30

_:f4 ON A NEW FORM OF THE PLATOMETER.

figure, then the rest of the framework will be turned through an equal angle,
and the index-sphere will roll on the hemisphere till it come into the position
represented in the figure. Then, if there be no slipping, the arc AP = BP, and
the angle ACP = SOP.

Next, let the instrument be moved backwards or forwards, so as to turn
the wheel Kk and the hemisphere U, then the index-sphere will be turned
about its axis Ss by the action of the hemisphere, but the ratio of their veloci-
ties will depend on their relative positions. If we draw PQ, PR, perpendiculars
from the point of contact on the two axes, then the angular motion of the
index-sphere will be to that of the hemisphere, as PQ is to PR ; that is, as
PQ is to QC, by the equal triangles POQ, PQC; that is, as ED is to DC,
by the similar triangles CQP, CDE.

Therefore the ratio of the angular velocities is as ED to DC, but since
DC is constant, this ratio varies as ED. We have now only to contrive some
way of making ED act as the generating line, and the machine is complete
(see art. 2).

11. The arm CF is moved in the following manner: — Tt is a rectangular
metal beam, fixed to the frame of the instrument, and parallel to the axis AH.
cEe is a little carriage which rolls along it, having two rollers on one side and
one on the other, which is pressed against the beam by a spring. This carriage
carries a vertical pin, E, turning in its socket, and having a collar above,
through which the arm CF works smoothly. The tracing point G is attached
to the carriage by a jointed frame eGe, which is so arranged that the point
may not bear too heavily on the paper.

12. When the machine is in action, the tracing point is placed on a point
in the boundary of the figure, and made to move round it always in one
direction till it arrives at the same point again. The up-and-down motion of
the tracing point moves the whole instrument over the paper, turns the wheel
I\, the hemisphere LI, and the index-sphere Bb ; while the lateral motion of
the tracing point moves the carriage E on the beam Tt, and so works the arm
CF and the framework CO; and so changes the relative velocities of the two
spheres, as has been explained.

13. In this way the instrument works by a perfect rolling motion, in what-
ever direction the tracing point is moved ; but since the accuracy of the result
depends on the equality of the arcs AP and BP, and since the smallest error

ON A NEW FORM OF THE PLATOMETEtt. 235

of adjustment would, in the course of time, produce a considerable deviation
from this equality, some contrivance is necessary to secure it. For this purpose
a wheel is fixed on the same axis with the ring SOs, and another of the same
size is fixed to the frame of the instrument, with its centre coinciding with the
vertical axis through C. These wheels are connected by two pieces of watch-
spring, which are arranged so as to apply closely to the edges of the wheels.
The first is firmly attached to the nearer side of the fixed wheel, and to the
farther side of the moveable wheel, and the second to the farther side of the
fixed wheel, and the nearer side of the moveable wheel, crossing beneath the
first steel band. In this way the spheres are maintained in their proper relative
position ; but since no instrument can be perfect, the wheels, by preventing
derangement, must cause some slight slipping, depending on the errors of work-
manship. This, however, does not ruin the pretensions of the instrument, for it
may be shown that the error introduced by slipping depends on the distance
through which the lateral slipping takes place ; and since in this case it must
be very small compared with its necessarily large amount in the other instru-
ments, the error introduced by it must be diminished in the same proportion.

14. I have shewn how the rotation of the index-sphere is proportional to
the area of the figure traced by the tracing point. This rotation must be
measured by means of a graduated circle attached to the sphere, and read off
by means of a vernier. The result, as measured in degrees, may be interpreted
in the following manner: —

Suppose the instrument to be placed with the arm CF coinciding with CD,
the equator Bb of the index-sphere touching the pole A of the hemisphere, and
the index of the vernier at zero : then let these four operations be performed :—

(1) Let the tracing point be moved to the right till DE = DC, and there-
fore DCE, AGP, and P0£ = 45».

(2) Let the instrument be rolled upwards till the wheel K has made a
complete revolution, carrying the hemisphere with it ; then, on account of the
equality of the angles SOP, PCA, the index-sphere will also make a complete
revolution.

(3) Let the arm CF be brought back again till F coincides with D.

(4) Let the instrument be rolled back again through a complete revolution
of the wheel K. The index-sphere will not rotate, because the point of contact
is at the pole of the hemisphere.

30—2

236 OX A NEW FORM OF THE PLATOMETER.

The tracing point has now traversed the boundary of a rectangle, whose
length is the circumference of the wheel A' and its breadth is equal to CD ;
and during this operation, the index-sphere has made a complete revolution.
360* on the sphere, therefore, correspond to an area equal to the rectangle con-
tained by the circumference of the wheel and the distance CD. The size of
the wheel K being known, different values may be given to CD, so as to make
the instrument measure according to any required scale. This may be done,
either by shifting the position of the beam 755, or by having several sockets
in the carriage E for the pin which directs the arm to work in.

15. If I have been too prolix in describing the action of an instrument
which has never been constructed, it is because I have myself derived great
satisfaction from following out the mechanical consequences of the mathematical
theorem on which the truth of this method depends. Among the other forms
of apparatus by which the action of the two spheres may be rendered available,
is one which might be found practicable in cases to which that here given
would not apply. In this instrument (Fig. 4) the areas are swept out by a
radius-vector of variable length, turning round a fixed point in the plane. The
area is thus swept out with a velocity varying as the angular velocity of the
radius-vector and the square of its length conjointly, and the construction of the
machine is adapted to the case as follows : —

The hemisphere is fixed on the top of a vertical pillar, about which the rest
of the instrument turns. The index-sphere is supported as before by a ring and
framework. This framework turns about the vertical pillar along with the tra-
cing point, but has also a motion in a vertical plane, which is communicated to
it by a curved slide connected with the tracing point, and which, by means of a
prolonged arm, moves the framework as the tracing point is moved to and from
the pillar.

The form of the curved slide is such, that the tangent of the angle of
inclination of the line joining the centres of the spheres with the vertical is
proportional to the square of the distance of the tracing point from the vertical
axis of the instrument. The curve which fulfils this condition is an hyperbola,
one of whose asymptotes is vertical, and passes through the tracing point, and
the other horizontal through the centre of the hemisphere.

The other parts of this instrument are identical with those belonging to
that already described.

VOL. I. PLATE II

•*.
•'

I

p<v,i 236

Cambridge University Pros

OX A NEW FORM OF THE PLATOMETER. 237

When the tracing point is made to traverse the boundary of a plane figure,
there is a continued rotation of the radius-vector combined with a change of
length. The rotation causes the index-sphere to roll on the fixed hemisphere,
while the length of the radius-vector determines the rate of its motion about its
axis, so that its whole motion measures the area swept out by the radius-vector
during the motion of the tracing point.

The areas measured by this instrument may either lie on one side of the
pillar, or they may extend all round it. In either case the action of the
instrument is the same as in the ordinary case. In this form of the instrument
we have the advantages of a fixed stand, and a simple motion of the tracing
point ; but there seem to be difficulties in the way of supporting the spheres
and arranging the slide ; and even then the instrument would require a tall
pillar, in order to take in a large area.

16. It will be observed that I have said little or nothing about the prac-
tical details of these instruments. Many useful hints will be found in the large
work on Platometers, by Professor T. Gonnellu, who has given us an account
of the difficulties, as well as the results, of the construction of his most
elaborate instrument. He has also given some very interesting investigations
into the errors produced by various irregularities of construction, although, as
far as I am aware, he has not even suspected the error which the sliding of
the index-wheel over the disc must necessarily introduce. With respect to this,
and other points relating to the working of the instru lent, the memoir of
Mr Sang, in the Transactions of this Society, is the rr )st complete that I
have met with. It may, however, be as well to state, t] at at the time when
I devised the improvements here suggested, I had not se n that paper, though
I had seen the instrument standing at rest in the Crystal Palace.

EDINBURGH, 30th January, 1855.

NOTE. — Since the design of the above instrument was submitted to the Society of Arts,
I have met with a description of an instrument combining simplicity of construction with
the power of adaptation to designs of any size, and at the same time more portable than
any other instrument of the kind. Although it does not act by perfect rolling, and there-
fore belongs to a different class of instruments from that described in this paper, I think
that its simplicity, and the beauty of the principle on which it acts, render it worth the
attention of engineers and mechanists, whether practical or theoretical. A full account of
this instrument is to be found in Moigno's " Cosmos," 5th year, Vol. VIII., Part vni., p. 213,
published 20th February 1856. Description et Throne du planim&tre polaire, invente" par
<f. Amsler, de Schaffouse en Suisse.

CAMBRIDGE, 30th April, 1856.

[From the Cambridge Philosophical Society Proceedings, VoL I. pp. 173 — 175.]

X. On the Elementary Tlieory of Optical Instruments.

THE object of this communication was to shew how the magnitude and
position of the image of any object seen through an optical instrument could
be ascertained without knowing the construction of the instrument, by means
of data derived from two experiments on the instrument. Optical questions
are generally treated of with respect to the pencils of rays which pass through
the instrument. A pencil is a collection of rays which have passed through one
point, and may again do so, by some optical contrivance. Now if we suppose
all the points of a plane luminous, each will give out a pencil of rays, and
that collection of pencils which passes through the instrument may be treated
as a beam of light. In a pencil only one ray passes through any point of
space, unless that point be the focus. In a beam an infinite number of rays,
corresponding each to some point in the luminous plane, passes through any
point ; and we may, if we choose, treat this collection of rays as a pencil
proceeding from that point. Hence the same beam of light may be decomposed
into pencils in an infinite variety of ways ; and yet, since we regard it as the
same collection of rays, we may study its properties as a beam independently
of the particular way in which we conceive it analysed into pencils.

Now in any instrument the incident and emergent beams are composed
of the same light, and therefore every ray in the incident beam has a
corresponding ray in the emergent beam. We do not know their path within
the instrument, but before incidence and after emergence they are straight
lines, and therefore any two points serve to determine the direction of each.

Let us suppose the instrument such that it forms an accurate image of a
plane object in a given position. Then every ray which passes through a given

ON THE ELEMENTARY THEORY OF OPTICAL INSTRUMENTS. 239

point of the object before incidence passes through the corresponding point of
the image after emergence, and this determines one point of the emergent ray.
If at any other distance from the instrument a plane object has an accurate
image, then there will be two other corresponding points given in the incident
and emergent rays. Hence if we know the points in which an incident ray
meets the planes of the two objects, we may find the incident ray by joining
the points of the two images corresponding to them.

It was then shewn, that if the image of a plane object be distinct, flat, and
similar to the object for two different distances of the object, the image of any
other plane object perpendicular to the axis will be distinct, flat and similar
to the object.

When the object is at an infinite distance, the plaile of its image is the
principal focal plane, and the point where it cuts the axis is the principal
focus. The line joining any point in the object to the corresponding point of
the image cuts the axis at a fixed point called the focal centre. The distance
of the principal focus from the focal centre is called the principal focal length,
or simply the^bcaZ length.

There are two principal foci, etc., formed by incident parallel rays passing
in opposite directions through the instrument. If we suppose light always to
pass in the same direction through the instrument, then the focus of incident
rays when the emergent rays are parallel is the first principal focus, and the
focus of emergent rays when the incident rays are parallel is the second
principal focus.

Corresponding to these we have first and second focal centres and focal
lengths.

Now let (?! be the focus of incident rays, P1 the foot of the perpendicular
from (?j on the axis, Qt the focus of emergent rays, P3 the foot of the corre-
sponding perpendicular, F^ the first and second principal foci, A^A^ the first and
second focal centres, then

lines being positive when measured in the direction of the light. Therefore
the position and magnitude of the image of any object is found by a simple
proportion.

240 OX THE ELEMENTARY THEORY OF OPTICAL INSTRUMENTS.

In one important class of instruments there are no. principal foci or focal
centres. A telescope in which parallel rays emerge parallel is an instance. In
such instruments, if m be the angular magnifying power, the linear dimensions

of the image are - of the object, and the distance of the image of the object

from the image of the object-glass is — s of the distance of the object from

the object-glass. Rules were then laid down for the composition of instruments,
and suggestions for the adaptation of this method to second approximations, and
the method itself was considered with reference to the labours of Cotes, Smith,
Euler, Lagrange, and Gauss on the same subject.

[From the Report of tJie British Association, 1856.]

XI. On a Method of Drawing the Theoretical Forms of Faraday's Lines of

Force without Calculation.

/

THE method applies more particularly to those cases in which the lines
are entirely parallel to one plane, such as the lines of electric currents in a
thin plate, or those round a system of parallel electric currents. In such cases,
if we know the forms of the lines of force in any two cases, we may combine
them by simple addition of the functions on which the equations of the lines
depend. Thus the system of lines in a uniform magnetic field is a series of
parallel straight lines at equal intervals, and that for an infinite straight electric
current perpendicular to the paper is a series of concentric circles whose radii
are in geometric progression. Having drawn these two sets of lines on two
separate sheets of paper, and laid a third piece above, draw a third set of lines
through the intersections of the first and second sets. This will be the system
of lines in a uniform field disturbed by an electric current. The most interesting
cases are those of uniform fields disturbed by a small magnet. If we draw a
circle of any diameter with the magnet for centre, and join those points in which
the circle cuts the lines of force, the straight lines so drawn will be parallel and
equidistant ; and it is easily shown that they represent the actual lines of
force in a paramagnetic, diamagnetic, or crystallized body, according to the
nature of the original lines, the size of the circle, &c. No one can study
Faraday's researches without wishing to see the forms of the lines of force.
This method, therefore, by which they may be easily drawn, is recommended
to the notice of electrical students.

VOL. I. 31

[From the Report of the British Association, 1856.]

XII. On the Unequal Sensibility of the Foramen Centrak to Light of

different Colours.

WHEN observing the spectrum formed by looking at a long vertical slit
through a simple prism, I noticed an elongated dark spot running up and down
in the blue, and following the motion of the eye as it moved up and doicn
the spectrum, but refusing to pass out of the blue into the other colours. It
was plain that the spot belonged both to the eye and to the blue part of the
spectrum. The result to which I have come is, that the appearance is due to
the yellow spot on the retina, commonly called the Foramen Centrale of Soem-
mering. The most convenient method of observing the spot is by presenting
to the eye in not too rapid succession, blue and yellow glasses, or, still better,
allowing blue and yellow papers to revolve slowly before the eye. In this way
the spot is seen in the blue. It fades rapidly, but is renewed every time the
yellow comes in to relieve the effect of the blue. By using a Nicol's prism
along with this apparatus, the brushes of Haidinger are well seen in connexion
with the spot, and the fact of the brushes being the spot analysed by polarized
light becomes evident. If we look steadily at an object behind a series of bright
bars which move in front of it, we shall see a curious bending of the bars as
they come up to the place of the yellow spot. The part which comes over the
spot seems to start in advance of the rest of the bar, and this would seem to
indicate a greater rapidity of sensation at the yellow spot than in the surround-
ing retina. But I find the experiment difficult, and I hope for better results
from more accurate observers.

[From the Report of the British Association, 1856.]

XIII. On the Tfieory of Compound Colours with reference to Mixtures of Blue

and Yellow Light.

WHEN we mix together blue and yellow paint, we obtain green paint. This
fact is well known to all who have handled colours ; and it is universally
admitted that blue and yellow make green. Red, yellow, and blue, being the
primary colours among painters, green is regarded as a secondary colour, arising
from the mixture of blue and yellow. Newton, however, found that the green
of the spectrum was not the same thing as the mixture of two colours of the
spectrum, for such a mixture could be separated by the prism, while the green
of the spectrum resisted further decomposition. But still it was believed that
yellow and blue would make a green, though not that of the spectrum. As
far as I am aware, the first experiment on the subject is that of M. Plateau,
who, before 1819, made a disc with alternate sectors of prussian blue and gam-
boge, and observed that, when spinning, the resultant tint was not green, but
a neutral gray, inclining sometimes to yellow or blue, but never to green. Prof.
J. D. Forbes of Edinburgh made similar experiments in 1849, with the same
result. Prof. Helmholtz of Konigsberg, to whom we owe the most complete
investigation on visible colour, has given the true explanation of this phenomenon.
The result of mixing two coloured powders is not by any means the same as
mixing the beams of light which flow from each separately. In the latter case
we receive all the light which comes either from the one powder or the other.
In the former, much of the light coming from one powder falls on particles of
the other, and we receive only that portion which has escaped absorption by one
or other. Thus the light coming from a mixture of blue and yellow powder,
consists partly of light coming directly from blue particles or yellow particles,
and partly of light acted on by both blue and yellow particles. This latter light
is green, since the blue stops the red, yellow, and orange, and the yellow stops

31—2

244 ON THE THEORY OF COMPOUND COLOURS.

the blue and violet. I have made experiments on the mixture of blue and
yellow light — by rapid rotation, by combined reflexion and transmission, by view-
ing them out of focus, in stripes, at a great distance, by throwing the colours
of the spectrum on a screen, and by receiving them into the eye directly ; and
I have arranged a portable apparatus by which any one may see the result of
this or any other mixture of the colours of the spectrum. In all these cases
blue and yellow do not make green. I have also made experiments on the
mixture of coloured powders. Those which I used principally were " mineral
blue" (from copper) and "chrome-yellow." Other blue and yellow pigments gave
curious results, but it was more difficult to make the mixtures, and the greens
were less uniform in tint. The mixtures of these colours were made by weight,
and were painted on discs of paper, which were afterwards treated in the manner
described in my paper " On Colour as perceived by the Eye," in the Transactions
of the Royal Society of Edinburgh, Vol. xxi. Part 2. The visible effect of the
colour is estimated in terms of the standard-coloured papers : — vermilion (V),
ultramarine (U), and emerald-green (E). The accuracy of the results, and their
significance, can be best understood by referring to the paper before mentioned.
I shall denote mineral blue by B, and chrome-yellow by Y ; and B, Y, means
a mixture of three parts blue and five parts yellow.

Coefficient
of brightness.

45

37

49

54

56

64

76

109

277

The columns V, U, E give the proportions of the standard colours which
are equivalent to 100 of the given colour; and the sum of V, U, E gives a co-
efficient, which gives a general idea of the brightness. It will be seen that the
first admixture of yellow diminishes the brightness of the blue. The negative
values of U indicate that a mixture of V, U, and E cannot be made equivalent
to the given colour. The experiments from which these results were taken had

Given Colour.

Standard Colours.
V. U. E.

B,

,

100

_ 2

36

7

B,

Y,,

100

= 1

18

17

B.

Y, ,

100

= 4

11

34

B,

Y, ,

100

= 9

5

40

B4

Y4,

100

= 15

1

40

B.

Y. ,

100

= 22

- 2

44

B.

Y. ,

100

= 35

-10

51

B,

Y7,

100

= 64

-19

64

Y, ,

100

= 180

-27

124

ON THE THEORY OF COMPOUND COLOURS. 245

the negative values transferred to the other side of the equation. They were
all made by means of the colour-top, and were verified by repetition at different
times. It may be necessary to remark, in conclusion, with reference to the mode
of registering visible colours in terms of three arbitrary standard colours, that it
proceeds upon that theory of three primary elements in the sensation of colour,
which treats the investigation of the laws of visible colour as a branch of human
physiology, incapable of being deduced from the laws of light itself, as set forth
in physical optics. It takes advantage of the methods of optics to study vision
itself; and its appeal is not to physical principles, but to our consciousness of
our own sensations.

[From the Report of Hie British Association, 1856.]

XIV. On an Instrument to illustrate Poinsdt's Theory of Rotation.

IN studying the rotation of a solid body according to Poins&t's method, we
have to consider the successive positions of the instantaneous axis of rotation
with reference both to directions fixed in space and axes assumed in the moving
body. The paths traced out by the pole of this axis on the invariable plane and
on the central ellipsoid form interesting subjects of mathematical investigation.
But when we attempt to follow with our eye the motion of a rotating body,
we find it difficult to determine through what point of the body the instantaneous
axis passes at any time, — and to determine its path must be still more difficult.
I have endeavoured to render visible the path of the instantaneous axis, and to
vary the circumstances of motion, by means of a top of the same kind as that
used by Mr Elliot, to illustrate precession*. The body of the instrument is a
hollow cone of wood, rising from a ring, 7 inches in diameter and 1 inch thick.
An iron axis, 8 inches long, screws into the vertex of the cone. The lower
extremity has a point of hard steel, which rests in an agate cup, and forms the
support of the instrument. An iron nut, three ounces in weight, is made to
screw on the axis, and to be fixed at any point; and in the wooden ring are
screwed four bolts, of three ounces, working horizontally, and four bolts, of one
ounce, working vertically. On the upper part of the axis is placed a disc of
card, on which are drawn four concentric rings. Each ring is divided into four
quadrants, which are coloured red, yellow, green, and blue. The spaces between
the rings are white. When the top is in motion, it is easy to see in which quad-
rant the instantaneous axis is at any moment and the distance between it and
the axis of the instrument ; and we observe, — 1st. That the instantaneous axis
travels in a closed curve, and returns to its original position in the body. 2ndly.

* Trantactions of the Royal Scottish Society of Arts, 1855.

ON AN INSTRUMENT TO ILLUSTRATE POINSOT's THEORY OF ROTATION. 247

That by working the vertical bolts, we can make the axis of the instrument
the centre of this closed curve. It will then be one of the principal axes of
inertia. 3rdly. That, by working the nut on the axis, we can make the order
of colours either red, yellow, green, blue, or the reverse. When the order of
colours is in the same direction as the rotation, it indicates that the axis of the
instrument is that of greatest moment of inertia. 4thly. That if we screw the
two pairs of opposite horizontal bolts to different distances from the axis, the
path of the instantaneous pole will no longer be equidistant from the axis, but
will describe an ellipse, whose longer axis is in the direction of the mean axis
of the instrument. 5thly. That if we now make one of the two horizontal axes
less and the other greater than the vertical axis, the instantaneous pole will
separate from the axis of the instrument, and the axis will incline more and more
till the spinning can no longer go on, on account of the 'obliquity. It is easy
to see that, by attending to the laws of motion, we may produce any of the
above effects at pleasure, and illustrate many different propositions by means of
the same instrument.

[From the Transactions of the Royal Society of Edinburgh, Vol. XXI. Part IV.]

XV. On a Dynamical Top, for exhibiting the phenomena of the motion of a
system of invariable form about a fcced point, with some suggestions as to
the Earth's motion.

(Read 20th April, 1857.)

To those who study the progress of exact science, the common spinning-top
is a symbol of the labours and the perplexities of men who had successfully
threaded the mazes of the planetary motions. The mathematicians of the last
age, searching through nature for problems worthy of their analysis, found in
this toy of their youth, ample occupation for their highest mathematical powers.

No illustration of astronomical precession can be devised more perfect than
that presented by a properly balanced top, but yet the motion of rotation has
intricacies far exceeding those of the theory of precession.

Accordingly, we find Euler and D'Alembert devoting their talent and their
patience to the establishment of the laws of the rotation of solid bodies.
Lagrange has incorporated his own analysis of the problem with his general
treatment of mechanics, and since his time M. Poins6t has brought the subject
under the power of a more searching analysis than that of the calculus, in
which ideas take the place of symbols, and intelligible propositions supersede
equations.

In the practical department of the subject, we must notice the rotatory
machine of Bohnenberger, and the nautical top of Troughton. In the first of
these instruments we have the model of the Gyroscope, by which Foucault has
been able to render visible the effects of the earth's rotation. The beautiful
experiments by which Mr J. Elliot has made the ideas of precession so familiar
to us are performed with a top, similar in some respects to Troughton's, though
not borrowed from his.

ON A DYNAMICAL TOP. 249

The top which I have the honour to spin before the Society, differs from
that of Mr Elliot in having more adjustments, and in being designed to exhibit
far more complicated phenomena.

The arrangement of these adjustments, so as to produce the desired effects,
depends on the mathematical theory of rotation. The method of exhibiting the
motion of the axis of rotation, by means of a coloured disc, is essential to the
success of these adjustments. This optical contrivance for rendering visible the
nature of the rapid motion of the top, and the practical methods of applying
the theory of rotation to such an instrument as the one before us, are the
grounds on which I bring my instrument and experiments before the Society
as my own.

I propose, therefore, in the first place, to give a brief- outline of such parts
of the theory of rotation as are necessary for the explanation of the phenomena
of the top.

I shall then describe the instrument with its adjustments, and the effect of
each, the mode of observing of the coloured disc when the top is in motion, and
the use of the top in illustrating the mathematical theory, with the method of
making the different experiments.

Lastly, I shall attempt to explain the nature of a possible variation in the
earth's axis due to its figure. This variation, if it exists, must cause a periodic
inequality in the latitude of every place on the earth's surface, going through its
period in about eleven months. The amount of variation must be very small,
but its character gives it importance, and the necessary observations are already
made, and only require reduction.

On the Theory of Rotation.

The theory of the rotation of a rigid system is strictly deduced from the
elementary laws of motion, but the complexity of the motion of the particles of
a body freely rotating renders the subject so intricate, that it has never been
thoroughly understood by any but the most expert mathematicians. Many who
have mastered the lunar theory have come to erroneous conclusions on this sub-
ject ; and even Newton has chosen to deduce the disturbance of the earth's axis
from his theory of the motion of the nodes of a free orbit, rather than attack
the problem of the rotation of a solid body.

VOL. I. 32

250 ON A DYNAMICAL TOP.

The method by which M. Poinsot has rendered the theory more manageable,
is by the liberal introduction of "appropriate ideas," chiefly of a geometrical
character, most of which had been rendered familiar to mathematicians by the
writings of Monge, but which then first became illustrations of this branch of
dynamics. If any further progress is to be made in simplifying and arranging
the theory, it must be by the method which Poins6t has repeatedly pointed out
as the only one which can lead to a true knowledge of the subject, — that of
proceeding from one distinct idea to another, instead of trusting to symbols and
equations.

An important contribution to our stock of appropriate ideas and methods has
lately been made by Mr R. B. Hayward, in a paper, " On a Direct Method of
estimating Velocities, Accelerations, and all similar quantities, with respect to axes,
moveable in any manner in Space." (Trans. Cambridge Phil. Soc. Vol. x. Part I.)

* In this communication I intend to confine myself to that part of the
subject which the top is intended to illustrate, namely, the alteration of the
position of the axis in a body rotating freely about its centre of gravity. I
shall, therefore, deduce the theory as briefly as possible, from two considera-
tions only, — the permanence of the original angular momentum in direction and
magnitude, and the permanence of the original vis viva.

The mathematical difficulties of the theory of rotation arise chiefly from
the want of geometrical illustrations and sensible images, by which we might
fix the results of analysis in our minds.

It is easy to understand the motion of a body revolving about a fixed axle.
Every point in the body describes a circle about the axis, and returns to its
original position after each complete revolution. But if the axle itself be in
motion, the paths of the different points of the body will no longer be circular
or re-entrant. Even the velocity of rotation about the axis requires a careful
definition, and the proposition that, in all motion about a fixed point, there is
always one line of particles forming an instantaneous axis, is usually given in
the form of a very repulsive mass of calculation. Most of these difficulties may
be got rid of by devoting a little attention to the mechanics and geometry of
the problem before entering on the discussion of the equations.

Mr Hayward, in his paper already referred to, has made great use of the
mechanical conception of Angular Momentum.

* 7th May, 1857. The paragraphs marked thus have 1*en rewritten since the paper was

ON A DYNAMICAL TOP. 251

DEFINITION. — The Angular Momentum of a particle about an axis is mea-
sured by the product of the mass of the particle, its velocity resolved in the normal
plane, and the perpendicular from the axis on the direction of motion.

* The angular momentum of any system about an axis is the algebraical
sum of the angular momenta of its parts.

As the rate of change of the linear momentum of a particle measures the
moving force which acts on it, so the rate of change of angular momentum
measures the moment of that force about an axis.

All actions between the parts of a system, being pairs of equal and opposite
forces, produce equal and opposite changes in the angular momentum of those
parts. Hence the whole angular momentum of the system is not affected by
these actions and re-actions. /

* When a system of invariable form revolves about an axis, the angular
velocity of every part is the same, and the angular momentum about the axis is
the product of the angular velocity and the moment of inertia about that axis.

* It is only in particular cases, however, that the whole angular momentum
can be estimated in this way. In general, the axis of angular momentum differs
from the axis of rotation, so that there will be a residual angular momentum
about an axis perpendicular to that of rotation, unless that axis has one of three
positions, called the principal axes of the body.

By referring everything to these three axes, the theory is greatly simplified.
The moment of inertia about one of these axes is greater than that about any
other axis through the same point, and that about one of the others is a mini-
mum. These two are at right angles, and the third axis is perpendicular to
their plane, and is called the mean axis.

* Let A, B, C be the moments of inertia about the principal axes through
the centre of gravity, taken in order of magnitude, and let ^ w2 &>3 be the
angular velocities about them, then the angular momenta will be A(olt Bu«
and C<a3.

Angular momenta may be compounded like forces or velocities, by the
law of the " parallelogram," and since these three are at right angles to each
other, their resultant is

+ CV = H (1), . ,,

and this must be constant, both in magnitude and direction in space, since no
external forces act on the body.

32—2

252 ON A DYNAMICAL TOP.

We shall call this axis of angular momentum the invariable axis. It is
perpendicular to what has been called the invariable plane. Poins6t calls it
the axis of the couple of impulsion. The direction-cosines of this axis in the
body are,

, AQ>I B(ot

= ' = ~> *

Since I, m and n vary during the motion, we need some additional
condition to determine the relation between them. We find this in the property
of the vis viva of a system of invariable form in which there is no friction.
The vis viva of such a system must be constant. We express this in the
equation

Atf + B^+Cu^V ................................. (2).

Substituting the values of «,, o>2, to, in terms of I, m, n,

Z* m| n'_JT
A+ B + C~H>'

Let ^=a\ B = ^' C=C>' Ht = e''

and this equation becomes

a'P+fc'm'+c'n^e1 ................................ (3),

and the equation to the cone, described by the invariable axis within the
body, is

(a1 - e*) x> + (bt-et) tf + (c'-e')zs = 0 ...................... (4).

The intersections of this cone with planes perpendicular to the principal
axes are found by putting x, y, or z, constant in this equation. By giving
e various values, all the different paths of the pole of the invariable axis,
corresponding to different initial circumstances, may be traced.

*In the figures, I have supposed a" = 100, 6J=107, and c*=110. The
first figure represents a section of the various cones by a plane perpendicular
to the axis of x, which is that of greatest moment of inertia. These sections
are ellipses having their major axis parallel to the axis of 6. The value of e'
corresponding to each of these curves is indicated by figures beside the curve.
The ellipticity increases with the size of the ellipse, so that the section
corresponding to e*=107 would be two parallel straight lines (beyond the bounds
of the figure), after which the sections would be hyperbolas.

ON A DYNAMICAL TOP. 253

*The second figure represents the sections made by a plane, perpendicular
to the mean axis. They are all hyperbolas, except when e3=107, when the
section is two intersecting straight lines.

The third figure shows the sections perpendicular to the axis of least
moment of inertia. From e'=110 to e2=107 the sections are ellipses, e2=107
gives two parallel straight lines, and beyond these the curves are hyperbolas.

*The fourth and fifth figures show the sections of the series of cones

made by a cube and a sphere respectively. The use of these figures is to

exhibit the connexion between the different curves described about the three
principal axes by the invariable axis during the motion of the body.

*We have next to compare the velocity of the invariable axis with respect
to the body, with that of the body itself round one of the principal axes.
Since the invariable axis is fixed in space, its motion relative to the body
must be equal and opposite to that of the portion of the body through which
it passes. Now the angular velocity of a portion of the body whose direction -
cosines are I, m, n, about the axis of x is

Substituting the values of <a,, a>,, <u,, in terms of I, m, n, and taking
account of equation (3), this expression becomes

1-P
Changing the sign and putting I = -jp. we have the angular velocity of

Ct £l

the invariable axis about that of x

— a*

always positive about the axis of greatest moment, negative about that of least
moment, and positive or negative about the mean axis according to the value
of e*. The direction of the motion in every case is represented by the arrows
in the figures. The arrows on the outside of each figure indicate the direction
of rotation of the body.

*If we attend to the curve described by the pole of the invariable axis

254 ON A DYNAMICAL TOP.

on the sphere in fig. 5, we shall see that the areas described by that point,
if projected on the plane of yz, are swept out at the rate

ta,

a'

Now the semi-axes of the projection of the spherical ellipse described by
the pole are

Dividing the area of this ellipse by the area described during one revo-
lution of the body, we find the number of revolutions of the body during
the description of the ellipse —

The projections of the spherical ellipses upon the plane of yz are all
similar ellipses, and described in the same number of revolutions; and in each
ellipse so projected, the area described in any time is proportional to the
number of revolutions of the body about the axis of x, so that if we measure
time by revolutions of the body, the motion of the projection of the pole of
the invariable axis is identical with that of a body acted on by an attractive
central force varying directly as the distance. In the case of the hyperbolas
in the plane of the greatest and least axis, this force must be supposed
repulsive. The dots in the figures 1, 2, 3, are intended to indicate roughly
the progress made by the invariable axis during each revolution of the body
about the axis of x, y and z respectively. It must be remembered that the
rotation about these axes varies with their inclination to the invariable axis,
so that the angular velocity diminishes as the inclination increases, and there-
fore the areas in the ellipses above mentioned are not described with uniform
velocity in absolute time, but are less rapidly swept out at the extremities of
the major axis than at those of the minor.

*When two of the axes have equal moments of inertia, or b = c, then
the angular velocity <a, is constant, and the path of the invariable axis is
circular, the number of revolutions of the body during one circuit of the
invariable axis, being

a1
&•-<»'•

ON A DYNAMICAL TOP. 255

The motion is in the same direction as that of rotation, or in the opposite
direction, according as the axis of x is that of greatest or of least moment
of inertia.

*Both in this case, and in that in which the three axes are unequal, the
motion of the invariable axis in the body may be rendered very slow by
diminishing the difference of the moments of inertia. The angular velocity of
the axis of x about the invariable axis in space is

'aa(l-Z')'

which is greater or less than o»,, as e* is greater or less than a?, and, when
these quantities are nearly equal, is very nearly the same as ^ itself This
quantity indicates the rate of revolution of the axle of the top about its
mean position, and is very easily observed.

*The instantaneous axis is not so easily observed. It revolves round the
invariable axis in the same time with the axis of x, at a distance which is very
small in the case when a, b, c, are nearly equal. From its rapid angular motion
in space, and its near coincidence with the invariable axis, there is no advantage
in studying its motion in the top.

*By making the moments of inertia very unequal, and in definite proportion
to each other, and by drawing a few strong lines as diameters of the disc, the
combination of motions will produce an appearance of epicycloids, which are the
result of the continued intersection of the successive positions of these lines, and
the cusps of the epicycloids lie in the curve in which the instantaneous axis
travels. Some of the figures produced in this way are very pleasing.

In order to illustrate the theory of rotation experimentally, we must have
a body balanced on its centre of gravity, and capable of having its principal
axes and moments of inertia altered in form and position within certain limits.
We must be able to make the axle of the instrument the greatest, least, or
mean principal axis, or to make it not a principal axis at all, and we must be
able to see the position of the invariable axis of rotation at any time. There
must be three adjustments to regulate the position of the centre of gravity,
three for the magnitudes of the moments of inertia, and three for the directions
of the principal axes, nine independent adjustments, which may be distributed
as we please among the screws of the instrument.

256 ON A DYNAMICAL TOP.

The form of the body of the instrument which I have found most suitable is
that of a bell (p. 262, fig. 6). C is a hollow cone of brass, R is a heavy
ring cast in the same piece. Six screws, with heavy heads, x, y, z, x', y1, z',
work horizontally in the ring, and three similar screws, I, m, n, work vertically
through the ring at equal intervals. AS is the axle of the instrument, SS is
a brass screw working in the upper part of the cone C, and capable of being
firmly clamped by means of the nut c. B is a cylindrical brass bob, which may
lie screwed up or down the axis, and fixed in its place by the nut b.

The lower extremity of the axle is a fine steel point, finished without emery,
and afterwards hardened. It runs in a little agate cup set in the top of the
pillar P. If any emery had been embedded in the steel, the cup would soon
be worn out. The upper end of the axle has also a steel point by which it may
be kept steady while spinning.

When the instrument is in use, a coloured disc is attached to the upper
end of the axle.

It will be seen that there are eleven adjustments, nine screws in the brass
ring, the axle screwing in the cone, and the bob screwing on the axle. The
advantage of the last two adjustments is, that by them large alterations can be
made, which are not possible by means of the small screws.

The first thing to be done with the instrument is, to make the steel point
at the end of the axle coincide with the centre of gravity of the whole. This
is done roughly by screwing the axle to the right place nearly, and then balancing
the instrument on its point, and screwing the bob and the horizontal screws till
the instrument will remain balanced in any position in which it is placed.

When this adjustment is carefully made, the rotation of the top has no
tendency to shake the steel point in the agate cup, however irregular the motion
may appear to be.

The next thing to be done, is to make one of the principal axes of the
central ellipsoid coincide with the axle of the top.

To effect this, we must begin by spinning the top gently about its axle,
steadying the upper part with the finger at first. If the axle is already a
principal axis the top will continue to revolve about its axle when the finger is
removed. If it is not, we observe that the top begins to spin about some other
axis, and the axle moves away from the centre of motion and then back to it
again, and so on, alternately widening its circles and contracting them.

ON A DYNAMICAL TOP. 257

It is impossible to observe this motion successfully, without the aid of the
coloured disc placed near the upper end of the axis. This disc is divided into
sectors, and strongly coloured, so that each sector may be recognised by its colour
when in rapid motion. If the axis about which the top is really revolving, falls
within this disc, its position may be ascertained by the colour of the spot at the
centre of motion. If the central spot appears red, we know that the invariable
axis at that instant passes through the red part of the disc.

In this way we can trace the motion of the invariable axis in the revolving
body, and we find that the path which it describes upon the disc may be a circle,
an ellipse, an hyperbola, or a straight line, according to the arrangement of the
instrument.

In the case in which the invariable axis coincides at 'first with the axle of
the top, and returns to it after separating from it for a time, its true path is
a circle or an ellipse having the axle in its circumference. The true principal
axis is at the centre of the closed curve. It must be made to coincide with the
axle by adjusting the vertical screws I, m, n.

Suppose that the colour of the centre of motion, when farthest from the
axle, indicated that the axis of rotation passed through the sector L, then the
principal axis must also lie in that sector at half the distance from the axle.

If this principal axis be that of greatest moment of inertia, we must raise
the screw I in order to bring it nearer the axle A. If it be the axis of least
moment we must lower the screw I. In this way we may make the principal
axis coincide with the axle. Let us suppose that the principal axis is that of
greatest moment of inertia, and that we have made it coincide with the axle of
the instrument. Let us also suppose that the moments of inertia about the
other axes are equal, and very little less than that about the axle. Let the top
be spun about the axle and then receive a disturbance which causes it to spin
about some other axis. The instantaneous axis will not remain at rest either
in space or in the body. In space it will describe a right cone, completing a
revolution in somewhat less than the time of revolution of the top. In the
body it will describe another cone of larger angle in a period which is longer
as the difference of axes of the body is smaller. The invariable axis will be
fixed in space, and describe a cone in the body.

The relation of the different motions may be understood from the following
illustration. Take a hoop and make it revolve about a stick which remains at
rest and touches the inside of the hoop. The section of the stick represents the

VOL. i. 33

258 ON A DYNAMICAL TOP.

path of the instantaneous axis in space, the hoop that of the same axis in the
body, and the axis of the stick the invariable axis. The point of contact repre-
sents the pole of the instantaneous axis itself, travelling many times round the
stick before it gets once round the hoop. It is easy to see that the direction in
which the instantaneous axis travels round the hoop, is in this case the same as
that in which the hoop moves round the stick, so that if the top be spinning in
the direction L, M, N, the colours will appear in the same order.

By screwing the bob B up the axle, the difference of the axes of inertia
may be diminished, and the time of a complete revolution of the invariable
axis in the body increased. By observing the number of revolutions of the top
in a complete cycle of colours of the invariable axis, we may determine the
ratio of the moments of inertia.

By screwing the bob up farther, we may make the axle the principal axis of
least moment of inertia.

The motion of the instantaneous axis will then be that of the point of
contact of the stick with the outside of the hoop rolling on it. The order of
colours will be N, M, L, if the top be spinning in the direction L, M, Nt and
the more the bob is screwed up, the more rapidly will the colours change, till
it ceases to be possible to make the observations correctly.

In calculating the dimensions of the parts of the instrument, it is necessary
to provide for the exhibition of the instrument with its axle either the greatest
or the least axis of inertia. The dimensions and weights of the parts of the top
which I have found most suitable, are given in a note at the end of this paper.

Now let us make the axes of inertia in the plane of the ring unequal. We
may do this by screwing the balance screws x and x1 farther from the axle
without altering the centre of gravity.

Let us suppose the bob B screwed up so as to make the axle the axis of
least inertia. Then the mean axis is parallel to xxl, and the greatest is at right
angles to xx1 in the horizontal plane. The path of the invariable axis on the
disc is no longer a circle but an ellipse, concentric with the disc, and having
its major axis parallel to the mean axis xx1.

The smaller the difference between the moment of inertia about the axle and
about the mean axis, the more eccentric the ellipse will be; and if, by screwing
the bob down, the axle be made the mean axis, the path of the invariable axis
will be no longer a closed curve, but an hyperbola, so that it will depart alto-
gether from the neighbourhood of the axle. When the top is in this condition

ON A DYNAMICAL TOP. 259

it must be spun gently, for it is very difficult to manage it when its motion
gets more and more eccentric.

When the bob is screwed still farther down, the axle becomes the axis of
greatest inertia, and xtf the least. The major axis of the ellipse described by
the invariable axis will now be perpendicular to axe1, and the farther the bob
is screwed down, the eccentricity of the ellipse will diminish, and the velocity
with which it is described will increase.

I have now described all the phenomena presented by a body revolving freely
on its centre of gravity. If we wish to trace the motion of the invariable axis
by means of the coloured sectors, we must make its motion very slow compared
with that of the top. It is necessary, therefore, to make the moments of inertia
about the principal axes very nearly equal, and in this case a very small change
in the position of any part of the top will greatly derange the position of the
principal axis. So that when the top is well adjusted, a single turn of one of
the screws of the ring is sufficient to make the axle no longer a principal axis,
and to set the true axis at a considerable inclination to the axle of the top.

All the adjustments must therefore be most carefully arranged, or we may
have the whole apparatus deranged by some eccentricity of spinning. The method
of making the principal axis coincide with the axle must be studied and prac-
tised, or the first attempt at spinning rapidly may end in the destruction of
the top, if not of the table on which it is spun.

On the Earth's Motion.

We must remember that these motions of a body about its centre of gra-
vity, are not illustrations of the theory of the precession of the Equinoxes.
Precession can be illustrated by the apparatus, but we must arrange it so that
the force of gravity acts the part of the attraction of the sun and moon in
producing a force tending to alter the axis of rotation. This is easily done by
bringing the centre of gravity of the whole a little below the point on which
it spins. The theory of such motions is far more easily comprehended than
that which we have been investigating.

But the earth is a body whose principal axes are unequal, and from the
phenomena of precession we can determine the ratio of the polar and equatorial
axes of the "central ellipsoid;" and supposing the earth to have been set in
motion about any axis except the principal axis, or to have had its original

33—2

260 ON A DYNAMICAL TOP.

axis disturbed in any way, its subsequent motion would be that of the top
when the bob is a little below the critical position.

The axis of angular momentum would have an invariable position in space,
and would travel with respect to the earth round the axis of figure with a velo-

C—A

city = «D -'—j — where at is the sidereal angular velocity of the earth. The apparent
A

pole of the earth would travel (with respect to the earth) from west to east

A

round the true pole, completing its circuit in -^ — -j sidereal days, which appears

O — A

to be about 325*6 solar days.

The instantaneous axis would revolve about this axis in space in about
a day, and would always be in a plane with the true axis of the earth and
the axis of angular momentum. The effect of such a motion on the apparent
position of a star would be, that its zenith distance would be increased and
diminished during a period of 325 '6 days. This alteration of zenith distance
is the same above and below the pole, so that the polar distance of the star
is unaltered. In fact the method of finding the pole of the heavens by obser-
vations of stars, gives the pole of the invariable axis, which is altered only by
external forces, such as those of the sun and moon.

There is therefore no change in the apparent polar distance of stars due to
this cause. It is the latitude which varies. The magnitude of this variation
cannot be determined by theory. The periodic time of the variation may be
found approximately from the known dynamical properties of the earth. The
epoch of maximum latitude cannot be found except by observation, but it must
be later in proportion to the east longitude of the observatory.

In order to determine the existence of such a variation of latitude, I have
examined the observations of Polaris with the Greenwich Transit Circle in the
years 1851-2-3-4. The observations of the upper transit during each month were
collected, and the mean of each month found. The same was done for the lower
transits. The difference of zenith distance of upper and lower transit is twice
the polar distance of Polaris, and half the sum gives the co-latitude of Greenwich.

In this way I found the apparent co-latitude of Greenwich for each month
of the four years specified.

There appeared a very slight indication of a maximum belonging to the set
of months,

March^ 51. Feb. 52. Dec. 52. Nov. 53. Sept. 54.

ON A DYNAMICAL TOP. 261

This result, however, is to be regarded as very doubtful, as there did not
appear to be evidence for any variation exceeding half a second of space, and
more observations would be required to establish the existence of so small a
variation at alL

I therefore conclude that the earth has been for a long time revolving
about an axis very near to the axis of figure, if not coinciding with it. The
cause of this near coincidence is either the original softness of the earth, or
the present fluidity of its interior. The axes of the earth are so nearly equal,
that a considerable elevation of a tract of country might produce a deviation
of the principal axis within the limits of observation, and the only cause which
would restore the uniform motion, would be the action of a fluid which would
gradually diminish the oscillations of latitude. The permanence of latitude essen-
tially depends on the inequality of the earth's axes, for if they had been all
equal, any alteration of the crust of the earth would have produced new prin-
cipal axes, and the axis of rotation would travel about those axes, altering the
latitudes of all places, and yet not in the least altering the position of the
axis of rotation among the stars.

Perhaps by a more extensive search and analysis of the observations of
different observatories, the nature of the periodic variation of latitude, if it exist,
may be determined. I am not aware of any calculations having been made to
prove its non-existence, although, on dynamical grounds, we have every reason
to look for some very small variation having the periodic time of 325 '6 days
nearly, a period which is clearly distinguished from any other astronomical cycle,
and therefore easily recognised.

262 ON A DYNAMICAL TOP.

NOTE.
Dimensions and Weights of the parts of th« Dynamical Top.

I. Body of the top-
Mean diameter of ring, 4 inches.
Section of ring, $ inch square.

The conical portion rises from the upper and inner edge of the ring, a
height of 1| inches from the base.

The whole body of the top weighs . 1 Ib. 7 oz.

Each of the nine adjusting screws has its screw 1 inch long, and the

screw and head together weigh 1 ounce. The whole weigh . 9 „

II. Axle, &c.—

Length of axle 5 inches, of which £ inch at the bottom is occupied by
the steel point, 3£ inches are brass with a good screw turned on it,
and the remaining inch is of steel, with a sharp point at the top.

The whole weighs 1£ „

The bob B has a diameter of 1'4 inches, and a thickness of '4. It weighs 2f „

The nuts 6 and c, for clamping the bob and the body of the top on the

axle, each weigh ^ oz. . 1 „

Weight of whole top 2 Ib. 5\ oz.

The best arrangement, for general observations, is to have the disc of card divided
into four quadrants, coloured with vermilion, chrome yellow, emerald green, and ultramarine.
These are bright colours, and, if the vermilion is good, they combine into a grayish tint
when the revolution is about the axle, and burst into brilliant colours when the axis is
disturbed. It is useful to have some concentric circles, drawn with ink, over the colours,
and about 12 radii drawn in strong pencil lines. It is easy to distinguish the ink from
the pencil lines, as they cross the invariable axis, by their want of lustre. In this way,
the path of the invariable axis may be identified with great accuracy, and compared with
theory.

VOL. I. PLATE III.

FIG. 3.

FIG. 2.

flC 6.

To fact page 262

Cambridge University Press

[From the Philosophical Magazine, Vol. xiv.]

XVI. Account of Experiments on the Perception of Colour.

To the Editors of the Philosophical Magazine and Journal.

GENTLEMEN,

THE experiments which I intend to describe were undertaken in order
to render more perfect the quantitative proof of the theory of three primary
colours. According to that theory, every sensation of colour in a perfect human
eye is distinguished by three, and only three, elementary qualities, so that in
mathematical language the quality of a colour may be expressed as a function
of three independent variables. There is very little evidence at present for
deciding the precise tints of the true primaries. I have ascertained that a
certain red is the sensation wanting in colour-blind eyes, but the mathematical
theory relates to the number, not to the nature of the primaries. If, with Sir
David Brewster, we assume red, blue, and yellow to be the primary colours, this
amounts to saying that every conceivable tint may be produced by adding
together so much red, so much yellow, and so much blue. This is perhaps the
best method of forming a provisional notion of the theory. It is evident that if
any colour could be found which could not be accurately defined as so much of
each of the three primaries, the theory would fall to the ground. Besides this,
the truth of the theory requires that every mathematical consequence of assuming
every colour to be the result of mixture of three primaries should also be true.

I have made experiments on upwards of 100 different artificial colours, con-
sisting of the pigments used in the arts, and their mechanical mixtures. These
experiments were made primarily to trace the effects of mechanical mixture on
various coloured powders ; but they also afford evidence of the truth of the
theory, that all these various colours can be referred to three primaries. The

264 EXPERIMENTS ON THE PERCEPTION OF COLOUR,

following experiments relate to the combinations of six well-defined colours only,
and I shall describe them the more minutely, as I hope to induce those who
have good eyes to subject them to the same trial of skill in distinguishing
tints.

The method of performing the experiments is described in the Transactions
of the Royal Society of Edinburgh, Vol. xxi. Part 2. The colour- top or teetotum
which I used may be had of Mr J. M. Bryson, Edinburgh, or it may be easily
extemporized. Any rotatory apparatus which will keep a disc revolving steadily
and rapidly in a good light, without noise or disturbance, and can be easily
stopped and shifted, will do as well as the contrivance of the spinning-top.

The essential part of the experiment consists in placing several discs of
coloured paper of the same size, and slit along a radius, over one another, so
that a portion of each is seen, the rest being covered by the other discs. By
sliding the discs over each other the proportion of each colour may be varied,
and by means of divisions on a circle on which the discs lie, the proportion of
each colour may be read off. My circle was divided into 100 parts.

On the top of this set of discs is placed a smaller set of concentric discs,
so that when the whole is in motion round the centre, the colour resulting from
the mixture of colours of the small discs is seen in the middle of that arising
from the larger discs. It is the object of the experimenter to shift the colours
till the outer and inner tints appear exactly the same, and then to read off the
proportions.

It is easy to deduce from the theory of three primary colours what must
be the number of discs exposed at one time, and how much of each colour must
appear.

Every colour placed on either circle consists of a certain proportion of each
of the primaries, and in order that the outer and inner circles may have precisely
the same resultant colour in every respect, there must be the same amount of
each of the primary colours in the outer and inner circles. Thus we have as
many conditions to fulfil as there are primary colours ; and besides these we
have two more, because the whole number of divisions in either the outer or
the inner circle is 100, so that if there are three primary colours there will be
five conditions to fulfil, and this will require five discs to be disposable, and
these must be arranged so that three are matched against two, or four against one.
If we take six different colours, we may leave out any one of the six, and
so form six different combinations of five colours. It is plain that these six

EXPERIMENTS ON THE PERCEPTION OF COLOUR. 265

combinations must be equivalent to two equations only, if the theory of three
primaries be true.

The method which I have found most convenient for registering the result
of an experiment, after an identity of tint has been obtained in the inner and
outer circles, is the following: —

Write down the names or symbols of the coloured discs each at the top of
a column, and underneath write the number of degrees of that colour observed,
calling it + when the colour is in the outer circle, and — when it is in the inner
circle; then equate the whole to zero. In this way the account of each colour
is kept in a separate column, and the equations obtained are easily combined and
reduced, without danger of confounding the colours of which the quantities have
been measured. The following experiments were made between the 3rd and llth
of September, 1856, about noon of each day, in a room fronting the north,
without curtains or any bright- coloured object near the window. The same
combination was never made twice in one day, and no thought was bestowed
upon the experiments except at the time of observation. Of course the gradua-
tion was never consulted, nor former experiments referred to, till each combi-
nation of colours had been fixed by the eye alone ; and no reduction was
attempted till all the experiments were concluded.

The coloured discs were cut from paper painted of the following colours : —
Vermilion, Ultramarine, Emerald-green, Snow-white, Ivory-black, and Pale
Chrome-yellow. They are denoted by the letters V, U, G, W, B, Y respectively.
These colours were chosen, because each is well distinguished from the rest, so
that a small change of its intensity in any combination can be observed. Two
discs of each colour were prepared, so that in each combination the colours might
occasionally be transposed from the outer circle to the inner.

The first equation was formed by leaving out vermilion. The remaining
colours are Ultramarine-blue, Emerald-green, White, Black, and Yellow. We
might suppose, that by mixing the blue and yellow in proper proportions, we
should get a green of the same hue as the emerald-green, but not so intense,
so that in order to match it we should have to mix the green with white to
dilute it, and with black to make it darker. But it is not in this way that we
have to arrange the colours, for our blue and yellow produce a pinkish tint, and
never a green, so that we must add green to the combination of blue and yellow,
to produce a neutral tint, identical with a mixture of white and black.

VOL. I. 34

266 EXPERIMENTS ON THE PERCEPTION OP COLOUR.

Blue, green, and yellow must therefore be combined on the large discs, and
stand on one side of the equation, and black and white, on the small discs, must
stand on the other side. In order to facilitate calculations, the colours are
always put down in the same order; but those belonging to the small discs
are marked negative. Thus, instead of writing

we write + 54U + 14G - 32W - 68B + 32Y = 0.

The sum of all the positive terms of such an equation is 100, being the
whole number of divisions in the circle. The sum of the negative terms is
also 100.

The second equation consists of all the colours except blue ; and in this
way we obtain six different combinations of five colours.

Each of these combinations was formed by the unassisted judgment of my
eye, on six different occasions, so that there are thirty-six independent observa-
tions of equations between five colours.

Table I. gives the actual observations, with their dates.

Table II. gives the result of summing together each group of six equations.

Each equation in Table II. has the sums of its positive and negative co-
efficients each equal to 600.

Having obtained a number of observations of each combination of colours,
we have next to test the consistency of these results, since theoretically two
equations are sufficient to determine all the relations among six colours. We
must therefore, in the first place, determine the comparative accuracy of the
different sets of observations. Table III. gives the averages of the errors of
each of the six groups of observations. It appears that the combination IV. is
the least accurately observed, and that VI. is the best.

Table IV. gives the averages of the errors in the observation of each colour
in the whole series of experiments. This Table was computed in order to detect
any tendency to colour-blindness in my own eyes, which might be less accurate
in discriminating red and green, than in detecting variations of other colours.
It appears, however, that my observations of red and green were more accurate
than those of blue or yellow. White is the most easily observed, from the

EXPERIMENTS ON THE PERCEPTION OF COLOUR. 267

brilliancy of the colour, and black is liable to the greatest mistakes. I would
recommend this method of examining a series of experiments as a means of
detecting partial colour-blindness, by the different accuracy in observing differ-
ent colours. The next operation is to combine all the equations according to
their values. Each was first multiplied by a coefficient proportional to its ac.-
curacy, and to the coefficient of white in that equation. The result of adding
all the equations so found is given in equation (W).

Equation (Y) is the result of similar operations with reference to the
yellow on each equation.

We have now two equations from which to deduce six new equations, by
eliminating each of the six colours in succession. We must first combine the
equations, so as to get rid of one of the colours, and then we must divide by
the sum of the positive or negative coefficients, so as to reduce the equations
to the form of the observed equations. The results of these operations are given
in Table V., along with the means of each group of six observations. It will
be seen that the differences between the results of calculation from two equations
and the six independent observed equations are very small. The errors in red
and green are here again somewhat less than in blue and yellow, so that there
is certainly no tendency to mistake red and green more than other colours.
The average difference between the observed mean value of a colour and the
calculated value is 77 of a degree. The average error of an observation in any
group from the mean of that group was '92. No observation was attempted
to be registered nearer than one degree of the top, or ^fa of a circle ; so that
this set of observations agrees with the theory of three primary colours quite
as far as the observations can warrant us in our calculations ; and I think that
the human eye has seldom been subjected to so severe a test of its power of
distinguishing colours. My eyes are by no means so accurate in this respect as
many eyes I have examined, but a little practice produces great improvement
even in inaccurate observers.

I have laid down, according to Newton's method, the relative positions of
the five positive colours with which I worked. It will be seen that W lies
within the triangle VUG, and Y outside that triangle.

The first combination, Equation I., consisted of blue, yellow, and green,
taken in such proportions that their centre of gravity falls at W.

34—2

•268 EXl'KIUMENTS ON THE PERCEPTION OF COLOUR.

In Equation II. a mixture of red and green, represented in the diagram
by the point 2, is seen to be equivalent to a mixture of white and yellow, also
represented by 2, which is a pale yellow tint.

Equation III. is between a mixture of blue and yellow and another of
white and red. The resulting tint is at the intersection of YU and WV ; that
is, at the point 3, which represents a pale pink grey.

Equation IV. is between VG and UY, that is, at 4, a dirty yellow.

Equation V. is between a mixture of white, red, and green, and a mixture
of blue and yellow at the point 5, a pale dirty yellow.

Equation VI. has W. for its resulting tint.

Blue, U.

Bed, V^ -MA -^ O. Green.

Y, Yellow.

Of all the resulting tints, that of Equation IV. is the furthest from white ;

and we find that the observations of this equation are affected with the greatest

errors. Hence the importance of reducing the resultant tint to as nearly a
neutral colour as possible.

It is hardly necessary for me to observe, that the whole of the numerical
results which I have given apply only to the coloured papers which I used,
and to them only when illuminated by daylight from the north at mid-day in
September, latitude 55*. In the evening, or in winter, or by candlelight, the
results are very different. I believe, however, that the results would differ far
less if observed by different persons, than if observed under different lights ;
for the apparatus of vision is wonderfully similar in different eyes, and even in
colour-blind eyes the system of perception is not different, but defective.

EXPERIMENTS ON THE PERCEPTION OF COLOUR.

269

TABLE I. — The observations arranged in groups.

Equation I

v=o.

+u.

+G. -W.

-B.

+Y.

Equation IV.

-V. +U.

-G. W=0.

+B.

+ Y,

1856, Sept.

3.

0

54

12 34

66

34

1856, Sept. 3.

62 15

38 0

53

32

4.

0

58

14 31

69

28

4.

63 17

37 0

46

37

5.

0

55

12 32

68

33

5.

64 16

36 0

50

34

6.

0

54

14 32

68

32

6.

62 19

38 0

46

35

8.

0

54

14 32

68

32

8.

62 19

38 0

47

34

'J.

0

53

15 32

68

32

9.

63 17

37 0

49

34

Equation

II.

-V.

U=0.

-G. +W.

+B.

+Y.

Equation V.

+ V. -U.

+ G. +W.

B=0.

-Y.

Sept.

3.

59

0

41 9

71

20

Sept. 3.

56 47

28 16

0

53

4.

61

0

39 9

68

23

4.

57 50

25 18

0

50

5.

61

0

39 9

67

24

5.

56 49

24 20

0

51

6.

59

0

41 10

66

24

6.

55 47

27 18

0

53

8.

60

0

40 9

69

22

8.

54 49

26 20

0

51

9.

61

0

39 9

68

23

11.

56 50

27 17

0

50

Equation m. +V.

-U.

G=0. +W.

+B.

-Y.

Equation VI.

+ V. +U.

+ G. -W.

-B.

Y=0.

Sept.

3.

20

56

0 28

52

44

Sept. 3.

38 27

35 24

76

0

i.

23

58

0 30

47

42

4.

39 27

34 24

76

0

5.

24

56

0 29

47

44

5.

40 26

34 24

76

0

6.

20

56

0 31

49

44

6.

38 28

34 24

76

0

a

21

57

0 29

50

43

8.

39 28

33 24

76

0

'j.

21

58

0 29

50

42

11.

39 27

34 23

77

0

TABLE II. — The sums of the observed equations.

V.

TJ. G. W.

B.

Y.

Equation I.

0

+

328 + 81 -193

-407

+ 191

II.

-361

0 -239 + 55

+ 409

+ 136

III.

+ 129

_

341 0 + 176

+ 295

-259

IV.

-376

+

103 -224 0

+ 291

+ 206

V.

+ 334

292 + 157 + 109

0

-308

...

VI.

+ 233

+

163 +204 -143

-457

0

TABLE III. — The averages

of the errors

of the several equations

from the

means expressed in

ifo

parts of a circle.

Equations.

I.

n.

in. IV.

V.

VI.

Errors.

•94

•85

1-05 1-17

1-08

•40

TABLE IV. — The averages of the errors of the several colours from the means in

a circle.

parts of

Colours.
Errors.

V.
•83

U.
•99

G.
•80

W.
•61

B.
1-15

T.
1-09

Average error on the whole '92.

The equations from which the reduced results were obtained were calculated as follow : —

Equation for (W) = (II) + 2 (III) + (V)-2 (I) -4 (VI).
Equation for (Y) = 2 (I) + 2 (II) - 3 (III) + 2 (IV) - 3 (V).

270

EXPERIMENTS ON THE PERCEPTION OF COLOUR.

These operations being performed, gave

V. U. G. W. B.

(W) + 701 + 2282 + 1060-1474-3641 + 1072 = 0.
(Y) +2863 -2761 + 1235 + 1131 ••• 299-2767 = 0.

Y.

Kr-.ni tlu-se were obtained the following results by elimination: —

TABLE V.

Equation

T / From
l' (From

(W) and (Y)
olwervation

0
0

-54-1

-54-7

-13
-13

0

5

+ 32-0
+ 32-1

+ 68-0
+ 67-9

-32-0
-31-8

n.{

From
From

(W)and(Y)
observation

- 59-6
-60-2

0
0

-40-
-39-

1
8

+ 10
+ 9

•4

•2

+ 66

+ 68

•0
•2

+ 23-6
+ 22-6

m.{

From
From

(W) and (Y)
observation

-21-7
-21-5

+ 57-4
+ 56-8

0
0

-30
-29

..i
•3

-48
-49

•1

•2

+ 42-6
+ 432

IV. {

From
From

(W) and (Y)
observation

-62-4
-62-7

-• is-i;
+ 17-2

-37-

-37-

6
3

0

0

+ 45

+ 48

•"
•5

+ 35-7
+ 34-3

v'{

From

From

(W) and (Y)
observation

+ 55-6
+ 55-7

-49-0

-48-7

+ 25'
+ 26

2
1

+ 19

+ 18

•2
•2

0

0

-51-0
-51-3

v,{

From
From

(W) and (Y)
observation

-38-8

-26-6
-27-2

-33
-34'

7

0

+ 22
+ 28

•7

•3

+ 77
+ 76

2

0

0

JAMES CLERK MAXWELL.

M..UR, June 13, 1857.

[From The Quarterly Journal of Pure and Applied Mathematics, Vol. II.]

XVII. On the General Laws of Optical Instruments.

THE optical effects of compound instruments have been generally deduced
from those of the elementary parts of which they are composed. The formulae
given in most works on Optics for calculating the effect of each spherical sur-
face are simple enough, but, when we attempt to carry on our calculations from
one of these surfaces to the next, we arrive at fractional expressions so com-
plicated as to make the subsequent steps very troublesome.

Euler (Acad. R. de Berlin, 1757, 1761. Acad. K. de Paris, 1765) has attacked
these expressions, but his investigations are not easy reading. Lagrange (Acad.
Berlin, 1778, 1803) has reduced the case to the theory of continued fractions
and so obtained general laws.

Gauss (Dioptrische Untersuchungen, Gottingen, 1841) has treated the subject
with that combination of analytical skill with practical ability which he displays
elsewhere, and has made use of the properties of principal foci and principal
planes. An account of these researches is given by Prof. Miller in the third
volume of Taylor's Scientific Memoirs. It is also given entire in French by
M. Bravais in Liouville's Journal for 1856, with additions by the translator.

The method of Gauss has been followed by Prof. Listing in his Treatise
on the Dioptrics of the Eye (in Wagner's Handworterbuch der Physiologic) from
whom I copy these references, and by Prof. Helmholtz in his Treatise on
Physiological Optics (in Karsten's Cyclopadie).

The earliest general investigations are those of Cotes, given in Smith's
Optics, n. 76 (1738). The method there is geometrical, and perfectly general,
but proceeding from the elementary cases to the more complex by the method
of mathematical induction. Some of his modes of expression, as for instance his
measure of "apparent distance," have never come into use, although his results
may easily be expressed more intelligibly ; and indeed the whole fabric of

272 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

Geometrical Optics, as conceived by Cotes and laboured by Smith, has fallen
into neglect, except among the writers before named. Smith tells us that it
was with reference to these optical theorems that Newton said "If Mr Cotes
had lived we might have known something."

The investigations which I now offer are intended to show how simple and
how general the theory of instruments may be rendered, by considering the
optical effects of the entire instrument, without examining the mechanism by
which those effects are obtained. I have thus established a theory of " perfect
instruments," geometrically complete in itself, although I have also shown, that
no instrument depending on refraction and reflexion, (except the plane mirror)
can be optically perfect. The first part of this theory was communicated to
the Philosophical Society of Cambridge, 28th April, 1856, and an abstract will
be found in the Philosophical Magazine, November, 1856. Propositions VIII.
and IX. are now added. I am not aware that the last has been proved before.

In the following propositions I propose to establish certain rules for deter-
mining, from simple data, the path of a ray of light after passing through any
optical instrument, the position of the conjugate focus of a luminous point, and
the magnitude of the image of a given object. The method which I shall use
does not require a knowledge of the internal construction of the instrument and
derives all its data from two simple experiments.

There are certain defects incident to optical instruments from which, in the
elementary theory, we suppose them to be free. A perfect instrument must
fulfil three conditions:

I. Every ray of the pencil, proceeding from a single point of the object,
must, after passing through the instrument, converge to, or diverge from, a
single point of the image. The corresponding defect, when the emergent rays
have not a common focus, has been appropriately called (by Dr Whewell)
Astigmatism.

II. If the object is a plane surface, perpendicular to the axis of the
instrument, the image of any point of it must also lie in a plane perpendicular
to the axis. When the points of the image lie in a curved surface, it is said
to have the defect of curvature.

III. The image of an object on this plane must be similar to the object,
whether its linear dimensions be altered or not ; when the image is not similar
to the object, it is said to be distorted.

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 273

An image free from these three defects is said to be perfect.

In Fig. 1, p. 285, let A^a^ represent a plane object perpendicular to the
axis of an instrument represented by I., then if the instrument is perfect, as
regards an object at that distance, an image A^a^ will be formed by the
emergent rays, which will have the following properties :

I. Every ray, which passes through a point a, of the object, will pass
through the corresponding point a2 of the image.

II. Every point of the image will lie in a plane perpendicular to the axis.

III. The figure A.fi.fi, will be similar and similarly situated to the figure

Now let us assume that the instrument is also perfect as regards an object

in the plane P-,&A perpendicular to the axis through Bu and that the image

of such an object is in the plane -Ba6jj83 and similar to the object, and we
shall be able to prove the following proposition :

PROP. I. If an instrument give a perfect image of a plane object at two
different distances from the instrument, all incident rays having a common focus
will have a common focus after emergence.

Let P, be the focus of incident rays. Let P,oA be any incident ray.
Then, since every ray which passes through a, passes through aa, its image after
emergence, and since every ray which passes through 6, passes through b,, the
direction of the ray P,a,&, after emergence must be ajbt.

Similarly, since a, and /?, are the images of a, and /?,, if P,a,j8, be any
other ray, its direction after emergence will be cg33.

Join a,a,, 6,/J,, c^a,, b^3; then, since the parallel planes Aja,^ and Bf>^
are cut by the plane of the two rays through P,, the intersections e^o, and
frjS, are parallel.

Also, their images, being similarly situated, are parallel to them, therefore
a,o, is parallel to fc-jS,, and the lines aab, and a^ are in the same plane, and
therefore either meet in a point P3 or are parallel.

Now take a third ray through P,, not in the plane of the two former.
After emergence it must either cut both, or be parallel to them. If it cuts
both it must pass through the point P2, and then every other ray must pass
through P,, for no line can intersect three lines, not in one plane, without
passing through their point of intersection. If not, then all the emergent rays

VOL. I. 35

274 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

are parallel, which is a particular case of a perfect pencil. So that for every
position of the focus of incident rays, the emergent pencil is free from astig-
matism.

PROP. II. In an instrument, perfect at two different distances, the image
of any plane object perpendicular to the axis will be free from the defects of
curvature and distortion.

Through the point Pl of the object draw any line P& in the plane of
the object, and through P,Q, draw a plane cutting the planes Alt -B, in the lines
0,0,, 6j8,. These lines will be parallel to P,<?, and to each other, wherefore
also their images, 0,0,, &^8,, will be parallel to PlQl and to each other, and
therefore in one plane.

Now suppose another plane drawn through P,<?, cutting the planes Al and
BI in two other lines parallel to PjQ^ These will have parallel images in the
planes A, and Bt, and the intersection of the planes passing through the two
pairs of images will define the line PtQt which will be parallel to them, and
therefore to P&, and will be the image of P,(?,. Therefore PtQt, the image
of PjQj is parallel to it, and therefore in a plane perpendicular to the axis.
Now if all corresponding lines in any two figures be parallel, however the lines
be drawn, the figures are similar, and similarly situated.

From these two propositions it follows that an instrument giving a perfect
image at two different distances will give a perfect image at all distances. We
have now only to determine the simplest method of finding the position and
magnitude of the image, remembering that wherever two rays of a pencil inter-
sect, all other rays of the pencil must meet, and that all parts of a plane
object have their images in the same plane, and equally magnified or diminished.

PROP. III. A ray is incident on a perfect instrument parallel to the axis,
to find its direction after emergence.

Let a,6, (fig. 2) be the incident ray, -*•!,«, one of the planes at which an
object has been ascertained to have a perfect image. Afi^ that image, similar
to Ala1 but in magnitude such that Ajit = xAlal.

Similarly let Btbt be the image of Bfa, and let B^ — yB^. Also let
-4,-B^c, and AtBi = ci.

Then since a, and bt are the images of a, and &„ the line F0Jbt will be
the direction of the ray after emergence, cutting the axis in Ft, (unless x = y,

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 275

when 0,6, becomes parallel to the axis). The point F, may be found, by
remembering that Ala1 = B1b1, A1at = xA1al, B})1 = yB}>l. We find —

y-x

Let <7a be the point at which the emergent ray is at the same distance
from the axis as the incident ray, draw gfr^ perpendicular to the axis, then
we have

y-x

Similarly, if a^lFl be a ray, which, after emergence, becomes parallel to
the axis ; and g^Cr, a line perpendicular to the axis, equal to the distance of
the parallel emergent ray, then ,

x-y

Definitions.

I. The point Flt the focus of incident rays when the emergent rays are
parallel to the axis, is called the first principal focus of the instrument.

II. The plane G^l at which incident rays through Fl are at the same
distance from the axis as they are after emergence, is called the first princi-
pal plane of the instrument. F& is called the first focal length.

III. The point Fz, the focus of emergent rays when the incident rays
are parallel, is called the second principal focus.

IV. The plane &&, at which the emergent rays are at the same distance
from the axis, as before incidence, is called the second principal plane, and
Ffrt is called the second focal length.

When x = y, the ray is parallel to the axis, both at incidence and emerg-
ence, and there are no such points as F and G. The instrument is then
called a telescope. x(=y) is called the linear magnifying power and is denoted

ft

by I, and the ratio • ' is denoted by n, and may be called the elongation.

ci

In the more general case, in which x and y are different, the principal
foci and principal planes afford the readiest means of finding the position of
images.

35—2

276 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

PROP. IV. Given the principal foci and principal planes of an instrument,
to find the relations of the foci of the incident and emergent pencils.

Let Flt Ft (fig. 3) be the principal foci, (?,, (?, the principal planes, Qt
the focus of incident light, <?,P, perpendicular to the axis.

Through Ql draw the ray Q&F^ Since this ray passes through Fl it
emerges parallel to the axis, and at a distance from it equal to G&. Its
direction after emergence is therefore Qtgt where Gtgt = G^. Through Ql draw
P,y, parallel to the axis. The corresponding emergent ray will pass through
Ftt and will cut the second principal plane at a distance Gtyt=Gly1, so that
Fty, is the direction of this ray after emergence.

Since both rays pass through the focus of the emergent pencil, Qt, the
point of intersection, is that focus. Draw Qft perpendicular to the axis.
Then P& = Glyl = Gtyt, and <?,£, = £,#, = P,&. By similar triangles FJP& and

And by similar triangles F,P3Qt and FtGtyt

P&( = G^ • **& ::GJP,: F,P,.
We may put these relations into the concise form

PA _ P.Q.

and the values of FJPt and PtQt are

, p Q _

/' A' *™~ P f '>•

These expressions give the distance of the image from Ft measured along the
axis, and also the perpendicular distance from the axis, so that they serve to
determine completely the position of the image of any point, when the princi-
pal foci and principal planes are known.

PROP. V. To find the focus of emergent rays, when the instrument is a
telescope.

Let Ql (fig. 4) be the focus of incident rays, and let <2,a,&, be a ray
parallel to the axis ; then, since the instrument is telescopic, the emergent
ray Qtatbt will be parallel to the axis, and QtPt = l. (?,P,.

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 277

Let Q1a1J51 be a ray through Blt the emergent ray will be Q^a^B^, and
= l.A

Now

> A T>

so that ~n^5 — ~A~i^ — n> a constant ratio.

"

Cor. If a point C be taken on the axis of the instrument so that

then

Def. The point C is called the centre of the telescope.

It appears, therefore, that the image of an object in a telescope has its
dimensions perpendicular to the axis equal to I times the corresponding dimen-
sions of the object, and the distance of any part from the plane through C
equal to n times the distance of the corresponding part of the object. Of
course all longitudinal distances among objects must be multiplied by n to
obtain those of their images, and the tangent of the angular magnitude of an

object as seen from a given point in the axis must be multiplied by to

it/

obtain that of the image of the object as seen from the image of the given
point. The quantity - is therefore called the angular magnifying power, and

71

is denoted by m.

PROP. VI. To find the principal foci and principal planes of a combina-
tion of two instruments having a common axis.

Let /, /' (fig. 5) be the two instruments, G^F^ the principal foci and
planes of the first, GiF^F^G, those of the second, r^^r, those of the com-
bination. Let the ray g^fi^g* pass through both instruments, and let it be
parallel to the axis before entering the first instrument. It will therefore pass
through Ft the second principal focus of the first instrument, and through gt
so that

On emergence from the second instrument it will pass through <j>3 the
focus conjugate to F,, and through gt' in the second principal plane, so that

278 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

tr/gr,' = 6,'gr,'. ^ is by definition the second principal focus of the combination
of instruments, and if r,y, be the second principal plane, then r,yJ=(rIgr1.

We have now to find the positions of <£, and F,.
By Prop. IV., we have

Or, the distance of the principal focus of the combination, from that of the
second instrument, is equal to the product of the focal lengths of the second
instrument, divided by the distance of the second principal focus of the first
instrument from the first of the second. From this we get

Now, by the pairs of similar triangles tf»Gt'gt', <£ray, and FjGrfgf, FtG.gv

IVy, %, _ F&

Multiplying the two sides of the former equation respectively by the first and
last of these equal quantities, we get

r j _ j, •__»*

Or, the second focal distance of a combination is the product of the second
local lengths of its two components, divided by the distance of their consecutive
principal foci.

If we call the focal distances of the first instrument y, and ft, those of
the second /,' and /,', and those of the combination /„ /„ and put
then the positions of the principal foci are found from the values

d ' •'
and the focal lengths of the combination from

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 279

When d = 0, all these values become infinite, and the compound instrument
becomes a telescope.

PROP. VII. To find the linear magnifying power, the elongation, and the
centre of the instrument, when the combination becomes a telescope.

Here (fig. 6) the second principal focus of the first instrument coincides at F
with the first of the second. (In the figure, the focal distances of both instru-
ments are taken in the opposite direction from that formerly assumed. They are
therefore to be regarded as negative.)

In the first place, F,' is conjugate to Flt for a pencil whose focus before
incidence is Fl will be parallel to the axis between the instruments, and will
converge to F{ after emergence.

Also if G& be an object in the first principal plane, G.^ will be its first
image, equal to itself, and if Hh be its final image

,TT_ FG' . G,'Ft'

f~ >
/,

Now the linear magnifying power is - , , and the elongation is -

because Ft' and H are the images of Fl and G^ respectively ; therefore

/=-, and n=r.

J* A/2

The angular magnifying power = m = - = — 4^ .

The centre of the telescope is at the point C, such that

When n becomes 1 the telescope has no centre. The effect of the instrument
is then simply to alter the position of an object by a certain distance measured
along the axis, as in the case of refraction through a plate of glass bounded by
parallel planes. In certain cases this constant distance itself disappears, as in
the case of a combination of three convex lenses of which the focal lengths are

280 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

4, 1, 4 and the distances 4 and 4. This combination simply inverts every object
without altering its magnitude or distance along the axis.

The preceding theory of perfect instruments is quite independent of the
mode in which the course of the rays is changed within the instrument, as
we are supposed to know only that the path of every ray is straight before
it enters, and after it emerges from the instrument. We have now to con-
sider, how far these results can be applied to actual instruments, in which
the course of the rays is changed by reflexion or refraction. We know that
such instruments may be made so as to fulfil approximately the conditions of
a perfect instrument, but that absolute perfection has not yet been obtained.
Let us inquire whether any additional general law of optical instruments can
be deduced from the laws of reflexion and refraction, and whether the imper-
fection of instruments is necessary or removeable.

The following theorem is a necessary consequence of the known laws of
reflexion and refraction, whatever theory we adopt.

If we multiply the length of the parts of a ray which are in different
media by the indices of refraction of those media, and call the sum of these
products the reduced path of the ray, then :

I. The extremities of all rays from a given origin, which have the same
reduced path, lie in a surface normal to those rays.

II. When a pencil of rays is brought to a focus, the reduced path from
the origin to the focus is the same for every ray of the pencil.

In the undulatory theory, the " reduced path " of a ray is the distance
through which light would travel in space, during the time which the ray
takes to traverse the various media, and the surface of equal "reduced paths"
is the wave-surface. In extraordinary refraction the wave-surface is not always
normal to the ray, but the other parts of the proposition are true in this and all
other cases.

From this general theorem in optics we may deduce the following propo-
sitions, true for all instruments depending on refraction and reflexion.

- PROP. VIII. In any optical instrument depending on refraction or reflex-
ion, if 0,0,, &,/?, (fig. 7) be two objects and 0,0,, &j/3, their images, AlBl the
distance of the objects, AtBt that of the images, /x, the index of refraction of

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 281

the medium in which the objects are, /A, that of the medium in which the
images are, then

x b^ _ a.,0, x

'

approximately, when the objects are small.

Since a, is the image of alt the reduced path of the ray a&a* will be
equal to that of a^a,, and the reduced paths of the rays o^o, and aj^a., will
be equal.

Also because b^ and bfy are conjugate foci, the reduced paths of the
rays 6,0,6, and b^b^, and of /JjOj/J, and &o.j8, will be equal. So that the
reduced paths

6,0, + o,62 = 6,a, + a.h,

. •. a,6, + ajS, + 0,6, + a^ = aj8i + Oj&j + aa&2 + a.j/8,,

these beuig still the reduced patfis of the rays, that is, the length of each
ray multiplied by the index of refraction of the medium.

If the figure is symmetrical about the axis, we may write the equation

/*i (« A - «A) = /*» («j/8i ~ 0*6,),
where aj8,, &c. are now the actual lengths of the rays so named.

Now a& = A& + i (a^ + 6g8,)f,

so that o^— ^7~«Ax

x

and

, , , a,o, x 6.J/8.,

Simikrly ^3 (a A - « A) = /

,. a,a, x

So that the equation u,-

VOL, I. 36

282 ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS.

is true accurately, and since when the objects are small, the denominators are
nearly 2.4,5, and 2AtBt, the proposition is proved approximately true.

Using the expressions of Pi-op. III., this equation becomes

1 xu
*- = *-/•

ci C

are

Now by Pn^ IIL, when x and y are different, the focal lengths /, and f,

f- J_
' J'~C'-'

x-y y-x

therefore ^ = -^ = - by the present theorem.

f, ct ^

So that in any instrument, not a telescope, the focal lengths are directly as
the indices of refraction of the media to which they belong. If, as in most
cases, these media are the same, then the two focal distances are equal.

When x = y, the instrument becomes a telescope, and we have, by Prop. V.,

s*

I = x, and » = - ; and therefore by this theorem

/"*» »

We may find ? experimentally by measuring the actual diameter of the
image of a known near object, such as the aperture of the object glass. If O be
the diameter of the aperture and o that of the circle of light at the eye-hole
(which is its image), then

From this we find the elongation and the angular magnifying power

ii u, 1

n = — r, and m = - T .
/*, frl

When /*, = /*,, as in ordinary cases, m = y= — , which is Gauss' rule for deter-

t 0

mining the magnifying power of a telescope.

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 283

PROP. IX. It is impossible, by means of any combination of reflexions
and refractions, to produce a perfect image of an object at two different distances,
unless the instrument be a telescope, and

l = n = ^-1, m=l.
P*

It appears from the investigation of Prop. VIII. that the results there
obtained, if true when the objects are very small, will be incorrect when the
objects are large, unless

and it is easy to prove that this cannot be, unless all the lines in the one figure
are proportional to the corresponding lines in the other. /

In this way we might show that we cannot in general have an astigmatic,
plane, undistorted image of a plane object. But we can prove that we cannot
get perfectly focussed images of an object in two positions, even at the expense
of curvature and distortion.

We shall first prove that if two objects have perfect images, the reduced
path of the ray joining any given points of the two objects is equal to that
of the ray joining the corresponding points of the images.

Let a2 (fig. 8) be the perfect image of at and & of &. Let

Draw a,lDl parallel to the axis to meet the plane JBlf and aj)3 to the plane
of Bv

Since everything is symmetrical about the axis of the instrument we shall
have the angles DlB1^1 = DtBj3, = 0, then in either figure, omitting the suffixes,

6.

It has been shown in Prop. VIII. that the difference of the reduced paths
of the rays «,&„ a^ in the object must be equal to the difference of the reduced
paths of a262, a^ in the image. Therefore, since we may assume any value for 6

Pi •/(«." + 61* + GI - 2«A cos 0) - p., J(a* + 622 + c2" - 2a262 cos 6}

36—2

284 ON THE CKNKUAL LAWS OF OPTICAL INSTRUMENTS.

is constant for all values of 6. This can be only when

fh x/(«,' + V + c,') = fr s/K + 6,' + c,'),
and

which shows that the constant must vanish, and that the lengths of lines
joining corresponding points of the objects and of the images must be inversely
as the indices of refraction before incidence and after emergence.

Next let ABC, DEF (fig. 9) represent three points in the one object

and three points in the other object, the figure being drawn to a scale so that

all the lines in the figure are the actual lines multiplied by /*,. The lines of

the figure represent the reduced paths of the rays between the corresponding
points of the objects.

Now it may be shown that the form of this figure cannot be altered with-

out altering the length of one or more of the nine lines joining the points ABC

to DEF. Therefore since the reduced paths of the rays in the image are equal

to those in the object, the figure must represent the image on a scale of fj.
to 1, and therefore the instrument must magnify every part of the object alike

and elongate the distances parallel ,to the axis in the same proportion. It is
therefore a telescope, and m=l.

If fr = fa, the image is exactly equal to the object, which is the case in
reflexion in a plane mirror, which we know to be a perfect instrument for all
distances.

The only case in which by refraction at a single surface we can get a
perfect image of more than one point of the object, is when the refracting
surface is a sphere, radius r, index \L, and when the two objects are spherical

7*

surfaces, concentric with the sphere, their radii being - , and r ; and the two
images also concentric spheres, radii p.r, and r.

In this latter case the image is perfect, only at these particular distances
and not generally.

I am not aware of any other case in which a perfect image of an object
can be formed, the rays being straight before they enter, and after they emerge
from the instrument. The only case in which perfect astigmatism for all pencils
has hitherto been proved to exist, was suggested to me by the consideration

\

VOL. I. PLATE IV.

Ob

To f<ir* pnfff 284

Cambridge University Press

ON THE GENERAL LAWS OF OPTICAL INSTRUMENTS. 285

of the structure of the crystalline lens in fish, and was published in one of
the problem-papers of the Cambridge and Dublin Mathematical Journal. My
own method of treating that problem is to be found in that Journal, for
February, 1854. The case is that of a medium whose index of refraction varies
with the distance from a centre, so that if //,„ be its value at the centre, a
a given line, and r the distance of any point where the index is /A, then

a2

The path of every ray within this medium is a circle in a plane passing through
the centre of the medium.

Every ray from a point in the medium, distant &/ from the centre, will

«2
converge to a point on the opposite side of the centre and distant from it j- .

It will be observed that both the object and the image are included in
the variable medium, otherwise the images would not be perfect. This case
therefore forms no exception to the result of Prop. IX., in which the object and
image are supposed to be outside the instrument.

Aberdeen, 12th Jan., 185&

[From the Proceedings of the Royal Society of Edinburgh, Vol. IV.]

XVIII. On Tfieories of the Constitution of Saturn's Rings.

THE planet Saturn is surrounded by several concentric flattened rings, which
appear to be quite free from any connection with each other, or with the planet,
except that due to gravitation.

The exterior diameter of the whole system of rings is estimated at about
176,000 miles, the breadth from outer to inner edge of the entire system,
36,000 miles, and the thickness not more than 100 miles.

It is evident that a* system of this kind, so broad and so thin, must
depend for its stability upon the dynamical equilibrium between the motions of
each part of the system, and the attractions which act on it, and that the
cohesion of the parts of so large a body can have no effect whatever on its
motions, though it were made of the most rigid material known on earth. It
is therefore necessary, in order to satisfy the demands of physical astronomy,
to explain how a material system, presenting the appearance of Saturn's Rings,
can be maintained in permanent motion consistently with the laws of gravitation.
The principal hypotheses which present themselves are these —

I. The rings are solid bodies, regular or irregular.
II. The rings are fluid bodies, liquid or gaseous.
III. The rings are composed of loose materials.

The results of mathematical investigation applied to the first case are, —
1st. That a uniform ring cannot have a permanent motion.

2nd. That it ia possible, by loading one side of the ring, to produce
stability of motion, but that this loading must be very great compared with
the whole mass of the rest of the ring, being as 82 to 18.

ON THEORIES Cflf THE CONSTITUTION OF SATURN'S RINGS. 287

3rd. That this- loading must not only be very great, but very nicely
adjusted; because, if it were less than '81, or more than '83 of the whole,
the motion would be unstable.

The mode in which such a system would be destroyed would be by the
collision between the planet and the inside of the ring.

And it is evident that as no loading so enormous in comparison with the
ring actually exists, we are forced to consider the rings as fluid, or at least
not solid ; and we find that, in the case of a fluid ring, waves would be gene-
rated, which would break it up into portions, the number of which would
depend on the mass of Saturn directly, and on that of the ring inversely.

It appears, therefore, that the only constitution possible for such a ring is
a series of disconnected masses, which may be fluid or s61id, and need not be
equal. The complicated internal motions of such a ring have been investigated,
and found to consist of four series of waves, which, when combined together,
will reproduce any form of original disturbance with all its consequences. The
motion of one of these waves was exhibited to the Society by means of a small
mechanical model made by Ramage of Aberdeen.

This theory of the rings, being indicated by the mechanical theory as the
only one consistent with permanent motion, is further confirmed by recent obser-
vations on the inner obscure ring of Saturn. The limb of the planet is seen
through the substance of this ring, not refracted, as it would be through a
gas or fluid, but in its true position, as would be the case if the light passed
through interstices between the separate particles composing the ring.

As the whole investigations are shortly to be published in a separate form,
the mathematical methods employed were not laid before the Society.

XIX. On the Stability of the motion of Saturn's Rings.

[An Essay, which obtained the Adams Prize for the year 1856, in the University

of Cambridge.]

ADVERTISEMENT.

THE Subject of the Prize was announced in the following terms : —

The University having accepted a fund, raised by several members of St John's College,
for the purpose of founding a Prize to be called the ADAMS PRIZE, for the best Essay
on some subject of Pure Mathematics, Astronomy, or other branch of Natural Philosophy,
the Prize to be given once in two years, and to be open to tfte competition of all persons
who have at any time been admitted to a degree in tliis University : —

The Examiners give Notice, that the following is the subject for the Prize to be adjudged
in 1857:—

The Motions of Saturn's Rings.

*#* The problem may be treated on the supposition that the system of Rings is exactly or
very approximately concentric with Saturn and symmetrically disposed about the plane of his Equator,
and different hypotheses may be made respecting the physical constitution of the Rings. It may
be supposed (1) that they are rigid: (2) that they are fluid, or in part aeriform: (3) that they
consist of masses of matter not mutually coherent The question will be considered to be answered
by ascertaining on these hypotheses severally, whether the conditions of mechanical stability are
satisfied by the mutual attractions and motions of the Planet and the Rings.

It is desirable that an attempt should also be made to determine on which of the above
hypotheses the appearances both of the bright Rings and the recently discovered dark Ring may
be most satisfactorily explained; and to indicate any causes to which a change of form, such as
is supposed from a comparison of modern with the earlier observations to have taken place, may
be attributed.

E. GUEST, Vice-chancellor.

J. CHALLIS.

8. PARKINSON.

W. THOMSON.
.VarcA 23, 1855.

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 289

CONTENTS.

Nature of the Problem 290

Laplace's investigation of the Equilibrium of a Ring, and its minimum density 292

Hit proof that the plane of the Rings will follow that of Saturn's Equator — that a solid uniform Ring is

necessarily unstable 293

Further investigation required — Theory of an Irregular Solid Ring leads to the result that to ensure stability

the irregularity must be tnormous 294

Theory of a fluid or discontinuous Ring resolves itself into the investigation of a series of waves . . 295

PART I.

ON THE MOTION OF A RIGID BODY OF ANY FORM ABOUT
A SPHERE.

Equations of Motion 296

PROBLEM I. To find the conditions under which a uniform motion of the Ring is possible . . . 298

PROBLEM II. To find the equations of the motion when slightly disturbed 299

PROBLEM III. To reduce the three simultaneous equations of motion to the form of a single linear equation 300
PROBLEM IV. To determine whether the motion of the Ring is stable or unstable, by means of the relations

of the coefficients A, B, C 301

PROBLEM V. To find the centre of gravity, the radius of gyration, and the variations of the potential near

the centre of a circular ring of small but variable section 302

PROBLEM VI. To determine the condition of stability of the motion in terms of the coefficients f, g, h, which

indicate the distribution of mass in the ring 30C

RB60LT8. 1st, a uniform ring is unstable. 2nd, a ring varying in section according to the law of sines is
unstable. 3rd, a uniform ring loaded with a heavy particle may be stable, provided the mass of the
particle be between '815865 and '8279 of the whole. Case in which the ring is to the particle at 18
to 82 307

PART II.

ON THE MOTION OF A RING, THE PARTS OF WHICH ARE NOT
RIGIDLY CONNECTED.

1. General Statement of the Problem, and limitation to a nearly uniform ring 310

2. Notation 311

3. Expansion of a function in terms of sines and cosines of multiples of the variable .... 311

4. Magnitude and direction of attraction between two elements of a disturbed Ring 312

5. Resultant attractions on any one of a ring of equal satellites disturbed in any way . . . . 313
Note. Calculated values of these attractions in particular cases 314

6. Equations of motion of a satellite of the Ring, and biquadratic equation to determine the wave-velocity 315

7. A ring of satellites may always be rendered stable by increasing the mass of the central body . . 317

8. Relation between the number and mass of satellites and the mass of the central body necessary to

ensure stability. S>'4352M2R 318

9. Solution of the biquadratic equation when the mass of the Ring is small ; and complete expressions

for the motion of each satellite 319

10. Each satellite moves (relatively to the ring) in an ellipse 321

VOL. I. 37

290 ON THE STABILITY OP THE MOTION OP SATURN'S RINGS.

11. Each satellite motet absolutely in tpaet in a curve which it nearly an ellipse for the large valuet

of n, and a spiral of many nearly circular coil* when n it mall 321

13. Tke form of the ring at a given inttant it a tenet of undulations 322

13. These undulations travel round the ring with velocity - - relative to the ring, and » absolutely 323

14. General Solution of the Problem — Given the porition and motion of every satellite at any one time,
to calculate the petition and motion of any tateUite at any other time, provided that the condition

of ttability it fulfilled 323

15. Calculation of the efect of a periodic external disturbing force 326

16. Treatment of ditturbing foroet in general 328

17. Theory of free waww and forced waves 329

18. Motion of the ring when the condition* of ttability are not fulfilled. Two different ways in which.

the ring may be broken up 330

19. Motion of a ring of unequal tatettitet 335

20. Motion of a ring composed of a cloud of scattered particle* 336

21. Calculation of the forces arising from the displacements of such a system 337

22. Application to the cote of a ring of this kind. The mean density must be excessively small, which

it inconsistent with its moving as a whole 338

23. On the forces arising from inequalities in a thin stratum of gravitating incompressible fluid of
indefinite extent 338

24. Application to the cote of a flattened fluid ring, moving with uniform angular velocity. Such a

ring will be broken up into portions which may continue to revolve a* a ring of tatettitet . . . 344

ON THE MUTUAL PERTURBATIONS OF TWO RINGS.

25. Application of the general theory of free and forced waves 345

26. To determine the attractions between the rings 346

27. To form the equation* of motion 348

28. Method of determining the reaction of the forced u-art on the free wave which produced it . . 349

29. C<i*et in which the perturbations increase indefinitely 351

30. Application to the theory of an indefinite number of concentric rings 352

31. On the efect of long-continued disturbances on a system of rings - . . . 352

32. On the efect of collision* among the parts of a revolving system 354

33. On the effect of internal friction in a fluid ring 354

Recapitulation of the Theory of the Motion of a Rigid King. Reasons for rejecting the hypothesis of

rigidity 356

Recapitulation of the Theory of a Ring of Equal Satellites 360

Description of a working model shewing the motions of such a system 363

Theory of Rings of various constitution* 367

Mutual action of Two Rings 370

Catt of many concentric Rings, <k<:. 371

General Conclusions 372

APPENDIX. Extract of a letter from Professor W. Thomson, of Glasgow, giving a solution of t/ie Pro-
blem of a Rigid Ring 374

THERE are some questions in Astronomy, to which we are attracted rather
on account of their peculiarity, as the possible illustration of some unknown
principle, than from any direct advantage which their solution would afford to

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 291

mankind. The theory of the Moon's inequalities, though in its first stages it
presents theorems interesting to all students of mechanics, has been pursued into
such intricacies of calculation as can be followed up only by those who make
the improvement of the Lunar Tables the object of their lives. The value of
the labours of these men is recognised by all who are aware of the importance
of such tables in Practical Astronomy and Navigation. The methods by which
the results are obtained are admitted to be sound, and we leave to professional
astronomers the labour and the merit of developing them.

The questions which are suggested by the appearance of Saturn's Kings
cannot, in the present state of Astronomy, call forth so great an amount of
labour among mathematicians. I am not aware that any practical use has been
made of Saturn's Rings, either in Astronomy or in Navigation. They are too
distant, and too insignificant in mass, to produce any appreciable effect on the
motion of other parts of the Solar system ; and for this very reason it is diffi-
cult to determine those elements of their motion which we obtain so accurately
in the case of bodies of greater mechanical importance.

But when we contemplate the Rings from a purely scientific point of view,
they become the most remarkable bodies in the heavens, except, perhaps, those
still less useful bodies — the spiral nebulae. When we have actually seen that
great arch swung over the equator of the planet without any visible connexion,
we cannot bring our minds to rest. We cannot simply admit that such is the
case, and describe it as one of the observed facts in nature, not admitting or
requiring explanation. We must either explain its motion on the principles of
mechanics, or admit that, in the Saturnian realms, there can be motion regu-
lated by laws which we are unable to explain.

The arrangement of the rings is represented in the figure (l) on a scale
of one inch to a hundred thousand miles. S is a section of Saturn through
his equator, A, B and C are the three rings. A and B have been known for
200 years. They were mistaken by Galileo for protuberances on the planet itself,
or perhaps satellites. Huyghens discovered that what he saw was a thin flat
ring not touching the planet, and Ball discovered the division between A and B.
Other divisions have been observed splitting these again into concentric rings,
but these have not continued visible, the only well-established division being one
in the middle of A. The third ring C was first detected by Mr Bond, at
Cambridge U.S. on November 15, 1850; Mr Dawes, not aware of Mr Bond's
discovery, observed it on November 29th, and Mr Lassel a few days later. It

37—2

292 ON THE STABILITY OP THE MOTION OF SATURN'S RINGS.

gives little light compared with the other rings, and is seen where it crosses
the planet as an obscure belt, but it is so transparent that the limb of the
planet is visible through it, and this without distortion, shewing that the rays
of light have not passed through a transparent substance, but between the
scattered particles of a discontinuous stream.

It is difficult to estimate the thickness of the system ; according to the
best estimates it is not more than 100 miles, the diameter of A being 176,418
miles ; so that on the scale of our figure the thickness would be one thousandth
of an inch.

Such is the scale on which this magnificent system of concentric rings is
constructed ; we have next to account for their continued existence, and to
reconcile it with the known laws of motion and gravitation, so that by rejecting
every hypothesis which leads to conclusions at variance with the facts, we may
learn more of the nature of these distant bodies than the telescope can yet
ascertain. We must account for the rings remaining suspended above the planet,
concentric with Saturn and in his equatoreal plane ; for the flattened figure of the
section of each ring, for the transparency of the inner ring, and for the gradual
approach of the inner edge of the ring to the body of Saturn as deduced
from all the recorded observations by M. Otto Struvd (Sur les dimensions des
Anneaux de Saturne — Recueil de Me*moires Astronomiques, Poulkowa, 15 Nov.
1851). For an account of the general appearance of the rings as seen from the
planet, see Lardner on the Uranography of Saturn, Mem. of the Astronomical
Society, 1853. See also the article "Saturn" in Nichol's Cyclopaedia. of the
Physical Sciences.

Our curiosity with respect to these questions is rather stimulated than
appeased by the investigations of Laplace. That great mathematician, though
occupied with many questions which more imperiously demanded his attention,
has devoted several chapters in various parts of his great work, to points con-
nected with the Saturnian System.

He has investigated the law of attraction of a ring of small section on a
point very near it (Mec. Cel. Liv. in. Chap, vi.), and from this he deduces the
equation from which the ratio of the breadth to the thickness of each ring is
to be found,

^?£_ X(X-l)

~~

where R is the radius of Saturn, and p his density; a the radius of the ring,

ON THE STABILITY OF THE MOTION OP SATURN'S RINGS. 293

and p its density ; and X the ratio of the breadth of the ring to its thick-
ness. The equation for determining X when e is given has one negative root
which must be rejected, and two roots which are positive while e<0'0543, and
impossible when e has a greater value. At the critical value of e, X = 2'594
nearly.

The fact that X is impossible when e is above this value, shews that the
ring cannot hold together if the ratio of the density of the planet to that of
the ring exceeds a certain value. This value is estimated by Laplace at 1*3,
assuming a = 2R.

We may easily follow the physical interpretation of this result, if we observe
that the forces which act on the ring may be reduced to —

(1) The attraction of Saturn, varying inversely as the square of the dis-
tance from his centre.

(2) The centrifugal force of the particles of the ring, acting outwards, and
varying directly as the distance from Saturn's polar axis.

(3) The attraction of the ring itself, depending on its form and density,
and directed, roughly speaking, towards the centre of its section.

The first of these forces must balance the second somewhere near the mean
distance of the ring. Beyond this distance their resultant will be outwards,
within this distance it will act inwards.

If the attraction of the ring itself is not sufficient to balance these residual
forces, the outer and inner portions of the ring will tend to separate, and the
ring will be split up; and it appears from Laplace's result that this will be
the case if the density of the ring is less than ±$ of that of the planet.

This condition applies to all rings whether broad or narrow, of which the
parts are separable, and of which the outer and inner parts revolve with the
same angular velocity.

Laplace has also shewn (Liv. v. Chap, in.), that on account of the oblate-
ness of the figure of Saturn, the planes of the rings will follow that of Saturn's
equator through every change of its position due to the disturbing action of
other heavenly bodies.

Besides this, he proves most distinctly (Liv. in. Chap, vi.), that a solid uni-
form ring cannot possibly revolve about a central body in a permanent manner,
for the slightest displacement of the centre of the ring from the centre of the
planet would originate a motion which would never be checked, and would

294 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

inevitably precipitate the ring upon the planet, not necessarily by breaking the
ring, but by the inside of the ring falling on the equator of the planet.

He therefore infers that the rings are irregular solids, whose centres of
gravity do not coincide with their centres of figure. We may draw the con-
clusion more formally as follows, " If the rings were solid and uniform, their
motion would be unstable, and they would be destroyed. But they are not
destroyed, and their motion is stable ; therefore they are either not uniform or
not solid."

I have not discovered* either in the works of Laplace or in those of more
recent mathematicians, any investigation of the motion of a ring either not uni-
form or not solid. So that in the present state of mechanical science, we do
not know whether an irregular solid ring, or a fluid or disconnected ring, can
revolve permanently about a central body; and the Saturnian system still re-
mains an unregarded witness in heaven to some necessary, but as yet unknown,
development of the laws of the universe.

We know, since it has been demonstrated by Laplace, that a uniform solid
ring cannot revolve permanently about a planet. We propose in this Essay to
determine the amount and nature of the irregularity which would be required
to make a permanent rotation possible. We shall find that the stability of the
motion of the ring would be ensured by loading the ring at one point with a

* Since this was written, Prof. Challis has pointed out to me three important papers in Gould's
Astronomical Journal: — Mr G. P. Bond on the Kings of Saturn (May 1851) and Prof. B. Pierce of
Harvard University on the Constitution of Satum't Rings (June 1851), and on the Adams' Prise
Problem for 1856 (Sept. 1855). These American mathematicians have both considered the conditions
of statical equilibrium of a transverse section of a ring, and have come to the conclusion that the
rings, if they move each as a whole, must be very narrow compared with the observed rings, so
that in reality there must be a great number of them, each revolving with its own velocity. They
have also entered on the question of the fluidity of the rings, and Prof. Pierce has made an
investigation as to the permanence of the motion of an irregular solid ring and of a fluid ring.
The paper in which these questions are treated at large has not (so far as I am aware) been
published, and the references to it in Gould's Journal are intended to give rather a popular account
of the results, than an accurate outline of the methods employed. In treating of the attractions of
an irregular ring, he makes admirable use of the theory of potentials, but his published investi-
gation of the motion of such a body contains some oversights which are due perhaps rather to the
imperfections of popular language than to any thing in the mathematical theory. The only part of
the theory of a fluid ring which he has yet given an account of, is that in which he considers
the form of the ring at any instant as an ellipse ; corresponding to the case where » = o>, and
rn=l. As I had only a limited time for reading these papers, and as I could not ascertain the
methods ui«ed in the original investigations, I am unable at present to state how far the results of
this essay agree with or differ from those obtained by Prof. Pierce.

ON THE STABILITY OF THE MOTION OP SATURN'S RINGS. 295

heavy satellite about 4-$- times tlie weight of the ring, but this load, besides
being inconsistent with the observed appearance of the rings, must be far too
artificially adjusted to agree with the natural arrangements observed elsewhere,
for a very small error in excess or defect would render the ring again unstable.

We are therefore constrained to abandon the theory of a solid ring, and
to consider the case of a ring, the parts of which are not rigidly connected,
as in the case of a ring of independent satellites, or a fluid ring.

There is now no danger of the whole ring or any part of it being pre-
cipitated on the body of the planet. Every particle of the ring is now to be
regarded as a satellite of Saturn, disturbed by the attraction of a ring of
satellites at the same mean distance from the planet, each of which however is
subject to slight displacements. The mutual action of the parts of the ring will
be so small compared with the attraction of the planet, 'that no part of the
ring can ever cease to move round Saturn as a satellite.

But the question now before us is altogether different from that relating to
the solid ring. We have now to take account of variations in the form and
arrangement of the parts of the ring, as well as its motion as a whole, and
we have as yet no security that these variations may not accumulate till the
ring entirely loses its original form, and collapses into one or more satellites,
circulating round Saturn. In fact such a result is one of the leading doctrines
of the " nebular theory " of the formation of planetary systems : and we are
familiar with the actual breaking up of fluid rings under the action of "capil-
lary" force, in the beautiful experiments of M. Plateau.

In this essay I have shewn that such a destructive tendency actually exists,
but that by the revolution of the ring it is converted into the condition of
dynamical stability. As the scientific interest of Saturn's Rings depends at
present mainly on this question of their stability, I have considered their motion
rather as an illustration of general principles, than as a subject for elaborate
calculation, and therefore I have confined myself to those parts of the subject
which bear upon the question of the permanence of a given form of motion.

There is a very general and very important problem in Dynamics, the solu-
tion of which would contain all the results of this Essay and a great deal
more. It is this —

"Having found a particular solution of the equations of motion of any
material system, to determine whether a slight disturbance of the motion indi-

296 ON THE STABILITY OF THE MOTION OF SATURN'S KINGS.

cated by the solution would cause a small periodic variation, or a total
derangement of the motion."

The question may be made to depend upon the conditions of a maximum
or a minimum of a function of many variables, but the theory of the testa
for distinguishing maxima from minima by the Calculus of Variations becomes
so intricate when applied to functions of several variables, that I think it doubt-
ful whether the physical or the abstract problem will be first solved.

PART I.

ON THE MOTION OF A RIGID BODY OF ANY FORM ABOUT A SPHERE.

WE confine our attention for the present to the motion in the plane of
reference, as the interest of our problem belongs to the character of this motion,
and not to the librations, if any, from this plane.

Let S (Fig. 2) be the centre of gravity of the sphere, which we may call
Saturn, and R that of the rigid body, which we may call the Ring. Join RS,
and divide it in G so that

SG : GR :: R : S,

\

R and S being the masses of the Ring and Saturn respectively.

Then G will be the centre of gravity of the system, and its position will
be unaffected by any mutual action between the parts of the system. Assume G
as the point to which the motions of the system are to be referred. Draw GA
in a direction fixed in space.

Let AGR = 6, and SR = r,

o D

then GR = ~S+RT> and GS=~S+Rr>

so that the positions of S and R are now determined.

Let BRR be a straight line through R, fixed with respect to the substance
of the ring, and let BRK=<j>.

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 297

This determines the angular position of the ring, so that from the values
of r, 6, and <jt the configuration of the system may be deduced, as far as relates
to the plane of reference.

We have next to determine the forces which act between the ring and
the sphere, and this we shall do by means of the potential function due to
the ring, which we shall call V.

The value of V for any point of space S, depends on its position relatively
to the ring, and it is found from the equation

F /dm

W — £4 \ *-

where dm is an element of the mass of the ring, and rr is the distance of that
element from the given point, and the summation is extended over every element
of mass belonging to the ring. V will then depend entirely upon the position
of the point S relatively to the ring, and may be expressed as a function
of r, the distance of S from R, the centre of gravity of the ring, and <j>, the
angle which the line SR makes with the line RB, fixed in the ring.

A particle P, placed at S, will, by the theory of potentials, experience a

moving force P -j- in the direction which tends to increase r, and P j-r

dr r d(f>

in a tangential direction, tending to increase <f).

Now we know that the attraction of a sphere is the same as that of
a particle of equal mass placed at its centre. The forces acting between the

dV
sphere and the ring are therefore S -j— tending to increase r, and a tangential

1 dV
force S — jr > applied at S tending to increase <f>. In estimating the effect of

1 dV
this latter force on the ring, we must resolve it into a tangential force S - -j-r

dV
acting at R, and a couple S -j-r tending to increase <£.

We are now able to form the equations of motion for the planet and the
ring.

VOL. I. 38

298 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

For the planet

Rr cW R

df

, .

dt ^R di

Rr \ Rr dff*

For the centre of gravity of the ring,

Sr V dff\ S dV

- * '

d (I

J{dt\\

s+R

Sr \ & AM dV

s+R di

For the rotation of the ring about its centre of gravity,

»»<•+«-» a? .............................. <5>-

where A; ia the radius of gyration of the ring about its centre of gravity.

Equation (3) and (4) are necessarily identical with (1) and (2), and shew
that the orbit of the centre of gravity of the ring must be similar to that
of the Planet. Equations (l) and (3) are equations of areas, (2) and (4) are
those of the radius vector.

Equations (3), (4) and (5) may be thus written,

These are the necessary and sufficient data for determining the motion of
the ring, the initial circumstances being given.

PROB. I. To find the conditions under which a uniform motion of the
ring is possible.

By a uniform motion is here meant a motion of uniform rotation, during
which the position of the centre of the Planet with respect to the ring does
not change.

ON THE STABILITY OP THE MOTION OF SATURN'S RINGS. 299

In this case r and <j) are constant, and therefore V and its differential
coefficients are given. Equation (7) becomes,

dV

j~ = Q>
a/-
which shews that the angular velocity is constant, and that

R + SdV

Hence, -5-5 = 0, and therefore by equation (8),

%=« e*

Equations (9) and (10) are the conditions under which the uniform motion
is possible, and if they were exactly fulfilled, the uniform motion would go on
for ever if not disturbed. But it does not follow that if these conditions were
nearly fulfilled, or that if when accurately adjusted, the motion were slightly
disturbed, the motion would go on for ever nearly uniform. The effect of the
disturbance might be either to produce a periodic variation in the elements
of the motion, the amplitude of the variation being small, or to produce a
displacement which would increase indefinitely, and derange the system altogether.
In the one case the motion would be dynamically stable, and in the other it
would be dynamically unstable. The investigation of these displacements while
still very small will form the next subject of inquiry.

PROB. II. To find the equations of the motion when slightly disturbed.
Let r = ra, & = tat and <j> = <j>0 in the case of uniform motion, and let

r=r0 + rlt

when the motion is slightly disturbed, where rlt 0lt and ^, are to be treated
as small quantities of the first order, and their powers and products are to be

dV dV
neglected. We may expand -5— and -j-r by Taylor's Theorem,

dV=dV

dV^dV

38—2

300 ON THE STABILITY OF THE MOTION OP SATURN'S RINGS.

where the values of the differential coefficients on the right-hand side of the
equations are those in which r, stands for r, and fa for fa

and taking account of equations (9) and (10), we may write these equations,

Substituting these values in equations (6), (7), (8), and retaining all small
quantities of the first order while omitting their powers and products, we have
the following system of linear equations in rlt 0lf and fa,

=0 ......... (11),

= 0 ......... (12),

=0 ......... (13).

PROB. III. To reduce the three simultaneous equations of motion to the
form of a single linear equation.

Let us write n instead of the symbol -j- , then arranging the equations in
terms of rlt 6lt and fa, they may be written:

=0 ...... (14),

= 0 ...... (15),

- (SM) r, + (/BfcW) 0, + (BJfn9 -SN}fa =0 ...... (16).

Here we have three equations to determine three quantities r,, 0,, fa; but
it is evident that only a relation can be determined between them, and that
in the process for finding their absolute values, the three quantities will vanish
together, and leave the following relation among the coefficients,

ON THE STABILITY OF THE MOTION OF SATURN'S EINGS. 301

- [ZRrjan + (R + S) M} {ZRrjm} {RJfn1 - SN}
+ {Rtf - Rrf -(R + S) L} {RIM} {(R + S)N]

+ (SM) (flr.V) (R + S) M- (SM) (2Rrta>n) (R + S)N\=0 ...... (17).

+ {2Rrjun + (R + S)M} {Rtfri} {(R + S)M}

- {Rn* - Rw1 -(R + S)} {/2r.V} {RtW - SN}

By multiplying up, and arranging by powers of n and dividing by Rn',

this equation becomes

Ant + Bn*+C=0 ................................ (18),

where

...... (19).

C=R{(R+S)lt- 3Sr0'} <S + (R + S) {(R + S)V + Sr*} (LN-

Here we have a biquadratic equation in n which may be treated as a
quadratic in ?i2, it being remembered that n stands for the operation -5- .

PROB. IV. To determine whether the motion of the ring is stable or
unstable, by means of the relations of the coefficients A, B, C.

The equations to determine the forms of rlt 0lt and fa are all of the form

and if n be one of the four roots of equation (18), then

u = Dent

will be one of the four terms of the solution, and the values of rlt 0lt and
fa will differ only in the values of the coefficient D.

Let us inquire into the nature of the solution in different cases.

(1) If n be positive, this term would indicate a displacement which
must increase indefinitely, so as to destroy the arrangement of the system.

(2) If n be negative, the disturbance which it belongs to would gradually
die away.

(3) If n be a pure impossible quantity, of the form ±a«/— 1, then there
will be a term in the solution of the form D cos (at + a), and this would indi-

cate a periodic variation, whose amplitude is D, and period - - .

a

302 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

(4) If n be of the form b±J-la, the first term being positive and
the second impossible, there will be a term in the solution of the form

which indicates a periodic disturbance, whose amplitude continually increases
till it disarranges the system.

(5) If n be of the form — 6±«/ — la, a negative quantity and an im-
possible one, the corresponding term of the solution is

.Dc"6* cos (at + a),
which indicates a periodic disturbance whose amplitude is constantly diminishing.

It is manifest that the first and fourth cases are inconsistent with the
permanent motion of the system. Now since equation (18) contains only even
powers of n, it must have pairs of equal and opposite roots, so that every
root coming under the second or fifth cases, implies the existence of another
root belonging to the first or fourth. If such a root exists, some disturbance
may occur to produce the kind of derangement corresponding to it, so that
the system is not safe unless roots of the first and fourth kinds are altogether
excluded. This cannot be done without excluding those of the second and fifth
kinds, so that, to insure stability, all the four roots must be of the third kind,
that is, pure impossible quantities.

That this may be the case, both values of n* must be real and negative,
and the conditions of this are —

1st. That A, B, and C should be of the same sign,
2ndly. That & > 1A C.

When these conditions are fulfilled, the disturbances will be periodic and
consistent with stability. When they are not both fulfilled, a small disturbance
may produce total derangement of the system.

PROB. V. To find the centre of gravity, the radius of gyration, and the
variations of the potential near the centre of a circular ring of small but variable
section.

Let a be the radius of the ring, and let 6 be the angle subtended at the
centre between the radius through the centre of gravity and the line through
a given point in the ring. Then if p be the mass of unit of length of the

ON THE STABILITY OP THE MOTION OF SATURN'S KINGS. 303

ring near the given point, fi will be a periodic function of 0, and may there-
fore be expanded by Fourier's theorem in the series,

/u, = — {l + 2/cos0 + f#cos20 + f/ism20+2icos(30 + a) + &c.} (21),

i.TT(.t

where f, g, h, &c. are arbitrary coefficients, and R is the mass of the ring.

(l) The moment of the ring about the diameter perpendicular to the
prime radius is

-f'

Jo

cos OdO = Raf,

therefore the distance of the centre of gravity from the centre of the ring,

r0 = a/

(2) The radius of gyration of the ring about its centre in its own plane
is evidently the radius of the ring =a, but if k be that about the centre of
gravity, we have

(3) The potential at any point is found by dividing the mass of each
element by its distance from the given point, and integrating over the whole
mass.

Let the given point be near the centre of the ring, and let its position be
denned by the co-ordinates r' and »|>, of which r' is small compared with a.

The distance (p) between this point and a point in the ring is
^ = i{l +

r
The other terms contain powers of - - higher than the second.

We have now to determine the value of the integral,

o P

and ha multiplying the terms of (p.) by those of (-) , we need retain only
those which contain constant quantities, for all those which contain sines or

304 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

cosines of multiples of (i|» — 6) will vanish when integrated between the limits.
In this way we find

........... (22).

r

The other terms containing higher powers of — .

In order to express V in terms of r, and <f>lt as we have assumed in the
former investigation, we must put

r'cosi/» = -r1
r'sini/»= -

From which we find (-T-) = — if,

\drj. a'-7'

(24).

These results may be confirmed by the following considerations applicable to
any circular ring, and not involving any expansion or integration. Let of be
the distance of the centre of gravity from the centre of the ring, and let
the ring revolve about its centre with velocity ta. Then the force necessary
to keep the ring in that orbit will be —Rafoi1.

But let S be a mass fixed at the centre of the ring, then if

j_8
a"

every portion of the ring will be separately retained in its orbit by the attrac-
tion of S, so that the whole ring will be retained in its orbit. The resultant
attraction must therefore pass through the centre of gravity, and be

a1 '

therefore ^- = — R • , ,

ar a*

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 305

d? V d* V d"V
The equation _ + _+_ + 4trp = 0

is true for any system of matter attracting according to the law of gravitation.
If we bear in mind that the expression is identical in form with that which
measures the total efflux of fluid from a differential element of volume, where

-r- , -j— , -r- are the rates at which the fluid passes through its sides, we may

easily form the equation for any other case. Now let the position of a point
in space be determined by the co-ordinates r, tf> and z, where z is measured
perpendicularly to the plane of the angle $. Then by choosing the directions
of the axes x, y, z, so as to coincide with those of the radius vector r, the per-
pendicular to it in the plane of <f>, and the normal, we shall have

dx = dr, dy = rd<f>, dz = dz,

dV=dV dV^ldV d_V=dV
dx~ dr ' dy ~ r d<f>' dz ~~ dz '

The quantities of fluid passing through an element of area in each direction are

dv j±j dVl j j dv j±j

-j- rddtdz. -j-r -drdz, — j- rd<f>dr,

dr d<(> r dz

so that the expression for the whole efflux is

/oc\

— I — 7 |. ~r 7 . • laOji

r dr d'r r d<f> dz
which is necessarily equivalent to the former expression.

Now at the centre of the ring -ry- may be found by considering the attrac-

Uw

tion on a point just above the centre at a distance z,

dV

-j- = —It
dz

R

TV = — ; , when z = 0.
dz1 a' '

1 dV R ,

Also we know -=- = — ; , and r = of,

r dr a*

• i d'V I d'V nR

so that in any circular ring - + = 2

an equation satisfied by the former values of L and N.

VOL. i. 39

306 ON THE STABILITY OF THE MOTION OP SATURN'S RINGS.

By referring to the original expression for the variable section of the ring,
it appears that the effect of the coefficient / is to make the ring thicker on
one side and thinner on the other in a uniformly graduated manner. The effect
of g is to thicken the ring at two opposite sides, and diminish its section in
the parts between. The coefficient h indicates an inequality of the same kind,
only not symmetrically disposed about the diameter through the centre of
gravity.

Other terms indicating inequalities recurring three or more times in the
circumference of the ring, have no effect on the values of L, M and N. There is
one remarkable case, however, in which the irregularity consists of a single
heavy particle placed at a point on the circumference of the ring.

Let P be the mass of the particle, and Q that of the uniform ring on
which it is fixed, then R = P + Q,

(27).

PROB. VI. To determine the conditions of stability of the motion in terms
of the coefficients /, g, h, which indicate the distribution of mass in the ring.

The quantities which enter into the differential equation of motion (18)
are R, S, J?, r,, «*, L, M, N. We must observe that S is very large compared
with R, and therefore we neglect R in those terms in which it is added to S,
and we put

*• = a' (I-/').

5?

~

ON THE STABILITY OF THE MOTION OF SATUKN's RINGS. 307

Substituting these values in equation (18) and dividing by Jfa4/", we obtain
(!_/.) w< + (l_|/» + i/^)wV + (|-6/'-^-i7r + 2/25r)^ = 0 ...... (28).

The condition of stability is that this equation shall give both values of n2
negative, and this renders it necessary that all the coefficients should have the
same sign, and that the square of the second should exceed four times the
product of the first and third.

(1) Now if we suppose the ring to be uniform, f, g and h disappear,
and the equation becomes

7i4 + ttV + | = 0 ................................. (29),

which gives impossible values to n1 and indicates the instability of a uniform
ring. /

(2) If we make g and h = 0, we have the case of a ring thicker at one
side than the other, and varying in section according to the simple law of sines.
We must remember, however, that f must be less than \, in order that the
section of the ring at the thinnest part may be real. The equation becomes

The condition that the third term should be positive gives

/'<-375.
The condition that n' should be real gives

71/4-112/' + 32 negative,
which requires/' to be between '37445 and 1*2.

The condition of stability is therefore that /" should lie between

•37445 and "375,

but the construction of the ring on this principle requires that f" should be
less than '25, so that it is impossible to reconcile this form of the ring with
the conditions of stability.

(3) Let us next take the case of a uniform ring, loaded with a heavy
particle at a point of its circumference. We have then g = 3f, h = 0, and the
equation becomes

39—2

308 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

Dividing each term by 1—f, we get

= 0 ............ (32).

The first condition gives /less than '8279.

The second condition gives f greater than "815865.

Let us assume as a particular case between these limits /='82, which
makes the ratio of the mass of the particle to that of the ring as 82 to 18,
then the equation becomes

l-82n4+-8114nV+-9696<a4 = 0 ...................... (33),

which gives J~^ln = ±'5916co or ±'3076w.

These values of n indicate variations of r,, 0lt and $,, which are com-
pounded of two simple periodic inequalities, the period of the one being 1*69
revolutions, and that of the other 3 '2 51 revolutions of the ring. The relations
between the phases and amplitudes of these inequalities must be deduced from
equations (14), (15), (16), in order that the character of the motion may be
completely determined.

Equations (14), (15), (16) may be written as follows:

................ (34),

(35),
}^ = 0 ....... (36).

By eliminating one of the variables between any two of these equations,
we may determine the relation between the two remaining variables. Assuming
one of these to be a periodic function of t of the form A cos vt, and remem-

bering that n stands for the operation -j- , we may find the form of the other.
Thus, eliminating Ql between the first and second equations,

{(3-g)a-lshn}<t>l = Q ............ (37).

ON THE STABILITY OF THE MOTION OP SATUBN's RINGS. 309

7*

Assuming - = A sin vt, and ^ — Q cos (vt — £),

Of

{-v* + £j,<as (5 - g)} A cos vt + hot3 A sin vt +/co3 (3 - g) Qcoa (vt -0) + ^fhv'vQ sin (vt - ft).
Equating vt to 0, and to - , we get the equations

{S-ivo? (5 -g)} A =fo?Q {(3 -g) o> cos ft - %hv sin ft},
- ho? A =fo>*Q {(3 -g)ownft + $hv cos ft},
from which to determine Q and ft.

In all cases in which the mass is disposed symmetrically about the diameter
through the centre of gravity, h = 0 and the equations may be greatly simplified.

Let &1 = P cos (vt — a), then the second equation becomes
(i/3 + frS (3 + g)} A sin vt = 2Pfo>v sin (vt - a),

whence a = 0, P = * + *?(*+9) A ........................ (38).

2jotv

The first equation becomes

4.4 o)v cos i>< - 2Pfv* cos i/« + Q/" (3 -g) o? cos (vf - ft) = 0,

whence /f=0, ^ = 11^ ..................... (39)-

In the numerical example in which a heavy particle was fixed to the cir-
cumference of the ring, we have, when f='82,

* f'5916 Pf3'21 Qf- 1-229
(a~\'3076' A~ \_572' A\- 797'

so that if we put o>£ = 00 = the mean anomaly,

0 0-/8) ..................... (40),

&

6, = 3-21 A cos (-5916 0,-a) + 5725 cos (-3076^-^).... ........ (41),

^= -1-229 ^4 cos (-5916^ -a) -57975 cos ('30760,- ft) ... (42).

These three equations serve to determine r,, 01 and fa when the original
motion Js given. They contain four arbitrary constants A, B, a, ft. Now since

310 ON THE STABILITY OF THE MOTION OF 8ATURN*S RINGS.

the original values r,, 0,, ^,, and also their first differential coefficients with
respect to t, are arbitrary, it would appear that six arbitrary constants ought
to enter into the equation. The reason why they do not is that we assume
r, and 6, as the mean values of r and 6 in the actual motion. These quantities
therefore depend on the original circumstances, and the two additional arbitrary
constants enter into the values of re and 00. In the analytical treatment of the
problem the differential equation in n was originally of the sixth degree with a
solution 7i* = 0, which implies the possibility of terms in the solution of the
form Ct + D.

The existence of such terms depends on the previous equations, and we find
that a term of this form may enter into the value of 0, and that r, may contain
a constant term, but that in both cases these additions will be absorbed into
the values of 00 and r0.

PART II.

ON THE MOTION OF A RING, THE PARTS OF WHICH ARE NOT RIGIDLY CONNECTED.

1. IN the case of the Ring of invariable form, we took advantage of the
principle that the mutual actions of the parts of any system form at all times
u system of forces in equilibrium, and we took no account of the attraction
between one part of the ring and any other part, since no motion could result
from this kind of action. But when we regard the different parts of the ring
as capable of independent motion, we must take account of the attraction on
each portion of the ring as affected by the irregularities of the other parts, and
therefore we must begin by investigating the statical part of the problem in
order to determine the forces that act on any portion of the ring, as depending
on the instantaneous condition of the rest of the ring.

In order to bring the problem within the reach of our mathematical methods,
we limit it to the case in which the ring is nearly circular and uniform, and has
a transverse section very small compared with the radius of the ring. By
analysing the difficulties of the theory of a linear ring, we shall be better able
to appreciate those which occur in the theory of the actual rings.

ON THE STABILITY OF THE MOTION OP SATURN'S KINGS. 311

The ring which we consider is therefore small in section, and very nearly
circular and uniform, and revolving with nearly uniform velocity. The variations
from circular form, uniform section, and uniform velocity must be expressed by a
proper notation.

2. To express the position of an element of a variable ring at a given time
in terms of the original position of the element in the ring.

Let S (fig. 3) be the central body, and SA a direction fixed in space.

Let SB be a radius, revolving with the mean angular velocity w of the
ring, so that ASB = <at.

Let TT be an element of the ring in its actual position, and let P be the
position it would have had if it had moved uniformly with the mean velocity a
and had not been displaced, then BSP is a constant angle =s, and the value
of s enables us to identify any element of the ring.

The element may be removed from its mean position P hi three different
ways.

(1) By change of distance from S by a quantity pir = p.

(2) By change of angular position through a space Pp = a.

(3) By displacement perpendicular to the plane of the paper by a quantity £.

p, cr and £ are all functions of s and t. If we could calculate the attrac-
tions on. any element as depending on the form of these functions, we might
determine the motion of the ring for any given original disturbance. We cannot,
however, make any calculations of this kind without knowing the form of the
functions, and therefore we must adopt the following method of separating the
original disturbance into others of simpler form, first given in Fourier's Traite
de Chaleur.

3. Let U be a function of s, it is required to express U in a series of
sines and cosines of multiples of s between the values s = 0 and s = 2jr.

Assume U=Al cos s + A3 cos 2s + &c. + Am cos ms + An cos ns
+ Bl sin s + Bt cos 2s + &c. + Bm sin ws + Bn sin ns.

312 ON THE STABILITY OF THE MOTION OF SATUKN's RINGS.

Multiply by COB msds and integrate, then all terms of the form

J cos ms cos nsds and / cos ms sin nsds
will vanish, if we integrate from 5 = 0 to s = 2ir, and there remains

U cos msds = irAm, U sin msds = irBn .

If we can determine the values of these integrals in the given case, we
can find the proper coefficients Am, Bm, &c., and the series will then represent
the values of U from s = 0 to s = 2ir, whether those values be continuous or
discontinuous, and when none of those values are infinite the series will be
convergent.

In this way we may separate the most complex disturbances of a ring into
parts whose form is that of a circular function of s or its multiples. Each of
these partial disturbances may be investigated separately, and its effect on the
attractions of the ring ascertained either accurately or approximately.

4. To find the magnitude and direction of the attraction between two
elements of a disturbed ring.

Let P and Q (fig. 4) be the two elements, and let their original positions
be denoted by s, and s,, the values of the arcs BP, BQ before displacement.
The displacement consists in the angle BSP being increased by o-j and BSQ
by <TJ, while the distance of P from the centre is increased by pt and that of
Q by ps. We have to determine the effect of these displacements on the distance
PQ and the angle SPQ.

Let the radius of the ring be unity, and st — s1 = 20, then the original
value of PQ will be 2 sin 6, and the increase due to displacement

= (pa + pi) sin 6 + (<ra - o-,) cos 0.
We may write the complete value of PQ thus,

} ............... (1).

The original value of the angle SPQ was n~^» anc^ ^ne increase due to
displacement is i (p» — PI) cot 0 — £ (<r, — <r,),

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 313

so that we may write the values of sin SPQ and cos SPQ,

(2),
} ...... (3).

If we assume the masses of P and Q each equal to - R, where R is the

V-

mass of the ring, and /x the number of satellites of which it is composed, the
accelerating effect of the radial force on P is

and the tangential force

1 „ sin SPQ _ 1 R cos 0 , _^ _, , , , «., , ,

1 L — £

The normal force is - R r* , ,' .
p. 8 sin 0

5. Let us substitute for p, a- and £ their values expressed in a series of
sines and cosines of multiples of s, the terms involving ms being

P! = A cos (ms + a), pa = A cos (ms + a

cr, = B sin (ms + ft), o-,=

£ = C cos (ms + y), £3 =

The radial force now becomes

• 1 — A cos (ms + a) (1+ cos 2m0) + A sin (ms + a) sin 2m0 T

. — -n> -\-\A cos(ms + a) (1 — cos2»i0) cot20 — \A sin (ms + a)sin 2m#cot20 \ (6).

/A 4 sm 0

. + \B sin (ms + ft) ( 1 — cos 2m0) cot 0 — %B cos (ms + ft) sin 2m0 cot 0 J

The radial component of the attraction of a corresponding particle on the
other side of P may be found by changing the sign of 0. Adding the two
together, we have for the effect of the pair

1 J?

{l —A cos (ms + a) (2 cos5 m0 — sin2 m0 cot2 0)

} (7).

VOL. I. 40

:u4

ON THE STABILITY OF THE MOTION OF SATURN'S RINOS.

Let us put

, /
= S
\

sin'0
sin 2m0 cos

sm0

4sin

0 cos 6\
in'0 )

-r
=

.7 = 2

sn1

0 sin'

sin

(8)*,

where the summation extends to all the satellites on the same side of P,

x

that is, every value of 6 of the form - IT, where x is a whole number less
than .

The radial force may now be written

P = - R {K+ LA cos (ms + a) - MB cos (ms + ft)}

(9).

* The following values of several quantities which enter into these investigations are calculated for a
ring of 36 satellites.

£=24-5.

„ sin* m$ cos'tf ^COB'III* r

3f

.V

sin3^

sin 9

»» = 0 0

43 -43

0

0

m= I 32

32 -16

16

37

m= 2 107

28 26

25

115

»» = 3 212

25 81

28

221

w= 4 401

24 177

32

411

m= 9 975

20 468

30

986

n»=18 1569

18 767

0

1582

When /

t is very great,

TT

» „
L = -5259 when m = ^ ,

M

•

•4342

•3287 „

e

4'

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 315

The tangential force may be calculated in the same way, it is

T=- Jl{MAsm(ms + a) + NJBsin.(ms + P)} (10).

The normal force is

Z= --RJCcos(ms + -y) (11).

6. We have found the expressions for the forces which act upon each
member of a system of equal satellites which originally formed a uniform ring,
but are now affected with displacements depending on circular functions. If
these displacements can be propagated round the ring in the form of waves

/f¥L

with the velocity -, the quantities a, /3, and y will depend on t, and the

complete expressions will be

p = A cos (ms + nt + a) T

o- = Bsin.(ms + nt + P) L (12).

£ = C cos (ms + n't + y) J

Let us find in what cases expressions such as these will be true, and
what will be the result when they are not true.

Let the position of a satellite at any time be determined by the values
of r, <£, and £, where r is the radius vector reduced to the plane of reference,
<£ the angle of position measured on that plane, and £ the distance from it.
The equations of motion will be

.(13).

If we substitute the value of £ in the third equation and remember that r
is nearly =1, we find

n*m8+-BJ. (14).

As this expression is necessarily positive, the value of n is always real,
and the disturbances normal to the plane of the ring can always be propa-

40—2

316 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

gated as waves, and therefore can never be the cause of instability. We
therefore confine our attention to the motion in the plane of the ring ,-i>
deduced from the two former equations.

Putting r = 1 + p and <j) = <at + s + <r, and omitting powers and products of
p, a- and their differential coefficients,

(15).

Substituting the values of p and cr as given above, these equations become

f* *• Tl T^ I * . f*. f*t *•

\

— RM\ B cos (r>is + nt + /B) = 0 (16),

/*

1 1

— RM) A sin (ms+ nt + a) + (n~-\ — RN) Bsin (ms + nt + /3) = 0....(17).

p. p.

Putting for (ms + nt) any two different values, we find from the second
equation (17)

«=0 (18).

nd 2 -RM)A >' ±RN B-

/x, p.

and from the first (16) (tf + ZS-- RL + n") A + (2a>n+ - RAf) B~Q (20),

and a>*-S--RK=0 (21).

V-

Eliminating A and B from these equations, we get
n'-{3o>'-2S+~R(L-N)}n*

1 111

a hiquadratic equation to determine n.

For every real value of n there are terms in the expressions for p and cr

of the form

A cos (ins + nt + a).

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 317

For every pure impossible root of the form ±«/ — In' there are terms of
the forms

At*"'1 cos (ms + a).

Although the negative exponential coefficient indicates a continually diminishing
displacement which is consistent with stability, the positive value which neces-
sarily accompanies it indicates a continually increasing disturbance, which would
completely derange the system in course of time.

For every mixed root of the form ±\/ — In' + n, there are terms of the form

Af* cos (ms + nt + a).
If we take the positive exponential, we have a series of m waves travelling

71 /

with velocity and increasing in amplitude with the coefficient e+n>t. The

negative exponential gives us a series of m waves gradually dying away, but
the negative exponential cannot exist without the possibility of the positive one
having a finite coefficient, so that it is necessary for the stability of the motion
that the four values of n be all real, and none of them either impossible
quantities or the sums of possible and impossible quantities.

We have therefore to determine the relations among the quantities Kt L,
M, N, R, S, that the equation

n« _ {S+ 1 R (3K+ L - N)} n3

-4(a~RMn + {SS+ - R (K-L}} - RN- I R*M* = U= 0

r1 /* r" P1

may have four real roots.

7. In the first place, U is positive, when n is a large enough quantity,
whether positive or negative.

It is also positive when n = 0, provided S be large, as it must be, com-

pared with - RL, - RM and - RN.
P* /* /*

If we can now find a positive and a negative value of n for which U
is negative, there must be four real values of n for which U=0, and the four
roots will be real.

318 ON THE STABILITY OF THE MOTION OF SATURN'S KINGS.

Now if we put n=

which is negative if S be large compared to R.

So that a ring of satellites can always be rendered stable by increasing
the mass of the central body and the angular velocity of the ring.

The values of L, Af, and N depend on TO, the number of undulations in
the ring. When m = ~, the values of L and N will be at their maximum

£t

and M=0. If we determine the relation between S and R in this case so
that the system may be stable, the stability of the system for every other
displacement will be secured.

8. To find the mass which must be given to the central body in order
that a ring of satellites may permanently revolve round it.

We have seen that when the attraction of the central body is sufficiently
great compared with the forces arising from the mutual action of the satellites,
a permanent ring is possible. Now the forces between the satellites depend on
the manner in which the displacement of each satellite takes place. The con-
ception of a perfectly arbitrary displacement of all the satellites may be rendered
manageable by separating it into a number of partial displacements depending
on periodic functions. The motions arising from these small displacements will
take place independently, so that we have to consider only one at a time.

Of all these displacements, that which produces the greatest disturbing
forces is that in which consecutive satellites are oppositely displaced, that is,

when in = ^, for then the nearest satellites are displaced so as to increase as

A

much as possible the effects of the displacement of the satellite between them.
If we make fi a large quantity, we shall have

-5259, M=0, N=2L, K very small.

7T

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 319

Let - RL = x. then the equation of motion will be
P-

x)=U=Q ....................... (23).

The conditions of this equation having real roots are

S>x ....................................... (24),

(S-x)*>8x(3S-x) ..... ...................... (25).

The last condition gives the equation

whence S>2G'642x, or S<0-351a: ........................ (26).

The last solution is inadmissible because S must be greater than x, so that
the true condition is *S>25'649x,

> 25-649 -S^ '5259,
p. if

.................................... (27).

So that if there were 100 satellites in the ring, then

5>4352 R

is the condition which must be fulfilled in order that the motion arising from
every conceivable displacement may be periodic.

If this condition be not fulfilled, and if S be not sufficient to render the
motion perfectly stable, then although the motion depending upon long undu-
lations may remain stable, the short undulations will increase in amplitude till
some of the neighbouring satellites are brought into collision.

9. To determine the nature of the motion when the system of satellites
is of small mass compared with the central body.

The equation for the determination of n is

U= n< - {«' + - R (2K+ L - N}} n* - 4o> - RMn

Q ......... (28).

When R is very small we may approximate to the values of n by assuming
that two of them are nearly ± <a, and that the other two are small.

320

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

If we put n= ±0),

U= --

Therefore the corrected values of n are

n=±{(a+ _L R (2K + L-4N)} + —RM.
2p.oi u.(a

(29).

The small values of n are nea.rly±/3-RN: correcting them in the same
way, we find the approximate values

(30).

The four values of n are therefore

n1= -o»-

.__

„,= + 3-RN- —

V ft /ACD

RM

(31),

and the complete expression for p, so far as it depends on terms containing ms,
is therefore f» = -A.\ cos (ms + nj, + a,) + At cos (ms + nji + a,)

+ A, cos (ms + njt, + a,) + At cos (ms + ntt + at) ............... (32),

and there will be other systems, of four terms each, for every value of m in
the expansion of the original disturbance.

We are now able to determine the value of <r from equations (12), (20), by
putting /8 = a, and

2<un + - RM

B=

(33).

ON THE STABILITY OF THE MOTION OF SATUKN's RINGS. 321

So that for every term of p of the form

p = A cos(ms + nt + a) (34),

there is a corresponding term in cr,

2&m + - RM
a- = -^ A sin (ms + nt + a) (35).

10. Let us now fix our attention on the motion of a single satellite,
and determine its motion by tracing the changes of p and cr while t varies
and s is constant, and equal to the value of s corresponding to the satellite
in question.

We must recollect that p and cr are measured outwards and forwards from
an imaginary point revolving at distance 1 and velocity «, so that the motions
we consider are not the absolute motions of the satellite, but its motions
relative to a point fixed in a revolving plane. This being understood, we may
describe the motion as elliptic, the major axis being in the tangential direc-
tion, and the ratio of the axes being nearly 2 - , which is nearly 2 for n^ and «4

71

and is very large for n., and nt.

The time of revolution is — , or if we take a revolution of the ring as

ft

the unit of time, the time of a revolution of the satellite about its mean

position is - .
n

The direction of revolution of the satellite about its mean position is in
every case opposite to that of the motion of the ring.

11. The absolute motion of a satellite may be found from its motion
relative to the ring by writing

r=l+p = l+Acos (ms + nt + a),

— 2 - A sin (ms + nt + a).
n

VOL. I. 41

322 ON THE STABILITY OP THE MOTION OF SATURN'S RINGS.

When n is nearly equal to ±<a, the motion of each satellite in space is
nearly elliptic. The eccentricity is A, the longitude at epoch s, and the longi-
tude when at the greatest distance from Saturn is for the negative value n,

- R (2 K+ L - iM - 4N) t + (m + 1 ) * + a,
and for the positive value n4

-R(2K+L + 4M-tN)t~(m+l)s-a.

fid)

We must recollect that in all cases the quantity within brackets is negative,
so that the major axis of the ellipse travels forwards in both cases. The chief
difference between the two cases lies in the arrangement of the major axes of
the ellipses of the different satellites. In the first case as we pass from one
satellite to the next in front the axes of the two ellipses lie in the same
order. In the second case the particle in front has its major axis behind that
of the other. In the cases in which n is small the radius vector of each
satellite increases and diminishes during a periodic time of several revolutions.
This gives rise to an inequality, in which the tangential displacement far exceeds
the radial, as in the case of the annual equation of the Moon.

12. Let us next examine the condition of the ring of satellites at a given
instant. We must therefore fix on a particular value of t and trace the changes
of p and o- for different values of s.

From the expression for p we learn that the satellites form a wavy line,
which is furthest from the centre when (ms + nt + a) is a multiple of 2,ir, and
nearest to the centre for intermediate values.

From the expression for cr we learn that the satellites are sometimes in
advance and sometimes in the rear of their mean position, so that there are
places where the satellites are crowded together, and others where they are
drawn asunder. When n is positive, B is of the opposite sign to A, and the
crowding of the satellites takes place when they are furthest from the centre.
When n is negative, the satellites are separated most when furthest from the
centre, and crowded together when they approach it.

The form of the ring at any instant is therefore that of a string of beads
forming a re-entering curve, nearly circular, but with a small variation of distance

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 323

from the centre recurring m times, and forming m regular waves of trans-
verse displacement at equal intervals round the circle. Besides these, there are
waves of condensation and rarefaction, the effect of longitudinal displacement.
When n is positive the points of greatest distance from the centre are points
of greatest condensation, and when n is negative they are- points of greatest
rarefaction.

13. We have next to determine the velocity with which these waves of
disturbance are propagated round the ring. We fixed our attention on a par-
ticular satellite by making s constant, and on a particular instant by making t
constant, and thus we determined the motion of a satellite and the form of the
ring. We must now fix our attention on a phase of thex motion, and this we
do by making p or <r constant. This implies

ms + nt + a = constant,

ds _ n
dt~ in'

So that the particular phase of the disturbance travels round the ring with an

77i

angular velocity = relative to the ring itself. Now the ring is revolving

in space with the velocity ta, so that the angular velocity of the wave in space is

vr = o»-- (36).

m

Thus each satellite moves in an ellipse, while the general aspect of the
ring is that of a curve of m waves revolving with velocity w. This, however,
is only the part of the whole motion, which depends on a single term of the
solution. In order to understand the general solution we must shew how to
determine the whole motion from the state of the ring at a given instant.

14. Given the position and motion of every satellite at any one time, to
calculate the position and motion of every satellite at any other time, provided
that the condition of stability is fulfilled.

The position of any satellite may be denoted by the values of p and a- for
that satellite, and its velocity and direction of motion are then indicated by the

values of -/- and -j~ at the given instant.
at at

41—2

324 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

These four quantities may have for each satellite any four arbitrary values,
as the position and motion of each satellite are independent of the rest, at the
beginning of the motion.

Each of these quantities is therefore a perfectly arbitrary function of s, the
mean angular position of the satellite in the ring.

But any function of s from s = 0 to s = 2ir, however arbitrary or discontinuous,
can be expanded in a series of terms of the form A cos (.? + a) + A' cos (2s + a) + &c.
See § 3.

Let each of the four quantities p, -£ , <r, --j- be expressed in terms of such
a series, and let the terms in each involving tns be

(37),

(38),
o- = G cos (ms +g) .............................. (39),

(40).

These are the parts of the values of each of the four quantities which are
capable of being expressed in the form of periodic functions of ms. It is
evident that the eight quantities E, F, G, H, e, f, g, h, are all independent and
arbitrary.

The next operation is to find the values of L, M, N, belonging to disturb-
ances in the ring whose index is in [see equation (8)], to introduce these
values into equation (28), and to determine the four values of n, (»,, n,, n,, «4).

This being done, the expression for p is that given in equation (32), which
contains eight arbitrary quantities (At, At, At, Au a,, a,, a,, a4).

Giving t its original value in this expression, and equating it to Ecos(ms + e),
we get an equation which is equivalent to two. For, putting ms = 0, we have

Al cos <LI + At cos a, + A, cos a, + At cos a4 = E cos e ............ (41).

And putting ms = , we have another equation

£i

Al sin a, + A* sin o, + -4, sin a, + A4 sin a4 = E sin e

ON THE STABILITY OF THE MOTION OF SATUBN'g RINGS. 325

Differentiating (32) with respect to t, we get two other equations

— A^ sin a — &c. = F cos/ ........................... (43),

A^ cos a + &c. = Fsmf ........................... (44).

Bearing in mind that Bl} B^ &c. are connected with Alt At, &c. by equa-
tion (33), and that B is therefore proportional to A, we may write B = Af$,
where

2o>n + -

R- *

P- --

yS being thus a function of n and a known quantity.
The value of or then becomes at the epoch

cr = Aj3l sin (ww + o^+.&c. = (7 cos (ms + g),
from which we obtain the two equations

sin a, + &c. = Gcos g .............................. (45),

oj + cfec. = — 6r sin<7 ........................... (46).

Differentiating with respect to t, we get the remaining equations

^ coso, + &c. =Hcosh ......................... (47),

i! sin al + &c. = H sin h ........................... (48).

We have thus found eight equations to determine the eight quantities
A1} &c. and 04, &c. To solve them, we may take the four in which A1coBal,
&c. occur, and treat them as simple equations, so as to find ^4 j cos a,, &c. Then
taking those in which A1 sin a,, &c. occur, and determining the values of those
quantities, we can easily deduce the value of A1 and Oj, &c. from these.

We now know the amplitude and phase of each of the four waves whose
index is m. All other systems of waves belonging to any other index must
be treated in the same way, and since the original disturbance, however arbitrary,
can be broken up into periodic functions of the form of equations (37 — 40),
our solution is perfectly general, and applicable to every possible disturbance of
a ring fulfilling the condition of stability (27).

326 ON THE STABILITY OF THE MOTION OF SATURN'S EINOS.

15. We come next to consider the effect of an external disturbing force,
due either to the irregularities of the planet, the attraction of satellites, or
the motion of waves in other rings.

All disturbing forces of this kind may be expressed in series of which the
general term is

A cos^vt+ms + a),

where v is an angular velocity and m a whole number.

Let Pcos(ms + vt+p) be the central part of the force, acting inwards, and
Q sin (ms + vt + q) the tangential part, acting forwards. Let p = A cos (ms + rt + a)
and o- = -Bsin (ms + vt + ft), be the terms of p and <r which depend on the
external disturbing force. These will simply be added to the terms depending
on the original disturbance which we have already investigated, so that the
complete expressions for p and a- will be as general as before. In consequence
of the additional forces and displacements, we must add to equations (16) and
(17), respectively, the following terms:

{3<w' - - R (2K+ L)+tf}A cos (ms +vt + a)

+ (2<av + -RM)BcoB(ms + vt + @)-Pcos (ms + vt+p) = 0 ...... (49).

(2a>v + - RM) A sin (ms + vt + a)

+ (v' + - RN) E sin (ms + vt + /3) + Q Bm(ms + vt + q) = 0 ......... (50).

Making ms + vt = 0 in the first equation and - in the second,

8

co8a + (2<av + -RM)BcoBp-PcoBp = Q ...... (51).

fii

-RM)Acoaa+(v> + -RN)BcoBp + QcoBq = 0 ...... (52).

Then if we put
U' = tf-{<S+-

-4 -RMv

(53),

ON THE STABILITY OF THE MOTION OP SATURN'S RINGS. 327

we shall find the value of A cos a and B cos y8 ;

Acoaa = - - Pcos^H -- - Qcosq ............. (54).

Bcoafi= T£ Pcosp £, Qcosq (55),

Substituting sines for cosines in equations (51), (52), we may find the
values of A sin a and B sin y8.

Now U' is precisely the same function of v that U is of n, so that if v
coincides with one of the four values of n, U' will vanish, the coefficients A
and B will become infinite, and the ring will be destrpyed. The disturbing
force is supposed to arise from a revolving body, or an undulation of any kind

which has an angular velocity relatively to the ring, and therefore an

absolute angular velocity = a .

If then the absolute angular velocity of the disturbing body is exactly or
nearly equal to the absolute angular velocity of any of the free waves of the
ring, that wave will increase till the ring be destroyed.

The velocities of the free waves are nearly

1\ 1 r~l~ I /~T~ / 1\

— , o>+ fZ-RN. <a 3-RN, and <a (1 (56).

m/ m V P- m \ p \ m/

When the angular velocity of the disturbing body is greater than that of
the first wave, between those of the second and third, or less than that of
the fourth, U' is positive. When it is between the first and second, or between
the third and fourth, U' is negative.

Let us now simplify our conception of the disturbance by attending to the
central force only, and let us put p = 0, so that P is a maximum when ms + vt
is a multiple of 2-n. We find in this case a = 0, and £ = 0. Also

1

328 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

When U' is positive, A will be of the same sign as P, that is, the parts
of the ring will be furthest from the centre where the disturbing force towards
the centre is greatest. When U' is negative, the contrary will be the case.

When v is positive, B will be of the opposite sign to A, and the parts
of the ring furthest from the centre will be most crowded. When v is negative,
the contrary will be the case.

Let us now attend only to the tangential force, and let us put q = 0. We
find in this case also a = 0, /8 = 0,

Q (60).

The tangential displacement is here in the same or in the opposite direc-
tion to the tangential force, according as U' is negative or positive. The
crowding of satellites is at the points farthest from or nearest to Saturn
according as v is positive or negative.

16. The effect of any disturbing force is to be determined in the following
manner. The disturbing force, whether radial or tangential, acting on the ring
may be conceived to vary from one satellite to another, and to be different at
different times. It is therefore a perfectly arbitrary function of s and t.

Let Fourier's method be applied to the general disturbing force so as to
divide it up into terms depending on periodic functions of s, so that each term
is of the form F (t) cos (ras + a), where the function of t is still perfectly arbitrary.

But it appears from the general theory of the permanent motions of the
heavenly bodies that they may all be expressed by periodic functions of t
arranged in seriea Let vt be the argument of one of these terms, then the
corresponding term of the disturbance will be of the form

P cos (ms + vt + a).

This term of the disturbing force indicates an alternately positive and
negative action, disposed in m waves round the ring, completing its period

ON THE STABILITY OF THE MOTION OF SATUEN's RINGS. 329

o_
relatively to each particle in the time - , and travelling as a wave among

v
the particles with an angular velocity , the angular velocity relative to fixed

Ilif

qj

space being of course 01 — - . The whole disturbing force may be split up into

In*

terms of this kind.

17. Each of these elementary disturbances will produce its own wave in
the ring, independent of those which belong to the ring itself. This new wave,
due to external disturbance, and following different laws from the natural waves
of the ring, is called the forced wave. The angular velocity of the forced wave
is the same as that of the disturbing force, and its maxima and minima coin-
cide with those of the force, but the extent of the disturbance and its direction
depend on the comparative velocities of the forced wave and the four natural
waves.

When the velocity of the forced wave lies between the velocities of the
two middle free waves, or is greater than that of the swiftest, or less than
that of the slowest, then the radial displacement due to a radial disturbing
force is in the same direction as the force, but the tangential displacement
due to a tangential disturbing force is in the opposite direction to the force.

The radial force therefore in this case produces a positive forced wave, and
the tangential force a negative forced ivave.

When the velocity of the forced wave is either between the velocities of
the first and second free waves, or between those of the third and fourth, then
the radial disturbance produces a forced wave in the contrary direction to that
in which it acts, or a negative wave, and the tangential force produces a positive
wave.

The coefficient of the forced wave changes sign whenever its velocity passes
through the value of any of the velocities of the free waves, but it does so
by becoming infinite, and not by vanishing, so that when the angular velocity
very nearly coincides with that of a free wave, the forced wave becomes very
great, and if the velocity of the disturbing force were made exactly equal to
that of a free wave, the coefficient of the forced wave would become infinite.
In such a case we should have to readjust our approximations, and to find
whether such a coincidence might involve a physical impossibility.

VOL. I. 42

330 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

The forced wave which we have just investigated is that which would main-
tain itself in the ring, supposing that it had been set agoing at the commence-
ment of the motion. It is in fact the form of dynamical equilibrium of the
ring under the influence of the given forces. In order to find the actual motion
of the ring we must combine this forced wave with all the free waves, which
go on independently of it, and in this way the solution of the problem becomes
perfectly complete, and we can determine the whole motion under any given
initial circumstances, as we did in the case where no disturbing force acted.

For instance, if the ring were perfectly uniform and circular at the instant
when the disturbing force began to act, we should have to combine with the
constant forced wave a system of four free waves so disposed, that at the given
epoch, the displacements due to them should exactly neutralize those due to the
forced wave. By the combined effect of these four free waves and the forced
one the whole motion of the ring would be accounted for, beginning from its
undisturbed state.

The disturbances which are of most importance in the theory of Saturn's
rings are those which are produced in one ring by the action of attractive
forces arising from waves belonging to another ring.

The effect of this kind of action is to produce in each ring, besides its
own four free waves, four forced waves corresponding to the free waves of the
other ring. There will thus be eight waves in each ring, and the corresponding
waves in the two rings will act and react on each other, so that, strictly speak-
ing, every one of the waves will be in some measure a forced wave, although
the system of eight waves will be the free motion of the two rings taken
together. The theory of the mutual disturbance and combined motion of two
concentric rings of satellites requires special consideration.

18. On the motion of a ring of satellites when the conditions of stability
are not fulfilled.

We have hitherto been occupied with the case of a ring of satellites, the
stability of which was ensured by the smallness of mass of the satellites com-
pared with that of the central body. We have seen that the statically unstable
condition of each satellite between its two immediate neighbours may be com-
pensated by the dynamical effect of its revolution round the planet, and a planet
of sufficient mass can not only direct the motion of such satellites round its

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 331

own body, but can likewise exercise an influence over their relations to each
other, so as to overrule their natural tendency to crowd together, and distribute
and preserve them in the form of a ring.

"We have traced the motion of each satellite, the general shape of the
disturbed ring, and the motion of the various waves of disturbance round the
ring, and determined the laws both of the natural or free waves of the ring,
and of the forced waves, due to extraneous disturbing forces.

We have now to consider the cases in which such a permanent motion of
the ring is impossible, and to determine the mode in which a ring, originally
regular, will break up, in the different cases of instability.

The equation from which we deduce the conditions of stability is —

U = n* - jo>= + - R (2K+ L - N)\ ri1 - 4o> - RMn
I V- P

+ { 3o>' - - R (2K+ L)\ - RN - -2 KM* = 0.
II* J V" P"

The quantity, which, in the critical cases, determines the nature of the
roots of this equation, is JV. The quantity M in the third term is always
small compared with L and N when ra is large, that is, in the case of the
dangerous short waves. We may therefore begin our study of the critical cases
by leaving out the third term. The equation then becomes a quadratic in if,
and in order that all the values of n may be real, both values of na must be
real and positive.

The condition of the values of w* being real is

>0 (61),

/* ' P-'

which shews that CD* must either be about 14 times at least smaller, or about 14
times at least greater, than quantities like - RN.

That both values of n* may be positive, we must have

.1. J-\ / _ fr . f Tl T\ /\

3^-~R(2K + L)\-RN>0

..(62).
42—2

332 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

We must therefore take the larger value of to1, and also add the condition
that N be positive.

We may therefore state roughly, that, to ensure stability, - - , the coefficient

of tangential attraction, must lie between zero and i^V. If the quantity be
negative, the two small values of n will become pure impossible quantities. If
it exceed -fa?, all the values of n will take the form of mixed impossible
quantities.

If we write x for - RN, and omit the other disturbing forces, the equation
becomes U=nt-(ot'-x)nl + 3ottx = () (63),

whence n* = •£ (<u9 - x) ± % -Jot4 - 1 4a>ax + x* .................. (64).

If a; be small, two of the values of n are nearly ±(o, and the others are
small quantities, real when x is positive and impossible when x is negative.

If x be greater than (7— »/48)o)a, or — - nearly, the term under the radical
becomes negative, and the value of n becomes

......... (65),

where one of the terms is a real quantity, and the other impossible. Every
solution may be put under the form

n=p±J^lq ................................. (66),

where q = 0 for the case of stability, p = 0 for the pure impossible roots, and p
and q finite for the mixed roots.

Let us now adopt this general solution of the equation for n, and determine
its mechanical significance by substituting for the impossible circular functions
their equivalent real exponential functions.

Substituting the general value of n in equations (34), (35),

... (67),

(68).

• -A

ON THE STABILITY OP THE MOTION OF SATURN'S KINGS. 333

Introducing the exponential notation, these values become

cr= —

) (69),

2c*A f p(pz + qs + x)(eqt + e-^)sm(ms+pt + a)\ , ,

'(p* + qiy + 2(p'i-q')x + xa\ + q(p' + qi-x) (e9* - e'9*) cos (ms +j>« + a) J " ' ^ ''

We have now obtained a solution free from impossible quantities, and applicable
to every case.

When q = 0, the case becomes that of real roots, which we have already
discussed. When p = Q, we have the case of pure impossible roots arising from
the negative values of n\ The solutions corresponding to these roots are

(71).

<r = - (t«t-t-'>t) cos (ms + a). ................ (72).

The part of the coefficient depending on e~qt diminishes indefinitely as the
time increases, and produces no marked effect. The other part, depending on
e7', increases in a geometrical proportion as the time increases arithmetically, and
so breaks up the ring. In the case of x being a small negative quantity, (f is
nearly 3x, so that the coefficient of <r becomes

-3° A. , -

?

It appears therefore that the motion of each particle is either outwards and
backwards or inwards and forwards, but that the tangential part of the motion
greatly exceeds the normal part.

It may seem paradoxical that a tangential force, acting towards a position
of equilibrium, should produce instability, while a small tangential force from that
position ensures stability, but it is easy to trace the destructive tendency of
this apparently conservative force.

Suppose a particle slightly in front of a crowded part of the ring, then
if x is negative there will be a tangential force pushing it forwards, and this
force will cause its distance from the planet to increase, its angular velocity to
diminish, and the particle itself to fall back on the crowded part, thereby
increasing the irregularity of the ring, till the whole ring is broken up. In
the same way it may be shewn that a particle behind a crowded part will be
pushed into it. The only force which could preserve the ring from the effect

334 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

of this action, is one which would prevent the particle from receding from the
planet under the influence of the tangential force, or at least prevent the dimi-
nution of angular velocity. The transversal force of attraction of the ring is of
this kind, and acts in the right direction, but it can never be of sufficient magni-
tude to have the required effect. In fact the thing to be done is to render the
last term of the equation in n* positive when N is negative, which requires

and this condition is quite inconsistent with any constitution of the ring which
fulfils the other condition of stability which we shall arrive at presently.

We may observe that the waves belonging to the two real values of n,
±(a, must be conceived to be travelling round the ring during the whole time
of its breaking up, and conducting themselves like ordinary waves, till the
excessive irregularities of the ring become inconsistent with their uniform propa-
gation.

The irregularities which depend on the exponential solutions do not travel
round the ring by propagation among the satellites, but remain among the same
satellites which first began to move irregularly.

We have seen the fate of the ring when x is negative. When x is small
we have two small and two large values of n, which indicate regular waves,
as we have already shewn. As x increases, the small values of n increase, and
the large values diminish, till they meet and form a pair of positive and a
pair of negative equal roots, having values nearly ±'68w. When x becomes
greater than about -^ta', then all the values of n become impossible, of the
form p + -J — Iq, q being small when x first begins to exceed its limits, and p
being nearly ± '68a>.

2ir
The values of p and a- indicate periodic inequalities having the period - - ,

but increasing in amplitude at a rate depending on the exponential c". At the
beginning of the motion the oscillations of the particles are in ellipses as in the
case of stability, 'having the ratio of the axes about 1 in the normal direction
to 3 in the tangential direction. As the motion continues, these ellipses increase
in magnitude, and another motion depending on the second term of <r is com-
bined with the former, so as to increase the ellipticity of the oscillations and to

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 335

turn the major axis into an inclined position, so that its fore end points a little
inwards, and its hinder end a little outwards. The oscillations of each particle
round its mean position are therefore in ellipses, of which both axes increase
continually while the eccentricity increases, and the major axis becomes slightly
inclined to the tangent, and this goes on till the ring is destroyed. In the
mean time the irregularities of the ring do not remain among the same set of
particles as in the former case, but travel round the ring with a relative angular

ar\

velocity — — . Of these waves there are four, two travelling forwards among the

satellites, and two travelling backwards. One of each of these pairs depends
on a negative value of q, and consists of a wave whose amplitude continually
decreases. The other depends on a positive value of q, and is the destructive
wave whose character we have just described.

19. We have taken the case of a ring composed of equal satellites, as
that with which we may compare other cases in which the ring is constructed
of loose materials differently arranged.

In the first place let us consider what will be the conditions of a ring
composed of satellites of unequal mass. We shall find that the motion is of
the same kind as when the satellites are equal.

For by arranging the satellites so that the smaller satellites are closer
together than the larger ones, we may form a ring which will revolve uni-
formly about Saturn, the resultant force on each satellite being just sufficient
to keep it in its orbit.

To determine the stability of this kind of motion, we must calculate the
disturbing forces due to any given displacement of the ring. This calculation
will be more complicated than in the former case, but will lead to results of
the same general character. Placing these forces in the equations of motion,
we shall find a solution of the same general character as in the former case,
only instead of regular waves of displacement travelling round the ring, each
wave will be split and reflected when it comes to irregularities in the chain of
satellites. But if the condition of stability for every kind of wave be fulfilled,
the motion of each satellite will consist of small oscillations about its position
of dynamical equilibrium, and thus, on the whole, the ring will of itself assume
the arrangement necessary for the continuance of its motion, if it be originally
in a state not very different from that of equilibrium.

336 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

20. We now pass to the case of a ring of an entirely different construc-
tion. It is possible to conceive of a quantity of matter, either solid or liquid,
not collected into a continuous mass, but scattered thinly over a great extent
of space, and having its motion regulated by the gravitation of its parts to
each other, or towards some dominant body. A shower of rain, hail, or cinders
is a familiar illustration of a number of unconnected particles in motion ; the
visible stars, the milky way, and the resolved nebulae, give us instances of a
similar scattering of bodies on a larger scale. In the terrestrial instances we
see the motion plainly, but it is governed by the attraction of the earth, and
retarded by the resistance of the air, so that the mutual attraction of the
parts is completely masked. In the celestial cases the distances are so enor-
mous, and the time during which they have been observed so short, that we
can perceive no motion at all. Still we are perfectly able to conceive of ;i
collection of particles of small size compared with the distances between them,
acting upon one another only by the attraction of gravitation, and revolving
round a central body. The average density of such a system may be smaller
than that of the rarest gas, while the particles themselves may be of great
density ; and the appearance from a distance will be that of a cloud of vapour,
with this difference, that as the space between the particles is empty, the rays
of light will pass through the system without being refracted, as they would
have been if the system had been gaseous.

Such a system will have an average density which may be greater in some
places than others. The resultant attraction will be towards places of greater
average density, and thus the density of those places will be increased so us
to increase the irregularities of density. The system will therefore be statically
unstable, and nothing but motion of some kind can prevent the particles from
forming agglomerations, and these uniting, till all are reduced to one solid
mass.

We have already seen how dynamical stability can exist where there is
statical instability in the case of a row of particles revolving round a central
body. Let us now conceive a cloud of particles forming a ring of nearly uni-
form density revolving about a central body. There will be a primary effect of
inequalities in density tending to draw particles towards the denser parts of the
ring, and this will elicit a secondary effect, due to the motion of revolution,
tending in the contrary direction, BO as to restore the rings to uniformity. The

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 337

relative magnitude of these two opposing forces determines the destruction or
preservation of the ring.

To calculate these effects we must begin with the statical problem: — To
determine the forces arising from the given displacements of the ring.

The longitudinal force arising from longitudinal displacements is that which
has most effect in determining the stability of the ring. In order to estimate its
limiting value we shall solve a problem of a simpler form.

21. An infinite mass, originally of uniform density k, has its particles
displaced by a quantity £ parallel to the axis of x, so that g=Acosmx, to
determine the attraction on each particle due to this displacement.

The density at any point will differ from the original density by a quantity
k', so that

(73),

k'= — k~,~ = Akm sin mx .......................... (74).

The potential at any point will be V+ V, where V is the original potential,
and V depends on the displacement only, so that

ffiV d'V d*V

Now V is a function of x only, and therefore,

— sinmaj .......................... (76),

m v '

and the longitudinal force is found by differentiating V with respect to x.

dV
X= j— = ^TrkA cosmx = iirk£ ..................... (77).

Now let us suppose this mass not of infinite extent, but of finite section
parallel to the plane of yz. This change amounts to cutting off all portions
of the mass beyond a certain boundary. Now the effect of the portion so cut
off upon the longitudinal force depends on the value of m. When m is large,
so that the wave-length is small, the effect of the external portion is insensible,
so that the longitudinal force due to short waves is not diminished by cutting
off a great portion of the mass.

VOL. I. 43

338 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

22. Applying this result to the case of a ring, and putting s for x, and

<r for f we have

(r = A cos MS, and T= AirkA cos »w,

so that -RN=47rk,

P-

when 7/i is very large, and this is the greatest value of N.

The value of L has little effect on the condition of stability. If L and
J/ are both neglected, that condition is

o>'>27-856 (2irk) .............................. (78),

and if L be as much as $N, then

o>3>25-649 (2irk) .............................. (79),

so that it is not important whether we calculate the value of L or not.

The condition of stability is, that the average density must not exceed a
certain value. Let us ascertain the relation between the maximum density of
the ring and that of the planet.

Let 6 be the radius of the planet, that of the ring being unity, then the
mass of Saturn is ^nb'k' = CD* if k' be the density of the planet. If we assume
that the radius of the ring is twice that of the planet, as Laplace has done,
then & = £ and

= 334-2 to 3077 .............................. (80),

so that the density of the ring cannot exceed -g-J^ of that of the planet. Now
Laplace has shewn that if the outer and inner parts of the ring have the same
angular velocity, the ring will not hold together if the ratio of the density of
the planet to that of the ring exceeds 1*3, so that in the first place, our ring
cannot have uniform angular velocity, and in the second place, Laplace's ring
cannot preserve its form, if it is composed of loose materials acting on each
other only by the attraction of gravitation, and moving with the same angular
velocity throughout.

23. On the forces arising from inequalities of thickness in a thin stratum
of fluid of indefinite extent.

The forces which act on any portion of a continuous fluid are of two kinds,
the pressures of contiguous portions of fluid, and the attractions of all portions of
the fluid whether near or distant. In the case of a thin stratum of fluid, not

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 339

acted on by any external forces, the pressures are due mainly to the component
of the attraction which is perpendicular to the plane of the stratum. It is
easy to shew that a fluid acted on by such a force will tend to assume a
position of equilibrium, in which its free surface is plane ; and that any irregu-
larities will tend to equalise themselves, so that the plane surface will be one
of stable equilibrium.

It is also evident, that if we consider only that part of the attraction
which is parallel to the plane of the stratum, we shall find it always directed
towards the thicker parts, so that the effect of this force is to draw the fluid
from thinner to thicker parts, and so to increase irregularities and destroy
equilibrium.

The normal attraction therefore tends to preserve the stability of equilibrium,
while the tangential attraction tends to render equilibrium unstable.

According to the nature of the irregularities one or other of these forces
will prevail, so that if the extent of the irregularities is small, the normal
forces will ensure stability, while, if the inequalities cover much space, the
tangential forces will render equilibrium unstable, and break up the stratum into
l>eads.

To fix our ideas, let us conceive the irregularities of the stratum split up
into the form of a number of systems of waves superposed on one another,
then, by what we have just said, it appears, that very short waves will disap-
pear of themselves, and be consistent with stability, while very long waves will
tend to increase in height, and will destroy the form of the stratum.

In order to determine the law according to which these opposite effects
take place, we must subject the case to mathematical investigation.

Let us suppose the fluid incompressible, and of the density k, and let it
be originally contained between two parallel planes, at distances +c and — c
from that of (xy), and extending to infinity. Let us next conceive a series of
imaginary planes, parallel to the plane of (yz), to be plunged into the fluid
stratum at infinitesimal distances from one another, so as to divide the fluid
into imaginary slices perpendicular to the plane of the stratum.

Next let these planes be displaced parallel to the axis of x according to this
law — that if a: be the original distance of the plane from the origin, and £ its
displacement in the direction of x,

g=A cos rax (81).

43—2

340

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

According to this law of displacement, certain alterations will take place in
the distances between consecutive planes ; but since the fluid is incompressible,
and of indefinite extent in the direction of y, the change of dimension must
occur in the direction of z. The original thickness of the stratum was 2c. Let
its thickness at any point after displacement be 2c + 2£, then we must have

= 2c (82),

or

£= — c -~ =

sinmx ........................ (83).

Let us assume that the increase of thickness 2£ is due to an increase of £
at each surface ; this is necessary for the equilibrium of the fluid between the
imaginary planes.

We have now produced artificially, by means of these planes, a system of

n

waves of longitudinal displacement whose length is and amplitude A ; and

we have found that this has produced a system of waves of normal displace-
ment on each surface, having the same length, with a height =cmA.

In order to determine the forces arising from these displacements, we must,
in the first place, determine the potential function at any point of space, and
this depends partly on the state of the fluid before displacement, and partly
on the displacement itself We have, in all cases —

Within the fluid, p = k; beyond it, p = 0.
Before displacement, the equation is reduced to

Instead of assuming F=0 at infinity, we shall assume F=0 at the origin,
and since in this case all is symmetrical, we have

within the fluid

at the bounding planes

beyond them

dV

^ = — 2irh? ; -— ' = — fakz

dV
V— — Ztrkc' ; -y- = T

dV

F, = 2irkc ( T 2z ± c) ; -7- = T

•(86);

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 341

the upper sign being understood to refer to the boundary at distance 4-c, and
the lower to the boundary at distance — c from the origin.

Having ascertained the potential of the undisturbed stratum, we find that
of the disturbance by calculating the effect of a stratum of density Jc and
thickness £, spread over each surface according to the law of thickness already
found. By supposing the coefficient A small enough, (as we may do in calcu-
lating the displacements on which stability depends), we may diminish the
absolute thickness indefinitely, and reduce the case to that of a mere " super-
ficial density," such as is treated of in the theory of electricity. We have here,
too, to regard some parts as of negative density ; but we must recollect that we
are dealing with the difference between a disturbed and an undisturbed system,
which may be positive or negative, though no real mass can be negative.

Let us for an instant conceive only one of these surfaces to exist, and let
us transfer the origin to it. Then the law of thickness is

(83),

and we know that the normal component of attraction at the surface is the
same as if the thickness had been uniform throughout, so that

dV
^

on the positive side of the surface.
Also, the solution of the equation

d'V _

dx> + dz1'0'

consists of a series of terms of the form Ce'* sin ix.

Of these the only one with which we have to do is that in which i= —m.
Applying the condition as to the normal force at the surface, we get

7= ZirJccfT^A sin mx ........................... (87),

for the potential on the positive side of the surface, and

F=2dkee"Msin«na! .............................. (88),

on the negative side.

] (89);

342 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

Calculating the potentials of a pair of such surfaces at distances +c and — c
from the plane of xy, and calling V the sum of their potentials, we have for
the space between these planes

F,' = ZirkcA sin max-"" («"- + e""*)
beyond them Vt'-2irkcA sin mxe*au(tme + €

the upper or lower sign of the index being taken according as z is positive or
negative.

These potentials must be added to those formerly obtained, to get the
potential at any point after displacement.

We have next to calculate the pressure of the fluid at any point, on the
supposition that the imaginary planes protect each slice of the fluid from the
pressure of the adjacent slices, so that it is in equilibrium under the action of
the forces of attraction, and the pressure of these planes on each side. Now
in a fluid of density k, in equilibrium under forces whose potential is V, we

have always —

dp ,
=

so that if we know that the value of p is pt where that of V is F0, then at
any other point

Now, at the free surface of the fluid, jp = 0, and the distance from the
free surface of the disturbed fluid to the plane of the original surface is £, a
small quantity. The attraction which acts on this stratum of fluid is, in the
first place, that of the undisturbed stratum, and this is equal to 4-irkc, towards
that stratum. The pressure due to this cause at the level of the original
surface will be &irk?ct>, and the pressure arising from the attractive forces due
to the displacements upon this thin layer of fluid, will be small quantities of
the second order, which we neglect. We thus find the pressure when z=c to be,

p, = lirJc'c'mA sin mx.
The potential of the undisturbed mass when z = c is

and the potential of the disturbance itself for the same value of z, is

F/ = 2nkcA sin mx (I + e-"").

ON THE STABILITY OF THE MOTION OF SATUEN's RINGS. 343

So that we find the general value of p at any other point to be

p = 2-rrJf (a* - z2) + ZvfccA sin mx {2cm - 1 - C™" + 1™ (f + c'"12)} . . . (90).

This expression gives the pressure of the fluid at any point, as depending
on the state of constraint produced by the displacement of the imaginary planes.
The accelerating effect of these pressures on any particle, if it were allowed to
move parallel to x, instead of being confined by the planes, would be

_ldp
~k dx'

The accelerating effect of the attractions in the same direction is

dV

dx' /

so that the whole acceleration parallel to a? is

X= -ZirkmcA cosmz (2mc-e"smc- 1) .................. (91).

It is to be observed, that this quantity is independent of z, so that every
particle in the slice, by the combined effect of pressure and attraction, is urged
with the same force, and, if the imaginary planes were removed, each slice
would move parallel to itself without distortion, as long as the absolute dis-
placements remained small. We have now to consider the direction of the
resultant force X, and its changes of magnitude.

We must remember that the original displacement is A cos mx, if therefore
(2mc — c"2™0— 1) be positive, X will be opposed to the displacement, and the
equilibrium will be stable, whereas if that quantity be negative, X will actr
along with the displacement and increase it, and so constitute an unstable
condition.

It may be seen that large values of me give positive results and small
ones negative. The sign changes when

2,mc = 1-147 .................................... (92),

which corresponds to a wave-length

o_

X = 2c = 20(5-471) ........................... (93).

The length of the complete wave in the critical case is 5 '471 times the
thickness of the stratum. Waves shorter than this are stable, longer waves
are unstable.

344 ON THE STABILITY OP THE MOTION OF SATURN'S RINGS.

The quantity 2mc (2mc - c"*1" — 1),

has a minimum when 27nc='607 (94),

and the wave-length is 10*353 times the thickness of the stratum.

In this case 2roc (2mc — €'"""— 1)= -'509 (95),

and X= f50QirkA cos mx (96).

24. Let us now conceive that the stratum of fluid, instead of being infinite
in extent, is limited in breadth to about 100 times the thickness. The pressures
and attractions will not be much altered by this removal of a distant part of
the stratum. Let us also suppose that this thin but broad strip is bent round
in its own plane into a circular ring whose radius is more than ten times the
breadth of the strip, and that the waves, instead of being exactly parallel to
each other, have their ridges in the direction of radii of the ring. We shall
then have transformed our stratum into one of Saturn's Rings, if we suppose
those rings to be liquid, and that a considerable breadth of the ring has the
same angular velocity.

Let us now investigate the conditions of stability by putting

x = - 2irkmc(2mc - e"Jme - 1)

into the equation for n. We know that x must lie between 0 and — - to

1 o J

ensure stability. Now the greatest value of x in the fluid stratum is •509rr/:.
Taking Laplace's ratio of the diameter of the ring to that of the planet, this
gives 42 -5 as the minimum value of the density of the planet divided by that
of the fluid of the ring.

Now Laplace has shewn that any value of this ratio greater than 1'3 is
inconsistent with the rotation of any considerable breadth of the fluid at the
same angular velocity, so that our hypothesis of a broad ring with uniform
velocity is untenable.

But the stability of such a ring is impossible for another reason, namely,
that for waves in which 2wc> T147, x is negative, and the ring will be destroyed
by these short waves in the manner described at page (333).

When the fluid ring is treated, not as a broad strip, but as a filament of
circular or elliptic section, the mathematical difficulties are very much increased,

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 345

but it may be shown that in this case also there will be a maximum value
of x, which will require the density of the planet to be several times that of
the ring, and that in all cases short waves will give rise to negative values
of x, inconsistent with the stability of the ring.

It appears, therefore, that a ring composed of a continuous liquid mass
cannot revolve about a central body without being broken up, but that the
parts of such a broken ring may, under certain conditions, form a permanent
ring of satellites.

On the Mutual Perturbations of Two Rings.

25. We shall assume that the difference of the mean radii of the rings
is small compared with the radii themselves, but large compared with the
distance of consecutive satellites of the same ring. We shall also assume that
each ring separately satisfies the conditions of stability.

We have seen that the effect of a disturbing force on a ring is to produce
a series of waves whose number and period correspond with those of the dis-
turbing force which produces them, so that we have only to calculate the
coefficient belonging to the wave from that of the disturbing force.

Hence in investigating the simultaneous motions of two rings, we may
assume that the mutually disturbing waves travel with the same absolute
angular velocity, and that a maximum in one corresponds either to a maximum
or a minimum of the other, according as the coefficients have the same or
opposite signs.

Since the motions of the particles of each ring are affected by the disturbance
of the other ring, as well as of that to which they belong, the equations of
motion of the two rings will be involved in each other, and the final equation
for determining the wave-velocity will have eight roots instead of four. But as
each of the rings has four free waves, we may suppose these to originate forced
waves in the other ring, so that we may consider the eight waves of each ring
as consisting of four free waves and four forced ones.

In strictness, however, the wave- velocity of the "free" waves will be
affected by the existence of the forced waves which they produce in the other
ring, so that none of the waves are really " free " in either ring independently,
though the whole motion of the system of two rings as a whole is free.

VOL. i. 44

346 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

We shall find, however, that it is best to consider the waves first as free,
and then to determine the reaction of the other ring upon them, which is such
as to alter the wave-velocity of both, as we shall see.

The forces due to the second ring may be separated into three parts.

1st The constant attraction when both rings are at rest.

2nd. The variation of the attraction on the first ring, due to its own
disturbances.

3rd. The variation of the attraction due to the disturbances of the second
ring.

The first of these affects only the angular velocity. The second affects the
waves of each ring independently, and the mutual action of the waves depends
entirely on the third class of forces.

26. To determine the attractions between two rings.

Let R and a be the mass and radius of the exterior ring, R and a' those
of the interior, and let all quantities belonging to the interior ring be marked
with accented letters. (Fig. 5.)

1st. Attraction between the rings when at rest.

Since the rings are at a distance small compared with their radii, we may
calculate the attraction on a particle of the first ring as if the second were an
infinite straight line at distance a' — a from the first.

TV

The mass of unit of length of the second ring is - — ,, and the accelerating

effect of the attraction of such a filament on an element of the first ring is

jy

— —. - A inwards .............................. (97).

ira (a - a)

The attraction of the first ring on the second may be found by transposing
accented and unaccented letters.

In consequence of these forces, the outer ring will revolve faster, and the

inner ring slower than would otherwise be the case. These forces enter into

the constant terms of the equations of motion, and may be included in the
value of K.

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 347

2nd. Variation due to disturbance of first ring.

If we put a(l+p) for a in the last expression, we get the attraction
when the first ring is displaced. The part depending on p is

-- —, - ?rs P inwards ........................ (98).

ira (a — ay1

This is the only variation of force arising from the displacement of the
first ring. It affects the value of L in the equations of motion.

3rd. Variation due to waves in the second ring.

On account of the waves, the second ring varies in distance from the
first, and also in mass of unit of length, and each of these alterations produces
variations both in the radial and tangential force, so that there are four things
to be calculated :

1st. Radial force due to radial displacement.

2nd. Radial force due to tangential displacement.

3rd. Tangential force due to radial displacement.

4th. Tangential force due to tangential displacement.

1st. Put a'(l+p') for a', and we get the term in p'

-7

2nd. By the tangential displacement of the second ring the section is
iced hi the proportion
of the radial force equal to

reduced in the proportion of 1 to 1 — -, , , and therefore there is an alteration

— inwards = —/A' -1-7 say (100).

ira' (a -a') ds' ' ds'

3rd. By the radial displacement of the second ring the direction of the
filament near the part in question is altered, so that the attraction is no longer
radial but forwards, and the tangential part of the force is

''&' forwards ................. (101).

——x

TTO, (a — a) ds ds

44—2

348 ON THE STABILITY OF THE MOTION OF SATURN'S

4th. By the tangential displacement of the second ring a tangential force
arises, depending on the relation between the length of the waves and the
distance between the rings.

a— a [+°° xsinpx .

If we make m - —,— =p, and TO I — -f-j ax = U,
a, J - » ( 1 + ar)

TV

the tangential force is ,, — T\* H<r=v<r' .......... .............. (102).

TTCt ^Ct — ft j

We may now write down the values of X, ^ and v by transposing accented
and unaccented letters.

R (2a-a) R R ,.._.

\=- —Td-; p. = — T-, - \; v- — ,-, -- /\,n ......... (103).

va (a — ay ira(a—a) va(a—ay

Comparing these values with those of X', p', and v, it will be seen that
the following relations are approximately true when a is nearly equal to a':

X" n~v

27. To form the equations of motion.
*The original equations were

'

Putting p = A cos (ms + nt), a- = Bain (ms + nt),

p = A' cos (ms + nt), <r' = I? sin (ms + nt),

then o>'

. .
'

The corresponding equations for the second ring may be found by trans-
posing accented and unaccented letters. We should then have four equations
to determine the ratios of A, Bt A', R, and a resultant equation of the eighth
degree to determine n. But we may make use of a more convenient method,
since X', //, and v are small. Eliminating B we find

_
- \'A' + n'm&) n' + (p'mA' - v'B1} 2omJ =

* [The analysis in this article is somewhat unsatisfactory, the equations of motion employed being
those which were applicable in the case of a ring of radius unity. ED.]

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 349

Putting B = $A, A' = xA, K =

we have n4 - {a? ( + 2K) + L - N} ri> - 4col/rc + BtfN\ _ n _ . .

' 'p) n'x + (//m - v'P] 2o>nx) =

- . . ................. (108),

dn

i (109),

I*MI

dn \'n — u.'mB'n — 2 u'mw + 2 v'B'a /-.-•« \

whence -7-=- - (110).

cte 4rr — 2w

28. If we were to solve the equation for n, leaving out the terms involving
x, we should find the wave-velocities of the four free waves of the first ring,
supposing the second ring to be prevented from being disturbed. But in reality
the waves in the first ring produce a disturbance in the second, and these in
turn react upon the first ring, so that the wave-velocity is somewhat different
from that which it would be in the supposed case. Now if x be the ratio
of the radial amplitude of displacement in the second ring to that in the first,
and if ft be a value of n supposing x = 0, then by Maclaurin's theorem,

n- 4-n4- — x (III}

/fr ^— T^ /fr T^ 7 •(/••••••••••••••••••••••••••••••••l-L-L.Lf.

dx

The wave- velocity relative to the ring is , and the absolute . angular

velocity of the wave in space is

n n I dn t -\-\v\

m mm dx

= +p-qx (113),

n , I dn

where » = &> , and <7 = — -r-.

m m dx

Similarly in the second ring we should have

*'=p'-q-
x

and since the corresponding waves in the two rings must have the same abso-
lute angular velocity,

(114);

or p — qx=p' — q' -

350 ON THE STABILITY OP THE MOTION OP SATURN'S RINGS.

This is a quadratic equation in x, the roots of which are real when

(p-p'Y+W

is positive. When this condition is not fulfilled, the roots are impossible, and
the general solution of the equations of motion will contain exponential factors,
indicating destructive oscillations in the rings.

Since q and q' are small quantities, the solution is always real whenever
p and p' are considerably different. The absolute angular velocities of the two
pairs of reacting waves, are then nearly

p —p p —p

instead of p and p1, as they would have been if there had been no reaction
of the forced wave upon the free wave which produces it.

When p and p' are equal or nearly equal, the character of the solution
will depend on the sign of qq'. We must therefore determine the signs of q
and q' in such cases.

Putting f? = —,-, we may write the values of q and q'

^ m

n

(lie):

. /<u <<A <a w

, X + 2u.m -, - 4i> — -

, n \n n) n n

2 m" 4n"-2a>'2

Referring to the values of the disturbing forces, we find that

X' fif v Ra
X p, v Ra''

q n 4n" — 2o/" Ra

Since qq' is of the same sign as —, , we have only to determine whether

2n -- , and 2n' -- , are of the same or of different signs. If these quantities
are of the same sign, qq' is positive, if of different signs, qq' is negative.

ON THE STABILITY OF THE MOTION OP SATURN'S KINGS. 351

Now there are four values of n, which give four corresponding values of

a.'

In

n

nj= — a» + &c., 2«,— is negative,

"i

w, = — a small quantity, 2n3 -- is positive,

«, = + a small quantity, 2n, -- is negative,

n4 = ft> — &c., 2n4 -- is positive.

nt

The quantity with which we have to do is therefore positive for the even
orders of waves and negative for the odd ones, and the corresponding quantity
in the other ring obeys the same law. Hence when the waves which act upon
each other are either both of even or both of odd names, qq' will be positive,
but when one belongs to an even series, and the other to an odd series, qq'
is negative.

29. The values of p and p' are, roughly,

(118)-

Pi = ft) H &C., p. = ft) + &C., pt = ft) — &C., pt= (O 1- &C.

/ /

7>1' = ft)'H &c., »,' = ft)' + &c., »,' = ft/ — &c., »/ = &/ H&c.

m m

ft)' is greater than <a, so that j9t' is the greatest, and pt the least of these
values, and of those of the same order, the accented is greater than the unac-
cented. The following cases of equality are therefore possible under suitable
circumstances :

P*=Pt' (when m=l), pt=p,',

In the cases in the first column qq' will be positive, in those in the second
column qq' will be negative.

352 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

30. Now each of the four values of p is a function of m, the number
of undulations in the ring, and of a the radius of the ring, varying nearly
as a"'. Hence m being given, we may alter the radius of the ring till any
one of the four values of p becomes equal to a given quantity, say a given
value of p', so that if an indefinite number of rings coexisted, so as to form
a sheet of rings, it would be always possible to discover instances of the
equality of p and p' among them. If such a case of equality belongs to the
first column given above, two constant waves will arise in both rings, one
travelling a little faster, and the other a little slower than the free waves.
If the case belongs to the second column, two waves will also arise in each
ring, but the one pair will gradually die away, and the other pair will increase
in amplitude indefinitely, the one wave strengthening the other till at last both
rings are thrown into confusion.

The only way in which such an occurrence can be avoided is by placing
the rings at such a distance that no value of m shall give coincident values
of p and p1. For instance, if &/ > 2o», but <a < 3<u, no such coincidence is possible.
For pl is always less than pf, it is greater than pt when m = 1 or 2, and less
than pi when m is 3 or a greater number. There are of course an infinite
number of ways in which this noncoincidence might be secured, but it is plain
that if a number of concentric rings were placed at small intervals from each
other, such coincidences must occur accurately or approximately between some
pairs of rings, and if the value of (p—p'Y is brought lower than —4qq', there
will be destructive interference.

This investigation is applicable to any number of concentric rings, for, by
the principle of superposition of small displacements, the reciprocal actions of
any pair of rings are independent of all the rest.

31. On the effect of long-continued disturbances on a system of rings.

The result of our previous investigations has been to point out several
ways in which disturbances may accumulate till collisions of the different par-
ticles of the rings take place. After such a collision the particles will still
continue to revolve about the planet, but there will be a loss of energy in
the system during the collision which can never be restored. Such collisions
however will not affect what is called the Angular Momentum of the system
about the planet, which will therefore remain constant.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 353

Let M be the mass of the system of rings, and Sm that of one ring
whose radius is r, and angular velocity a> = $*r~*. The angular momentum of
the ring is

half its vis viva is ^coVSm = ^Sr'1 8m.

The potential energy due to Saturn's attraction on the ring is

-Sr-i&m.

The angular momentum of the whole system is invariable, and is

S*2 (r*Sm) = A .............................. (119).

/
The whole energy of the system is the sum of half the vis viva and the

potential energy, and is

E ........................... (120).

A is invariable, while E necessarily diminishes. We shall find that as E
diminishes, the distribution of the rings must be altered, some of the outer
rings moving outwards, while the inner rings move inwards, so as either to
spread out the whole system more, both on the outer and on the inner edge
of the system, or, without affecting the extreme rings, to diminish the density
or number of the rings at the mean distance, and increase it at or near the
inner and outer edges.

Let us put x = i*, then A = S^t (xdm) is constant.

-T 2 (xdm)

Now let BI = -*/J /,

2 (am)

and x = #! + x't

then we may write

= 1dm (Vs - 2 -j+ 8 -7-&C.),

.Cj .//!

= -^ 2 (dm) -^2 (x'dm) + —t2 (»*8»i) -&c ..... (121).

X, Xl Xl

VOL. I. 45

354 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

Now 2 (dm) = M a constant, 2 (x'dm) = 0, and 2 (x^Sm) is a quantity which
increases when the rings are spread out from the mean distance either way,
x' being subject only to the restriction 2 (x'dm) = 0. But 2 (x^dm) may
increase without the extreme values of x' being increased, provided some other
values be increased.

32. In fact, if we consider the very innermost particle as moving in an
ellipse, and at the further apse of its orbit encountering another particle
belonging to a larger orbit, we know that the second particle, when at the
same distance from the planet, moves the faster. The result is, that the
interior satellite will receive a forward impulse at its further apse, and will
move in a larger and less eccentric orbit than before. In the same way one
of the outermost particles may receive a backward impulse at its nearer apse,
and so be made to move in a smaller and less eccentric orbit than before.
When we come to deal with collisions among bodies of unknown number, size,
and shape, we can no longer trace the mathematical laws of their motion with
any distinctness. All we can now do is to collect the results of our investi-
gations and to make the best use we can of them in forming an opinion as
to the constitution of the actual rings of Saturn which are still in existence
and apparently in steady motion, whatever catastrophes may be indicated by
the various theories we have attempted.

33. To find the Loss of Energy due to interned f notion in a broad Fix id
Ring, the parts of which revolve about the Planet, each with the velocity of a
satellite at the same distance.

Conceive a fluid, the particles of which move parallel to the axis of x
with a velocity u, u being a function of z, then there will be a tangential pres-
sure on a plane parallel to xy

du ., f

= u. -j- on unit of area

r dz

due to the relative sliding of the parts of the fluid over each other.

In the case of the ring we have

w = 5*r-1.
The absolute velocity of any particle is <ar. That of a particle at distance

(r + 8r) is

d

- (wr)8r.

ON THE STABILITY OF THE MOTION OF SATUKN's RINGS. 355

If the angular velocity had been uniform, there would have been no sliding,
and the velocity would have been

<ar + <a8r.
The sliding is therefore

d<» ~
r -7- Sr,
dr

and the friction on unit of area perpendicular to r is pr -y- .

The loss of Energy, per unit of area, is the product of the sliding by the
friction,

or,

dr

Sr in unit of time.

The loss of Energy in a part of the Ring whose radius is r, breadth
Sr, and thickness c, is

dear R

27rrcu -j- or.
dr\

In the case before us it is •§ Tr/i^Scr"1 Sr.

If the thickness of the ring is uniform between r = a and r = b, the whole
loss of Energy is

in unit of time.

Now half the vis viva of an elementary ring is

irpcrBr 1*0)* = npcSBr,

and this between the limits r — a and r = b gives

irpcS (a — 6).

The potential due to the attraction of S is twice this quantity with the
sign changed, so that

E= —irpcS(a — b),

dE /I 1

and

1 dE_ _ p !_
E dt = " * p ab '

45—2

356 ON THE STABILITY OP THE MOTION OF SATURN'S RINGS.

Now Professor Stokes finds */- = 0'0564 for water,

' P

and =0'116 for air,

taking the unit of space one English inch, and the unit of time one second.
We may take a = 88,209 miles, and 6 = 77,636 for the ring A; and a = 75,845,
and 6 = 58,660 for the ring B. We may also take one year as the unit of
time. The quantity representing the ratio of the loss of energy in a year to
the whole energy is

l_d_E = 1 .

E dt "60,880,000,000,000

1
811(1 39,540,000,000,000 fo1

showing that the effect of internal friction in a ring of water moving with
steady motion is inappreciably small. It cannot be from this cause therefore
that any decay can take place in the motion of the ring, provided that no
waves arise to disturb the motion.

Recapitulation of the Theory of the Motion of a Rigid Ring.

The position of the ring relative to Saturn at any given instant is defined
by three variable quantities.

1st. The distance between the centre of gravity of Saturn and the centre
of gravity of the ring. This distance we denote by r.

2nd. The angle which the line r makes with a fixed line in the plane of
the motion of the ring. This angle is called 6.

3rd. The angle between the line r and a line fixed with respect to the
ring so that it coincides with r when the ring is in its mean position. This is
the angle <j>.

The values of these three quantities determine the position of the ring so
far as its motion in its own plane is concerned. They may be referred to as
the radius vector, longitude, and angle of libration of the ring.

The forces which act between the ring and the planet depend entirely upon
their relative positions. The method adopted above consists in determining the

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 357

potential (V) of the ring at the centre of the planet in terms of r and <$>. Then
the work done by any displacement of the system is measured by the change
of VS during that displacement. The attraction between the centre of gravity

dV
of the Ring and that of the planet is —S-j—, and the moment of the couple

dV
tending to turn the ring about its centre of gravity is S-j-r.

It is proved in Problem V, that if a be the radius of a circular ring, r^ = af
the distance of its centre of gravity from the centre of the circle, and R the

mass of the ring, then, at the centre of the ring, -,- = -- -„/, -TJ = 0-

CtT* Ot Ct(b

It also appears that •r^ = ^—, , (1 +5r)» which is positive when g > — 1,

Cb

7JTT J3

and that -y-r-2 = £ — f (3 — g), which is positive when g<3.

If -j-j- is positive, then the attraction between the centres decreases as the

distance increases, so that, if the two centres were kept at rest at a given

d'V
distance by a constant force, the equilibrium would be unstable. If -5-75 is positive,

then the forces tend to increase the angle of libration, in whichever direction
the libration takes place, so that if the ring were fixed by an axis through its
centre of gravity, its equilibrium round that axis would be unstable.

In the case of the uniform ring with a heavy particle on its circumference
whose weight = '82 of the whole, the direction of the whole attractive force of
the ring near the centre will pass through a point lying in the same radius as
the centre of gravity, but at a distance from the centre = fa. (Fig. 6.)

If we call this point O, the line SO will indicate the direction and position
of the force acting on the ring, which we may call F.

It is evident that the force F, acting on the ring in the line OS, will tend
to turn it round its centre of gravity R and to increase the angle of libration
KRO. The direct action of this force can never reduce the angle of libration
to zero again. To understand the indirect action of the force, we must recollect
that the centre of gravity (R) of the ring is revolving about Saturn in the
direction of the arrows, and that the ring is revolving about its centre of gravity

358 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

with nearly the same velocity. If the angular velocity of the centre of gravity
about Saturn were always equal to the rotatory velocity of the ring, there
would be no libration.

Now suppose that the angle of rotation of the ring is in advance of the
longitude of its centre of gravity, so that the line RO has got in advance of
SRK by the angle of libration KRO. The attraction between the planet and
the ring is a force F acting in SO. We resolve this force into a couple, whose
moment is F'RN, and a force F acting through R the centre of gravity of the
ring.

The couple affects the rotation of the ring, but not the position of its centre
of gravity, and the force RF acts on the centre of gravity without affecting the
rotation.

Now the couple, in the case represented in the figure, acts in the positive
direction, so as to increase the angular velocity of the ring, which was already
greater than the velocity of revolution of R about S, so that the angle of
libration would increase, and never be reduced to zero.

The force RF does not act in the direction of S, but behind it, so that it
becomes a retarding force acting upon the centre of gravity of the ring. Now
the effect of a retarding force is to cause the distance of the revolving body to
decrease and the angular velocity to increase, so that a retarding force increases
the angular velocity of R about S.

The effect of the attraction along SO in the case of the figure is, first, to
increase the rate of rotation of the ring round R, and secondly, to increase the
angular velocity of R about & If the second effect is greater than the first,
then, although the line RO increases its angular velocity, SR will increase its
angular velocity more, and will overtake RO, and restore the ring to its original
position, so that SRO will be made a straight line as at first. If this accelerat-
ing effect is not greater than the acceleration of rotation about R due to the
couple, then no compensation will take place, and the motion will be essentially
unstable.

If in the figure we had drawn <£ negative instead of positive, then the
couple would have been negative, the tangential force on R accelerative, r would
have increased, and in the cases of stability the retardation of 6 would be greater
than that of (6 +<£), and the normal position would be restored, as before.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 359

The object of the investigation is to find the conditions under which this
compensation is possible.

It is evident that when SRO becomes straight, there is still a difference
of angular velocities between the rotation of the ring and the revolution of
the centre of gravity, so that there will be an oscillation on the other side,
and the motion will proceed by alternate oscillations without limit.

If we begin with r at its mean value, and <f> negative, then the rotation
of the ring will be retarded, r will be increased, the revolution of r will be
more retarded, and thus <f> will be reduced to zero. The next part of the
motion will reduce r to its mean value, and bring <f> to its greatest positive
value. Then r will diminish to its least value, and <f> will vanish. Lastly r
will return to the mean value, and <f> to the greatest negative value.

It appears from the calculations, that there are, in general, two different
ways in which this kind of motion may take place, and that these may have
different periods, phases, and amplitudes. The mental exertion required in follow-
ing out the results of a combined motion of this kind, with all the variations of
force and velocity during a complete cycle, would be very great in proportion to
the additional knowledge we should derive from the exercise.

The result of this theory of a rigid ring shows not only that a perfectly
uniform ring cannot revolve permanently about the planet, but that the irregu-
larity of a permanently revolving ring must be a very observable quantity, the
distance between the centre of the ring and the centre of gravity being between
•8158 and '8279 of the radius. As there is no appearance about the rings
justifying a belief in so great an irregularity, the theory of the solidity of the
rings becomes very improbable.

When we come to consider the additional difficulty of the tendency of the
fluid or loose parts of the ring to accumulate at the thicker parts, and thus
to destroy that nice adjustment of the load on which stability depends, we
have another powerful argument against solidity.

And when we consider the immense size of the rings, and their comparative
thinness, the absurdity of treating them as rigid bodies becomes self-evident.
An iron ring of such a size would be not only plastic but semifluid under the
forces which it would experience, and we have no reason to believe these rings
to be artificially strengthened with any material unknown on this earth.

360 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

Recapitulation of tfie Theory of a Ring of equal Satellites.

In attempting to conceive of the disturbed motion of a ring of unconnected
satellites, we have, in the first place, to devise a method of identifying each
satellite at any given time, and in the second place, to express the motion of
every satellite under the same general formula, in order that the mathematical
methods may embrace the whole system of bodies at once.

By conceiving the ring of satellites arranged regularly in a circle, we may
easily identify any satellite, by stating the angular distance between it and a
known satellite when so arranged. If the motion of the ring were undisturbed,
this angle would remain unchanged during the motion, but, in reality, the
satellite has its position altered in three ways : 1st, it may be further from
or nearer to Saturn; 2ndly, it may be in advance or in the rear of the position
it would have had if undisturbed ; 3rdly, it may be on one side or other of
the mean plane of the ring. Each of these displacements may vary in any way
whatever as we pass from one satellite to another, so that it is impossible
to assign beforehand the place of any satellite by knowing the places of the
rest. § 2.

The formula, therefore, by which we are enabled to predict the place of
every satellite at any given time, must be such as to allow the initial position
of every satellite to be independent of the rest, and must express all future
positions of that satellite by inserting the corresponding value of the quantity
denoting time, and those of every other satellite by inserting the value of the
angular distance of the given satellite from the point of reference. The three
displacements of the satellite will therefore be functions of two variables — the
angular position of the satellite, and the time. When the time alone is made
to vary, we trace the complete motion of a single satellite ; and when the time
is made constant, and the angle is made to vary, we trace the form of the
ring at a given time.

It is evident that the form of this function, in so far as it indicates the
state of the whole ring at a given instant, must be wholly arbitrary, for the
form cf the ring and its motion at starting are limited only by the condition
that the irregularities must be small. We have, however, the means of breaking
up any function, however complicated, into a series of simple functions, so that
the value of the function between certain limits may be accurately expressed

ON THE STABILITY OF THE MOTION OF SATURN'S KINGS. 361

as the sum of a series of sines and cosines of multiples of the variable. This
method, due to Fourier, is peculiarly applicable to the case of a ring returning
into itself, for the value of Fourier's series is necessarily periodic. We now
regard the form of the disturbed ring at any instant as the result of the
superposition of a number of separate disturbances, each of which is of the nature
of a series of equal waves regularly arranged round the ring. Each of these
elementary disturbances is characterised by the number of undulations in it, by
their amplitude, and by the position of the first maximum in the ring. § 3.

When we know the form of each elementary disturbance, we may calculate
the attraction of the disturbed ring on any given particle in terms of the con-
stants belonging to that disturbance, so that as the actual displacement is the
resultant of the elementary displacements, the actual attraction will be the
resultant of the corresponding elementary attractions, and therefore the actual
motion will be the resultant of all the motions arising from the elementary
disturbances. We have therefore only to investigate the elementary disturbances
one by one, and having established the theory of these, we calculate the actual
motion by combining the series of motions so obtained.

Assuming the motion of the satellites in one of the elementary disturbances
to be that of oscillation about a mean position, and the whole motion to be
that of a uniformly revolving series of undulations, we find our supposition to
be correct, provided a certain biquadratic equation is satisfied by the quantity
denoting the rate of oscillation. § 6.

When the four roots of this equation are all real, the motion of each
satellite is compounded of four different oscillations of different amplitudes and
periods, and the motion of the whole ring consists of four series of undulations,
travelling round the ring with different velocities. When any of these roots
are impossible, the motion is no longer oscillatory, but tends to the rapid
destruction of the ring.

To determine whether the motion of the ring is permanent, we must assure
ourselves that the four roots of this equation are real, whatever be the number
of undulations in the ring; for if any one of the possible elementary disturb-
ances should lead to destructive oscillations, that disturbance might sooner or
later commence, and the ring would be destroyed.

Now the number of undulations in the ring may be any whole number
from one up to half the number of satellites. The forces from which danger
VOL. I. 46

362 ON THE STABILITY OF THE MOTION OF SATURN S RINGS.

is to be apprehended are greatest when the number of undulations is greatest,
and by taking that number equal to half the number of satellites, we find the
condition of stability to be

where S is the mass of the central body, R that of the ring, and p the number
of satellites of which it is composed. § 8. If the number of satellites be too
great, destructive oscillations will commence, and finally some of the satellites
will come into collision with each other and unite, so that the number of
independent satellites will be reduced to that which the central body can retain
and keep in discipline. When this has taken place, the satellites will not only
be kept at the proper distance from the primary, but will be prevented by its
preponderating mass from interfering with each other.

We next considered more carefully the case in which the mass of the ring
is very small, so that the forces arising from the attraction of the ring are
small compared with that due to the central body. In this case the values
of the roots of the biquadratic are all real, and easily estimated. § 9.

If we consider the motion of any satellite about its mean position, as
referred to axes fixed in the plane of the ring, we shall find that it describes
an ellipse in the direction opposite to that of the revolution of the ring, the
periodic time being to that of the ring as <a to n, and the tangential ampli-
tude of oscillation being to the radial as 2o> to n. § 10.

The absolute motion of each satellite in space is nearly elliptic for the large
values of n, the axis of the ellipse always advancing slowly in the direction of
rotation. The path of a satellite corresponding to one of the small values of
n is nearly circular, but the radius slowly increases and diminishes during a
period of many revolutions. §11.

The form of the ring at any instant is that of a re-entering curve, having
m alternations of distance from the centre, symmetrically arranged, and m points
of condensation, or crowding of the satellites, which coincide with the points of
greatest distance when n is positive, and with the points nearest the centre
when n is negative. § 12.

71

This system of undulations travels with an angular velocity -- relative to

71

the ring, and at — - in space, so that during each oscillation of a satellite a

771

complete wave passes over it. § 14.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 363

To exhibit the movements of the satellites, I have made an arrangement
J»y which 36 little ivory balls are made to go through the motions belonging
to the first or fourth series of waves. (Figs. 7, 8.)

The instrument stands on a pillar A, in the upper part of which turns
the cranked axle (7(7. On the parallel parts of this axle are placed two wheels,
RR and TT, each of which has 36 holes at equal distances in a circle near
its circumference. The two circles are connected by 36 small cranks of the
form KK, the extremities of which turn in the corresponding holes of the two
wheels. That axle of the crank K which passes through the hole in the wheel
S is bored, so as to hold the end of the bent wire which carries the satellite S.
This wire may be turned in the hole so as to place the bent part carrying
the satellite at any angle with the crank. A pin P, which passes through the
top of the pillar, serves to prevent the cranked axle from turning ; and a pin Q,
passing through the pillar horizontally, may be made to fix the wheel R, by
inserting it in a hole in one of the spokes of that wheel. There is also a
handle //, which is in one piece with the wheel T, and serves to turn the axle.

Now suppose the pin P taken out, so as to allow the cranked axle to
turn, and the pin Q inserted in its. hole, so as to prevent the wheel R from
revolving ; then if the crank C be turned by means of the handle //, the
wheel T will have its centre carried round in a vertical circle, but will remain
parallel to itself during the whole motion, so that every point in its plane will
describe an equal circle, and all the cranks K will be made to revolve exactly
as the large crank C does. Each satellite will therefore revolve in a small
circular orbit, in the same time with the handle H, but the position of each
satellite in that orbit may be arranged as we please, according as we turn the
wire which supports it in the end of the crank.

In fig. 8, which gives a front view of the instrument, the satellites are so
placed that each is turned 60° further round in its socket than the one behind
it. As there are 36 satellites, this process will bring us back to our starting-
point after six revolutions of the direction of the arm of the satellite ; and
therefore as we have gone round the ring once in the same direction, the arm
of the satellite will have overtaken the radius of the ring five times.

Hence there will be five places where the satellites are beyond their mean
distance from the centre of the ring, and five where they are within it, so
that we have here a series of five undulations round the circumference of the

46—2

3G4 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

ring. In this case the satellites are crowded together when nearest to the centre,
so that the case is that of the first series of waves, when m = 5.

Now suppose the cranked axle C to be turned, and all the small cranks
K to turn with it, as before explained, every satellite will then be carried
round on its own arm in the same direction ; but, since the direction of the
arms of different satellites is different, their phases of revolution will preserve
the same difference, and the system of satellites will still be arranged in five
undulations, only the undulations will be propagated round the ring in the
direction opposite to that of the revolution of the satellites.

To understand the motion better, let us conceive the centres of the orbits
of the satellites to be arranged in a straight line instead of a circle, as in
fig. 10. Each satellite is here represented in a different phase of its orbit, so
that as we pass from one to another from left to right, we find the position
of the satellite in its orbit altering in the direction opposite to that of the
hands of a watch. The satellites all lie in a trochoidal curve, indicated by
the line through them in the figure. Now conceive every satellite to move in
its orbit through a certain angle in the direction of the arrows. The satellites
will then lie in the dotted line, the form of which is the same as that of
the former curve, only shifted in the direction of the large arrow. It appears,
therefore, that as the satellites revolve, the undulation travels, so that any
part of it reaches successively each satellite as it comes into the same phase
of -rotation. It therefore travels from those satellites which are most advanced
in phase to those which are less so, and passes over a complete wave-length
in the time of one revolution of a satellite.

Now if the satellites be arranged as in fig. 8, where each is more advanced
in phase as we go round the ring in the direction of rotation, the wave will
travel in the direction opposite to that of rotation, but if they are arranged
as in fig. 12, where each satellite is less advanced in phase as we go round
the ring, the wave will travel in the direction of rotation. Fig. 8 represents
the first series of waves where m = 5, and fig. 12 represents the fourth series
where m = 7. By arranging the satellites in their sockets before starting, we
might make m equal to any whole number, from 1 to 18. If we chose any
number above 18 the result would be the same as if we had taken a number
as much below 18 and changed the arrangement from the first wave to the
fourth.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 365

In this way we can exhibit the motions of the satellites in the first and
fourth waves. In reality they ought to move in ellipses, the major axes being
twice the minor, whereas in the machine they move in circles : but the character
of the motion is the same, though the form of the orbit is different.

We may now show these motions of the satellites among each other, com-
bined with the motion of rotation of the whole ring. For this purpose we
put in the pin P, so as to prevent the crank axle from turning, and take
out the pin Q so as to allow the wheel R to turn. If we then turn the
wheel T, all the small cranks will remain parallel to the fixed crank, and the
wheel R will revolve at the same rate as T. The arm of each satellite will
continue parallel to itself during the motion, so that the satellite will describe
a circle whose centre is at a distance from the centre of R, equal to the arm
of the satellite, and measured in the same direction. In our theory of real
satellites, each moves in an elh'pse, having the central body in its focus, but
this motion in an eccentric circle is sufficiently near for illustration. The
motion of the waves relative to the ring is the same as before. The waves
of the first kind travel faster than the ring itself, and overtake the satellites,
those of the fourth kind travel slower, and are overtaken by them.

In fig. 11 we have an exaggerated representation of a ring of twelve satel-
lites affected by a wave of the fourth kind where m = 2. The satellites here lie in
an elh'pse at any given instant, and as each moves round in its circle about
its mean position, the ellipse also moves round in the same direction with half
their angular velocity. In the figure the dotted line represents the position of
the ellipse when each satellite has moved fonvard into the position represented
by a dot.

Fig. 13 represents a wave of the first kind where m=2. The satellites at
any instant lie in an epitrochoid, which, as the satellites revolve about their
mean positions, revolves in the opposite direction with half their angular velocity,
so that when the satellites come into the positions represented by the dots,
the curve in which they lie turns round hi the opposite direction and forms the
dotted curve.

In fig. 9 we have the same case as in fig. 13, only that the absolute orbits
of the satellites in space are given, instead of their orbits about their mean
positions in the ring. Here each moves about the central body in an eccentric

366 ON THE STABILITY OF THE MOTION OP SATURN'S RINGS.

circle, which in strictness ought to be an ellipse not differing much from the
circle.

As the satellites move in their orbits in the direction of the arrows, the
curve which they form revolves in the same direction with a velocity 1£ times
that of the ring.

By considering these figures, and still more by watching the actual motion
of the ivory balls in the model, we may form a distinct notion of the motions
of the particles of a discontinuous ring, although the motions of the model are
circular and not elliptic. The model, represented on a scale of one-third in figs.
7 and 8, was made in brass by Messrs. Smith and Ramage of Aberdeen.

We are now able to understand the mechanical principle, on account of
which a massive central body is enabled to govern a numerous assemblage of
satellites, and to space them out into a regular ring ; while a smaller central
body would allow disturbances to arise among the individual satellites, and
collisions to take place.

When we calculated the attractions among the satellites composing the ring,
we found that if any satellite be displaced tangentially, the resultant attraction
will draw it away from its mean position, for the attraction of the satellites it
approaches will increase, while that of those it recedes from will diminish, so that
its equilibrium when in the mean position is unstable with respect to tangential
displacements ; and therefore, since every satellite of the ring is statically unstable
between its neighbours, the slightest disturbance would tend to produce collisions
among the satellites, and to break up the ring into groups of conglomerated
satellites.

But if we consider the dynamics of the problem, we shall find that this
effect need not necessarily take place, and that this very force which tends
towards destruction may become the condition of the preservation of the ring.
Suppose the whole ring to be revolving round a central body, and that one
satellite gets in advance of its mean position. It will then be attracted forwards,
its path will become less concave towards the attracting body, so that its distance
from that body will increase. At this increased distance its angular velocity
will be less, so that instead of overtaking those in front, it may by this means
be made to fall back to its original position. Whether it does so or not must
depend on the actual values of the attractive forces and on the angular velocity
of the ring. When the angular velocity is great and the attractive forces small,

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 367

the compensating process will go on vigorously, and the ring will be preserved.
When the angular velocity is small and the attractive forces of the ring great,
the dynamical effect will not compensate for the disturbing action of the forces
and the ring will be destroyed.

If the satellite, instead of being displaced forwards, had been originally
behind its mean position in the ring, the forces would have pulled it backwards,
its path would have become more concave towards the centre, its distance from
the centre would diminish, its angular velocity would increase, and it would
gain upon the rest of the ring till it got in front of its mean position. This
effect is of course dependent on the very same conditions as in the former case,
and the actual effect on a disturbed satellite would be to make it describe an
orbit about its mean position in the ring, so that if in' advance of its mean
position, it first recedes from the centre, then falls behind its mean position in
the ring, then approaches the centre within the mean distance, then advances
beyond its mean position, and, lastly, recedes from the centre till it reaches its
starting-point, after which the process is repeated indefinitely, the orbit being
always described in the direction opposite to that of the revolution of the
ring.

We now understand what would happen to a disturbed satellite, if all the
others were preserved from disturbance. But, since all the satellites are equally
free, the motion of one will produce changes in the forces acting on the rest,
and this will set them in motion, and this motion will be propagated from one
satellite to another round the ring. Now propagated disturbances constitute
waves, and all waves, however complicated, may be reduced to combinations of
simple and regular waves ; and therefore all the disturbances of the ring may
be considered as the resultant of many series of waves, of different lengths, and
travelling with different velocities. The investigation of the relation between
the length and velocity of these waves forms the essential part of the problem,
after which we have only to split up the original disturbance into its simple
elements, to calculate the effect of each of these separately, and then to combine
the results. The solution thus obtained will be perfectly general, and quite
independent of the particular form of the ring, whether regular or irregular at
starting. § 14.

We next investigated the effect upon the ring of an external disturbing
force. Having split up the disturbing force into components of the same type

368 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

with the waves of the ring (an operation which is always possible), we found
that each term of the disturbing force generates a "forced wave" travelling with
its own angular velocity. The magnitude of the forced wave depends not only
on that of the disturbing force, but on the angular velocity with which the dis-
turbance travels round the ring, being greater in proportion as this velocity
more nearly coincides with that of one of the "free waves" of the ring. We
also found that the displacement of the satellites was sometimes in the direction
of the disturbing force, and sometimes in the opposite direction, according to
the relative position of the forced wave among the four natural ones, producing
in the one case positive, and in the other negative forced waves. In treating
the problem generally, we must determine the forced waves belonging to every
term of the disturbing force, and combine these with such a system of free
waves as shall reproduce the initial state of the ring. The subsequent motion
of the ring is that which would result from the free waves and forced waves
together. The most important class of forced waves are those which are pro-
duced by waves in neighbouring rings. § 15.

We concluded the theory of a ring of satellites by tracing the process by
which the ring would be destroyed if the conditions of stability were not
fulfilled. We found two cases of instability, depending on the nature of the
tangential force due to tangential displacement. If this force be in the direction
opposite to the displacement, that is, if the parts of the ring are statically
stable, the ring will be destroyed, the irregularities becoming larger and larger
without being propagated round the ring. When the tangential force is in the
direction, of the tangential displacement, if it is below a certain value, the
disturbances will be propagated round the ring without becoming larger, and
we have the case of stability treated of at large. If the force exceed this value,
the disturbances will still travel round the ring, but they will increase in ampli-
tude continually till the ring falls into confusion. § 18.

We then proceeded to extend our method to the case of rings of different
constitutions. The first case was that of a ring of satellites of unequal size.
If the central body be of sufficient mass, such a ring will be spaced out, so that
the larger satellites will be at wider intervals than the smaller ones, and the
waves of disturbance will be propagated as before, except that there may be
reflected waves when a wave reaches a part of the ring where there is a change
in the average size of the satellites. § 19.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 369

The next case was that of an annular cloud of meteoric stones, revolving
uniformly about the planet. The average density of the space through which
these small bodies are scattered will vary with every irregularity of the motion,
and this variation of density will produce variations in the forces acting upon
the other parts of the cloud, and so disturbances will be propagated in this
ring, as in a ring of a finite number of satellites. The condition that such a
ring should be free from destructive oscillations is, that the density of the
planet should be more than three hundred times that of the ring. This would
make the ring much rarer than common air, as regards its average density,
though the density of the particles of which it is composed may be great.
Comparing this result with Laplace's minimum density of a ring revolving as
a whole, we find that such a ring cannot revolve as a whole, but that the inner
parts must have a greater angular velocity than the outer parts. § 20.

We next took up the case of a flattened ring, composed of incompressible
fluid, and moving with uniform angular velocity. The internal forces here arise
partly from attraction and partly from fluid pressure. We began by taking the
case of an infinite stratum of fluid affected by regular waves, and found the accurate
values of the forces in this case. For long waves the resultant force is in the
same direction as the displacement, reaching a maximum for waves whose
length is about ten times the thickness of the stratum. For waves about five
times as long as the stratum is thick there is no resultant force, and for shorter
waves the force is in the opposite direction to the displacement. § 23.

Applying these results to the case of the ring, we find that it will be
destroyed by the long waves unless the fluid is less than ^ of the density of
the planet, and that in all cases the short waves will break up the ring into
small satellites.

Passing to the case of narrow rings, we should find a somewhat larger
maximum density, but we should still find that very short waves produce forces
in the direction opposite to the displacement, and that therefore, as already
explained (page 333), these short undulations would increase in magnitude without
being propagated along the ring, till they had broken up the fluid filament into
drops. These drops may or may not fulfil the condition formerly given for the
stability of a ring of equal satellites. If they fulfil it, they will move as a
permanent ring. If they do not, short waves will arise and be propagated among
the satellites, with ever increasing magnitude, till a sufficient number of drops

VOL. I. 47

370 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

have been brought into collision, so as to unite and form a smaller number of
larger drops, which may be capable of revolving as a permanent ring.

We have already investigated the disturbances produced by an external
force independent of the ring; but the special case of the mutual perturbations
of two concentric rings is considerably more complex, because the existence of a
double system of waves changes the character of both, and the waves produced
react on those that produced them.

We determined the attraction of a ring upon a particle of a concentric
ring, first, when both rings are in their undisturbed state ; secondly, when the
particle is disturbed ; and, thirdly, when the attracting ring is disturbed by a
series of waves. § 26.

We then formed the equations of motion of one of the rings, taking in the
disturbing forces arising from the existence of a wave in the other ring, and
found the small variation of the velocity of a wave in the first ring as dependent
on the magnitude of the wave in the second ring, which travels with it. § 27.

The forced wave in the second ring must have the same absolute angular
velocity as the free wave of the first which produces it, but this velocity of
the free wave is slightly altered by the reaction of the forced wave upon it.
We find that if a free wave of the first ring has an absolute angular velocity
not very different from that of a free wave of the second ring, then if both
free waves be of even orders (that is, of the second or fourth varieties of waves),
or both of odd orders (that is, of the first or third), then the swifter of the
two free waves has its velocity increased by the forced wave which it produces,
and the slower free wave is rendered still slower by its forced wave ; and even
when the two free waves have the same angular velocity, their mutual action
will make them both split into two, one wave in each ring travelling faster,
and the other wave in each ring travelling slower, than the rate with which
they would move if they had not acted on each other.

But if one of the free waves be of an even order and the other of an odd
order, the swifter free wave will travel slower, and the slower free wave will
travel swifter, on account of the reaction of their respective forced waves. If
the two free waves have naturally a certain small difference of velocities, they
will be made to travel together, but if the difference is less than this, they
will again split into two pairs of waves, one pair continually increasing in

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 371

magnitude without limit, and the other continually diminishing, so that one
of the waves in each ring will increase in violence till it has thrown the ring
into a state of confusion.

There are four cases in which this may happen. The first wave of the
outer ring may conspire with the second or the fourth of the inner ring, the
second of the outer with the third of the inner, or the third of the outer with
the fourth of the inner. That two rings may revolve permanently, their distances
must be arranged so that none of these conspiracies may arise between odd
and even waves, whatever be the value of m. The number of conditions to
be fulfilled is therefore very great, especially when the rings are near together
and have nearly the same angular velocity, because then there are a greater
number of dangerous values of m to be provided for.

In the case of a large number of concentric rings, the stability of each pair
must be investigated separately, and if in the case of any two, whether con-
secutive rings or not, there are a pair of conspiring waves, those two rings will
be agitated more and more, till waves of that kind are rendered impossible by
the breaking up of those rings into some different arrangement. The presence
of the other rings cannot prevent the mutual destruction of any pair which
bear such relations to each other.

It appears, therefore, that in a system of many concentric rings there will
be continually new cases of mutual interference between different pairs of rings.
The forces which excite these disturbances being very small, they will be slow
of growth, and it is possible that by the irregularities of each of the rings the
waves may be so broken and confused (see § 19), as to be incapable of mounting
up to the height at which they would begin to destroy the arrangement of the
ring. In this way it may be conceived to be possible that the gradual dis-
arrangement of the system may be retarded or indefinitely postponed.

But supposing that these waves mount up so as to produce collisions among
the particles, then we may deduce the result upon the system from general
dynamical principles. There will be a tendency among the exterior rings to
remove further from the planet, and among the interior rings to approach the
planet, and this either by the extreme interior and exterior rings diverging
from each other, or by intermediate parts of the system moving away from the
mean ring. If the interior rings are observed to approach the planet, while it

47—2

372 ON THE STABILITY OF THE MOTION OF SATURN 8 KINGS.

is known that none of the other rings have expanded, then the cause of the
change cannot be the mutual action of the parts of the system, but the resistance
of some medium in which the rings revolve. § 31-

There is another cause which would gradually act upon a broad fluid ring
of which the parts revolve each with the angular velocity due to its distance
from the planet, namely, the internal friction produced by the slipping of the
concentric rings with different angular velocities. It appears, however (§ 33),
that the effect of fluid friction would be insensible if the motion were regular.

Let us now gather together the conclusions we have been able to draw
from the mathematical theory of various kinds of conceivable rings.

We found that the stability of the motion of a solid ring depended on
so delicate an adjustment, and at the same time so unsymmetrical a distribution
of mass, that even if the exact condition were fulfilled, it could scarcely last
long, and if it did, the immense preponderance of one side of the ring would
be easily observed, contrary to experience. These considerations, with others
derived from the mechanical structure of so vast a body, compel us to abandon
any theory of solid rings.

We next examined the motion of a ring of equal satellites, and found that
if the mass of the planet is sufficient, any disturbances produced in the arrange-
ment of the ring will be propagated round it in the form of waves, and will not
introduce dangerous confusion. If the satellites are unequal, the propagation of
the waves will no longer be regular, but disturbances of the ring will in this,
as in the former case, produce only waves, and not growing confusion. Sup-
posing the ring to consist, not of a single row of large satellites, but of a cloud
of evenly distributed unconnected particles, we found that such a cloud must
have a very small density in order to be permanent, and that this is inconsistent
with its outer and inner parts moving with the same angular velocity. Supposing
the ring to be fluid and continuous, we found that it will be necessarily broken
up into small portions.

We conclude, therefore, that the rings must consist of disconnected particles ;
these may be either solid or liquid, but they must be independent. The entire
system of rings must therefore consist either of a series of many concentric rings,
each moving with its own velocity, and having its own systems of waves, or else
of a confused multitude of revolving particles, not arranged in rings, and
continually coming into collision with each other.

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 373

Taking the first case, we found that in an indefinite number of possible
cases the mutual perturbations of two rings, stable in themselves, might mount
up in time to a destructive magnitude, and that such cases must continually
occur in an extensive system like that of Saturn, the only retarding cause being
the possible irregularity of the rings.

The result of long-continued disturbance was found to be the spreading
out of the rings in breadth, the outer rings pressing outwards, while the inner
rings press inwards.

The final result, therefore, of the mechanical theory is, that the only system
of rings which can exist is one composed of an indefinite number of unconnected
particles, revolving round the planet with different velocities according to their
respective distances. These particles may be arranged in 'series of narrow rings,
or they may move through each other irregularly. In the first case the destruc-
tion of the system will be very slow, in the second case it will be more rapid,
but there may be a tendency towards an arrangement in narrow rings, which
may retard the process.

We are not able to ascertain by observation the constitution of the two
outer divisions of the system of rings, but the inner ring is certainly transparent,
for the limb of Saturn has been observed through it. It is also certain, that
though the space occupied by the ring is transparent, it is not through the
material parts of it that Saturn was seen, for his limb was observed without
distortion ; which shows that there was no refraction, and therefore that the
rays did not pass through a medium at all, but between the solid or liquid
particles of which the ring is composed. Here then we have an optical argument
in favour of the theory of independent particles as the material of the rings.
The two outer rings may be of the same nature, but not so exceedingly rare
that a ray of light can pass through their whole thickness without encountering
one of the particles.

Finally, the two outer rings have been observed for 200 years, and it appears,
from the careful analysis of all the observations by M. Struve, that the second
ring is broader than when first observed, and that its inner edge is nearer the
planet than formerly. The inner ring also is suspected to be approaching the
planet ever since its discovery in 1850. These appearances seem to indicate
the same slow progress of the rings towards separation which we found to be
the result of theory, and the remark, that the inner edge of the inner ring is

374 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

most distinct, seems to indicate that the approach towards the planet is less
rapid near the edge, as we had reason to conjecture. As to the apparent
unchangeableness of the exterior diameter of the outer ring, we must remember
that the outer rings are certainly far more dense than the inner one, and that
a small change in the outer rings must balance a great change in the inner
one. It is possible, however, that some of the observed changes may be due
to the existence of a resisting medium. If the changes already suspected should
be confirmed by repeated observations with the same instruments, it will be
worth while to investigate more carefully whether Saturn's Rings are permanent
or transitionary elements of the Solar System, and whether in that part of
the heavens we see celestial immutability, or terrestrial corruption and generation,
and the old order giving place to new before our own eyes.

APPENDIX.

On the Stability of the Steady Motion of a Rigid Body about a Fixed Centre of Force.
By PROFESSOR W. THOMSON (communicated in a letter).

THE body will be supposed to be symmetrical on the two sides of a certain plane
containing the centre of force, and no motion except that of parts of the body parallel
to the plane will be considered. Taking it as the plane of construction, let G (fig. 14)
be the centre of gravity of the body, and 0 a point at which the resultant attraction of
the body is in the line OG towards G. Then if the body be placed with 0 coinciding
with the centre of force, and set in a state of rotation about that point as an axis, with

an angular velocity equal to A/ \r> (where / denotes the attraction of the body on a

unit of matter at 0, S the amount of matter in the central body, M. the mass of the
revolving body, and a the distance OG), it will continue, provided it be perfectly undis-
turbed, to revolve uniformly at this rate, and the attraction Sf on the moving body will
be constantly balanced by the centrifugal force ufaM of its motion.

Let us now suppose the motion to be slightly disturbed, and let it be required to
investigate the consequences. Let X, S, Y, be rectangular axes of reference revolving
uniformly with the angular velocity o>, round S, the fixed attracting point. Let x, y, be
the co-ordinates of G with reference to these axes, and let XS, YS denote the components

ON THE STABILITY OF THE MOTION OF SATURN'S RINGS. 375

of the whole force of attraction of S on the rigid body. Then since this force is in the
line through S, its moment round G is

SYx-SXy;

the components of the forces on the moving body being reckoned as positive when they
tend to diminish x and y respectively. Hence if k denote the radius of gyration of the
body round G, and if <f> denote the angle which OG makes with SX (i.e. the angle OOK),
the equations of motion are,

In the first place we see that one integral of these equations is

This is the "equation of angular momentum."

In considering whether the motion round S with velocity eo when 0 coincides with
8 is stable or unstable, we must find whether every possible motion with the same
" angular momentum " round 8 is such that it will never bring 0 to more than an infinitely
small distance from <S : that is to say, we must find whether, for every possible solution
iu which H = M (a1 + k*) a>, and for which the co-ordinates of 0 are infinitely small at one
time, these co-ordinates remain infinitely small. Let these values at time t be denoted
thus: /82V =£ and N0=ij; let OG be at first infinitely nearly parallel to OX, i.e. let </>
be infinitely small (the full solution will tell us whether or not $ remains infinitely small) ;
then, as long as $ is infinitely small, we have

x==a+ f , y = r) + a<l>,
and the equations of motion have the forms

and we may write the equation of angular momentum instead of the third equation,

If now we suppose £ and i) to be infinitely small, the last of these equations becomes

> + 2&,af+a=0 ...................................... (a).

376 ON THE STABILITY OF THE MOTION OF SATURN'S RINGS.

If p and q denote the components parallel and perpendicular to 00 of the attraction
of the body on a unit of matter at S, we have

X=pcos<f> — qain<f> = p, and Y=psm<f> + qcos <j> =p<j> + q,

since q and <f> are each infinitely small; and if we put V= potential at 8, and

<TV

_

" d? ' ~ drf '

then p =/- af - 717, q = -@r)- yf,

If we make these substitutions for A' and )'. and take into account that

/=»'<* f ................................................ (*),

the first and second equations of motion become

Combining equations (a), (c), and (e?), by the same method as that adopted in the text,
we find that the differential equation in f , 17, or <£, is of the form

where A = Jf,

B=a>* (2i* — o*) - ^~ {/fc'a + (a"

C = a>4 (*• - 3a") + w" 5 {(a" + £') (a + /8) - 4a'/9) + (a1

In comparing this result with that obtained in the Essay, we must put

r0 for a,
R for M,
B + S for 8,
L for o,
XrJ for &
Mr, for 7.

VOL. I. PLATE V.

Cambridge Umi/ertnty Press

[From the Philosophical Magazine for January and July, I860.]

XX. Illustrations of the Dynamical Theory of Gases*.

PART I.

/

ON THE MOTIONS AND COLLISIONS OF PEEFECTLY ELASTIC SPHERES.

So many of the properties of matter, especially when in the gaseous form,
can be deduced from the hypothesis that their minute parts are in rapid motion,
the velocity increasing with the temperature, that the precise nature of this
motion becomes a subject of rational curiosity. Daniel Bernouilli, Herapath,
Joule, Kronig, Clausius, &c. have shewn that the relations between pressure,
temperature, and density in a perfect gas can be explained by supposing the
particles to move with uniform velocity in straight lines, striking against the
sides ef the containing vessel and thus producing pressure. It is not necessary
to suppose each particle to travel to any great distance in the same straight
line ; for the effect in producing pressure will be the same if the particles
strike against each other ; so that the straight line described may be very short.
M. Clausius has determined the mean length of path in terms of the average
distance of the particles, and the distance between the centres of two particles
when collision takes place. We have at present no means of ascertaining either
of these distances ; but certain phenomena, such as the internal friction of gases,
the conduction of heat through a gas, and the diffusion of one gas through
another, seem to indicate the possibility of determining accurately the mean
length of path which a particle describes between two successive collisions. In
order to lay the foundation of such investigations on strict mechanical principles,
I shall demonstrate the laws of motion of an indefinite number of small, hard,
and perfectly elastic spheres acting on one another only during impact.

* Read at the Meeting of the British Association at Aberdeen, September 21, 1859.
VOL. I. 48

378 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASE&

If the properties of such a system of bodies are found to correspond to
those of gases, an important physical analogy will be established, which may
lead to more accurate knowledge of the properties of matter. If experiments
on gases are inconsistent with the hypothesis of these propositions, then our
theory, though consistent with itself, is proved to be incapable of explaining
the phenomena of gases. In either case it is necessary to follow out the
consequences of the hypothesis.

Instead of saying that the particles are hard, spherical, and elastic, we may
if we please say that the particles are centres of force, of which the action is
insensible except at a certain small distance, when it suddenly appears as a
repulsive force of very great intensity. It is evident that either assumption
will lead to the same results. For the sake of avoiding the repetition of a.
long phrase about these repulsive forces, I shall proceed upon the assumption
of perfectly elastic spherical bodies. If we suppose those aggregate molecules
which move together to have a bounding surface which is not spherical, then
the rotatory motion of the system will store up a certain proportion of the
whole via viva, as has been shewn by Clausius, and in this way we may
account for the value of the specific heat being greater than on the more
simple hypothesis.

On the Motion and Collision of Perfectly Elastic Spheres.

Prop. I. Two spheres moving in opposite directions with velocities inversely
as their masses strike one another ; to determine their motions after impact.

Let P and Q be the position of the centres at
impact ; AP, BQ the directions and magnitudes of
the velocities before impact ; Pa, Qb the same after
impact ; then, resolving the velocities parallel and per-
pendicular to PQ the line of centres, we find that
the velocities parallel to the line of centres are exactly
reversed, while those perpendicular to that line are

unchanged. Compounding these velocities again, we find that the velocity of
each ball is the same before and after impact, and that the directions before
and after impact lie in the same plane with the line of centres, and make equal
angles with it.

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 379

Prop. II. To find the probability of the direction of the velocity after
impact lying between given limits.

In order that a collision may take place, the line of motion of one of the
balls must pass the centre of the other at a distance less than the sum of
their radii ; that is, it must pass through a circle whose centre is that of the
other ball, and radius (s) the sum of the radii of the balls. Within this circle
every position is equally probable, and therefore the probability of the distance
from the centre being between r and r + dr is

2rdr

~7~'

Now let <f» be the angle A Pa between the original direction and the direction
after impact, then APN =%<$>, and r = s sin ^<f>, and the probability becomes

sn

The area of a spherical zone between the angles of polar distance <f> and tf> + d<f> is

ZTT sin <f>d<f> ;

therefore if 01 be any small area on the surface of a sphere, radius unity, the
probability of the direction of rebound passing through this area is

CM

4^;

so that the probability is independent of <f>, that is, all directions of rebound
are equally likely.

Prop. III. Given the direction and magnitude of the velocities of two
spheres before impact, and the line of centres at impact ; to find the velocities
after impact.

Let OA, OB represent the velocities before impact, so that if there had been
no action between the bodies they would
have been at A and B at the end of a
second. Join AB, and let G be their centre
of gravity, the position of which is not
affected by their mutual action. Draw GN
parallel to the line of centres at impact (not
necessarily in the plane AOB). Draw aGb

in the plane AGN, making NGa^NGA, and Ga=GA and Gb = GB; then by

48—2

380 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

Prop. I. Ga and Gb will be the velocities relative to G ; and compounding
these with OG, we have Oa and Ob for the true velocities after impact.

By Prop. II. all directions of the line aGb are equally probable. It appears
therefore that the velocity after impact is compounded of the velocity of the
centre of gravity, and of a velocity equal to the velocity of the sphere relative
to the centre of gravity, which may with equal probability be in any direction
whatever.

If a great many equal spherical particles were in motion in a perfectly
elastic vessel, collisions would take place among the particles, and their velocities
would be altered at every collision ; so that after a certain time the vis viva
will be divided among the particles according to some regular law, the average
number of particles whose velocity lies between certain limits being ascertainable,
though the velocity of each particle changes at every collision.

Prop. IV. To find the average number of particles whose velocities lie
between given limits, after a great number of collisions among a great number
of equal particles.

Let N be the whole number of particles. Let x, y, z be the components
of the velocity of each particle in three rectangular directions, and let the number
of particles for which x lies between x and x + dx, be Nf(x)dx, where f(x) is
a function of x to be determined.

The number of particles for which y lies between y and y + dy will be
Nf(y)dy ; and the number for which z lies between z and z + dz will be Nf(z)dz,
where f always stands for the same function.

Now the existence of the velocity x does not in any way affect that of
the velocities y or z, since these are all at right angles to each other and
independent, so that the number of particles whose velocity lies between x and
x + dx, and also between y and y + dy, and also between z and z + dz, is

Nf(x)f(y)f(z)dxdydz.

If we suppose the N particles to start from the origin at the same instant,
then this will be the number in the element of volume (dxdydt) after unit of
time, and the number referred to unit of volume will be

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 381

But the directions of the coordinates are perfectly arbitrary, and therefore this
number must depend on the distance from the origin alone, that is

Solving this functional equation, we find

If we make A positive, the number of particles will increase with the
velocity, and we should find the whole number of particles infinite. We there-
fore make A negative and equal to — j , so that the number between x and

x + dx is

NCe~-'dx.

Integrating from x = — oo to a; = + o> , we find the whole number of particles,

1

-

fix) is therefore 7=e a .

aVTT

Whence we may draw the following conclusions : —

1st. The number of particles whose velocity, resolved in a certain direction,
lies between x and x + dx is

dx ................................. (1).

avir
2nd. The number whose actual velocity lies between v and v + dv is

e*dv ................................. (2).

3rd. To find the mean value of v, add the velocities of all the particles
together and divide by the number of particles ; the result is

mean velocity = -r= ................................. (3).

V7T

4th. To find the mean value of v\ add all the values together and
divide by N,

mean value of zrs = |a2 .............................. (4).

This is greater than the square of the mean velocity, as it ought to be.

382 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

It appears from this proposition that the velocities are distributed among
the particles according to the same law as the errors are distributed among
the observations in the theory of the " method of least squares." The velocities
range from 0 to oo, but the number of those having great velocities is com-
jwiratively small. In addition to these velocities, which are in all directions
equally, there may be a general motion of translation of the entire system of
particles which must be compounded with the motion of the particles relatively
to one another. We may call the one the motion of translation, and the other
the motion of agitation.

PROP. V. Two systems of particles move each according to the law stated
in Prop. IV. ; to find the number of pairs of particles, one of each system,
whose relative velocity lies between given limits.

Let there be N particles of the first system, and N' of the second, then
NN' is the whole number of such pairs. Let us consider the velocities in the
direction of x only ; then by Prop. IV. the number of the first kind, whose
velocities are between x and x + dx, is

a -Jir
The number of the second kind, whose velocity is between x + y and x + y + dy, is

_

N'- -i=e * dy,
P-J*

where /8 is the value of a for the second system.

The number of pairs which fulfil both conditions is

-, l<* (

NN'--e « ~P~' dxdy.
afiir

Now x may have any value from — oo to + oo consistently with the difference
of velocities being between y and y + dy; therefore integrating between these
limits, we find

NN' -f===-f=edy ......................... (5)

'

for the whole number of pairs whose difference of velocity lies between y and
y + dy.

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

383

This expression, which is of the same form with (1) if we put NN' for
N, a' + f? for a2, and y for x, shews that the distribution of relative velocities
is regulated by the same law as that of the velocities themselves, and that
the mean relative velocity is the square root of the sum of the squares of the
mean velocities of the two systems.

Since the direction of motion of every particle in one of the systems may
be reversed without changing the distribution of velocities, it follows that the
velocities compounded of the velocities of two particles, one in each system, are
distributed according to the same formula (5) as the relative velocities.

PROP. VI. Two systems of particles move in the same vessel ; to prove
that the mean vis viva of each particle will become the same in the two
systems.

Let P be the mass of each particle of the first system, Q that of each
particle of the second. Let p, q be the mean veloci-
ties in the two systems before impact, and let p', q'
be the mean velocities after one impact. Let OA = p
and OB = q, and let AOB be a right angle; then, by
Prop. V., AB will be the mean relative velocity, OG will
be the mean velocity of the centre of gravity ; and drawing
aGb at right angles to OG, and making aG = AG and
bG = BG, then Oa will be the mean velocity of P after
impact, compounded of OG and Ga, and Ob will be that of Q after impact.

Now

P+Q

therefore

and

and

P+Q

(6)-

384 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASE3.

It appears therefore that the quantity Pp* — Qq* is diminished at every impact
in the same ratio, so that after many impacts it will vanish, and then

q_ O_

Now the mean vis viva is f Pa* = — Pp* for P, and ' - Qq* for Q ; and it is

o o

manifest that these quantities will be equal when Pp' = Qq*.

If any number of different kinds of particles, having masses P, Q, R and
velocities p, q, r respectively, move in the same vessel, then after many impacts

= Rr>, &c ............................... (7).

PROP. VII. A particle moves with velocity r relatively to a number of
particles of which there are N in unit of volume ; to find the number of these
which it approaches within a distance s in unit of time.

If we describe a tubular surface of which the axis is the path of the
particle, and the radius the distance s, the content of this surface generated
in unit of time will be wrs?, and the number of particles included in it will be

Nirrs* ....................................... (8),

which is the number of particles to which the moving particle approaches within
a distance s.

PROP. VIII. A particle moves with velocity v in a system moving according
to the law of Prop. IV.; to find the number of particles which have a velocity
relative to the moving particle between r and r + dr.

Let u be the actual velocity of a particle of the system, v that of the
original particle, and r their relative velocity, and 6 the angle between v and r,
then

tt* = i? + ?•* — 2 vr cos 6.

If we suppose, as in Prop. IV., all the particles to start from the origin at
once, then after unit of time the "density" or number of particles to unit of
volume at distance u will be

1 -"'

N- -e «*
••" « « "

a. IT'-

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 385

From this we have to deduce the number of particles in a shell whose centre
is at distance v, radius = r, and thickness = dr,

-I „ &•-»)« (r+»y

If -4= - {e~~^-e~~}dr .......................... (9),

which is the number required.

COR. It is evident that if we integrate this expression from r = 0 to
/• = oo , we ought to get the whole number of particles = N, whence the following
mathematical result,

,00 («-ay (x+af

dx.x(e~ «' -e~ •» ) = J-naa ..................... (10).

Jo

,

PROP. IX. Two sets of particles move as in Prop. V.; to find the number
of pairs which approach within a distance s in unit of time.

The number of the second kind which have a velocity between v and v + dv is

The number of the first kind whose velocity relative to these is between r
and r + dr is

N-= - (e f -
a^ir v v

and the number of pairs which approach within distance s in unit of time is

nn'-jrrs',

4 _£ (»-«* _(»•»*•)»

= KIT — ™ s'r've * {e a' —e •* } dr dv.

By the last proposition we are able to integrate with respect to v, and get

———

(<*'+£')*

Integrating this again from r = 0 to r = oo ,

(11)

is the number of collisions in unit of time which take place in unit of volume

between particles of different kinds, s being the distance of centres at collision.

VOL. r. 49

386 ILLUSTRATIONS OF THE DYNAMICAL THEORY OP GASES.

The number of collisions between two particles of the first kind, s, being the
striking distance, is

and for the second system it is

2N* J* V2J8V.

n _ 9/3

The mean velocities in the two systems are ' and -£. ; so that if Z, and lt

•JTT «/TT

be the mean distances travelled by particles of the first and second systems
between each collision, then

PROP. X. To find the probability of a particle reaching a given distance
before striking any other.

Let us suppose that the probability of a particle being stopped while
passing through a distance dx, is adx; that is, if N particles arrived at a
distance x, Nadx of them would be stopped before getting to a distance x + dx.
Putting this mathematically,

or N=Ce-*.

dx

Putting N=l when x = 0, we find e'" for the probability of a particle not
striking another before it reaches a distance x.

The mean distance travelled by each particle before striking is - = I. The

probability of a particle reaching a distance = nl without being struck is e~n.
(See a paper by M. Clausius, Philosophical Magazine, February 1859.)

If all the particles are at rest but one, then the value of a is

a = irs'N,

where * is the distance between the centres at collision, and .2V is the number
of particles in unit of volume. If v be the velocity of the moving particle
relatively to the rest, then the number of collisions in unit of time will be

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 387

and if vl be the actual velocity, then the number will be v^a. ; therefore

where v1 is the actual velocity of the striking particle, and v its velocity
relatively to those it strikes. If vt be the actual velocity of the other particles,
then v = -Jv? + v?. If vl = vt, then v = 72 vl , and

Note*. M. Clausius makes a = %irs1N.

PROP. XL In a mixture of particles of two different kinds, to find the
mean path of each particle. /

Let there be JV, of the first, and Nt of the second in unit of volume.
Let s, be the distance of centres for a collision between two particles of the
first set, 5, for the second set, and s for collision between one of each kind.
Let v1 and vt be the coefficients of velocity, Mw Mt the mass of each particle.

The probability of a particle J/t not being struck till after reaching a
distance xl by another particle of the same kind is

* [In the Philosophical Magazine of 1860, Vol. I. pp. 434 — 6 Clausius explains the method by
which he found hia value of the mean relative velocity. It is briefly as follows : If u, v be the
velocities of two particles their relative velocity is «/w* + if — 2uv cos 6 and the mean of this as
regards direction only, all directions of v being equally probable, is shewn to be

lu' IS

v + 3 — when u <= v. and u + = — when u > v.
3 v 3 u

If v-u these expressions coincide. Clausius in applying this result and putting u, v for the
mean velocities assumes that the mean relative velocity is given by expressions of the same form,
BO that when the mean velocities are each equal to u the mean relative velocity would be ^u.
This step is, however, open to objection, and in fact if we take the expressions given above for the
mean velocity, treating u and v as the velocities of two particles which may have any values between
0 and oo , to calculate the mean relative velocity we should proceed as follows : Since the number of

4 _2?
particles with velocities between w and u + du is N 3 / u'e •' du, the mean relative velocity is

16

rr (*+*)/ iv*\, 16 r /•» . ,(«-%.*)/ I«A , .

/ wVe \a- >' I u + g — I dudv + -^r- I I u'v'e Va» V/ I « + 5 — I du dv.
Jo Jo \ 3 u ) a'/JV /o JO \ •» • /

This expression, when reduced, leads to -j= Jo.1 + /3*, which is the result in the text. Ed.]

49—2

388 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

The probability of not being struck by a particle of the other kind in the same
distance is

e ••

Therefore the probability of not being struck by any particle before reaching a
distance x is

and if Z, be the mean distance for a particle of the first kind,

\ = fasM + it 1 1 + V-( s*Nt ...................... (12).

t, "v vt

Similarly, if Z, be the mean distance for a particle of the second kind,

N, ........................ (13).

,

The mean density of the particles of the first kind is NlMl = plt and that of
the second NtMi = pr If we put

(15),

a

and ^

C

PROP. XII. To find the pressure on unit of area of the side of the vessel
due to the impact of the particles upon it.

Let N= number of particles in unit of volume;
3/= mass of each particle ;
v = velocity of each particle ;
I = mean path of each particle ;

then the number of particles in unit of area of a stratum dz thick is

Ndz ........................................ (17).

The number of collisions of these particles in unit of time is

Ndzj ...................................... (18).

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 389

The number of particles which after collision reach a distance between nl and
(n + dn) I is

Nje~ndzdn (19).

The proportion of these which strike on unit of area at distance z is

•""' (*»):"*

the mean velocity of these in the direction of z is

•^ <*'>•

Multiplying together (19), (20), and (21), and M, we find the momentum at
impact

MN ~t (nV - z2) e- dz dn.
4n"l* v

Integrating with respect to z from 0 to nl, we get

X /V/ f\l/11 ¥lf>~ /"I'M
"Jr.iJ-t.iT c/ /frC \Julvt

Integrating with respect to n from 0 to oo , we get

for the momentum in the direction of z of the striking particles ; for the
momentum of the particles after impact is the same, but in the opposite
direction ; so that the whole pressure on unit of area is twice this quantity, or

This value of p is independent of I the length of path. In applying this
result to the theory of gases, we put MN=p, and v' = 3k, and then

which is Boyle and Mariotte's law. By (4) we have

'y> = fa2, .'. a2 = 2& (23).

We have seen that, on the hypothesis of elastic particles moving in straight
lines, the pressure of a gas can be explained by the assumption that the square
of the velocity is proportional directly to the absolute temperature, and inversely
to the specific gravity of the gas at constant temperature, so that at the same

390 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

pressure and temperature the value of NMtf is the same for all gases. But
we found in Prop. VI. that when two sets of particles communicate agitation
to one another, the value of Mif is the same in each. From this it appears
that N, the number of particles in unit of volume, is the same for all gases
at the same pressure and temperature. This result agrees with the chemical law,
that equal volumes of gases are chemically equivalent.

We have next to determine the value of I, the mean length of the path
of a particle between consecutive collisions. The most direct method of doing
this depends upon the feet, that when different strata of a gas slide upon
one another with different velocities, they act upon one another with a tan-
gential force tending to prevent this sliding, and similar in its results to the
friction between two solid surfaces sliding over each other in the same way.
The explanation of gaseous friction, according to our hypothesis, is, that particles
having the mean velocity of translation belonging to one layer of the gas, pass
out of it into another layer having a different velocity of translation ; and
by striking against the particles of the second layer, exert upon it a tangential
force which constitutes the internal friction of the gas. The whole friction
between two portions of gas separated by a plane surface, depends upon the
total action between all the layers on the one side of that surface upon all the
layers on the other side.

PROP. XIII. To find the internal friction in a system of moving particles.

Let the system be divided into layers parallel to the plane of xy, and
let the motion of translation of each layer be u in the direction of x, and
let u = A + Bz. We have to consider the mutual action between the layers on
the positive and negative sides of the plane xy. Let us first determine the
action between two layers dz and dz', at distances z and — z' on opposite sides
of this plane, each unit of area. The number of particles which, starting from
dz in unit of time, reach a distance between nl and (n + dn)l is by (19),

m

N j e~* dz dn.

The number of these which have the ends of their paths in the layer dz' is

N —-e~

The mean velocity in the direction of x which each of these has before impact
is A + Bz, and after impact A+Bz'; and its mass is Mt so that a mean

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 391

momentum =MB(z — z') is communicated by each particle. The whole action due
to these collisions is therefore

NMB ;A (z - z') e~n dz dz' dn.
2nP x

We must first integrate with respect to z' between z' = 0 and z' = z — nl; this
gives

\NMB (n'P - z2) e~n dz dn

for the action between the layer dz and all the layers below the plane xy.
Then integrate from z = 0 to z — nl,

Integrate from n = 0 to n = QO , and we find the whole friction between unit

of area above and below the plane to be

u du

where \L is the ordinary coefficient of internal friction,

1 Mv

, ,
(24),

where p is the density, I the mean length of path of a particle, and v the

, ., 2a /2k

mean velocity v = —f= = 2 / — ,

•J-JT V IT

* /> V 2T

Now Professor Stokes finds by experiments on air,

If we suppose \/A = 930 feet per second for air at 60°, and therefore the mean
velocity v = l505 feet per second, then the value of I, the mean distance
travelled over by a particle between consecutive collisions, =^4^000^ °f an
inch, and each particle makes 8,077,200,000 collisions per second.

A remarkable result here presented to us in equation (24), is that if this
explanation of gaseous friction be true, the coefficient of friction is independent
of the density. Such a consequence of a mathematical theory is very startling,
and the only experiment I have met with on the subject does not seem to
confirm it. We must next compare our theory with what is known of the
diffusion of gases, and the conduction of heat through a gas.

.! .i-2 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

PART II.

* ON THE PROCESS OF DIFFUSION OF TWO OR MORE KINDS OF MOVING PARTICLES

AMONG ONE ANOTHER.

We have shewn, in the first part of this paper, that the motions of a
system of many small elastic particles are of two kinds : one, a general motion
of translation of the whole system, which may be called the motion in mass ;
and the other a motion of agitation, or molecular motion, in virtue of which
velocities in all directions are distributed among the particles according to a
certain law. In the cases we are considering, the collisions are so frequent that
the law of distribution of the molecular velocities, if disturbed in any way,
will be re-established in an inappreciably short time ; so that the motion will
always consist of this definite motion of agitation, combined with the general
motion of translation.

When two gases are in communication, streams of the two gases might
run freely in opposite directions, if it were not for the collisions which take
place between the particles. The rate at which they actually interpenetrate each
other must be investigated. The diffusion is due partly to the spreading of the
particles by the molecular agitation, and partly to the actual motion of the
two opposite currents in mass, produced by the pressure behind, and resisted

* [The methods and results of this paper have been criticised by Clausing in a memoir published
in Poggendorffs Annalen, VoL cxv., and in the Philosophical Magazine, VoL xxm. His main objec-
tion is that the various circumstances of the strata, discussed in the paper, have not been sufficiently
represented in the equations. In particular, if there be a series of strata at different temperatures
perpendicular to the axis of x, then the proportion of molecules whose directions form with the
axis of x angles whose cosines lie between /t and /x + efyx is not l<lp. as has been assumed by Maxwell
throughout his work, but './A//t where If is a factor to be determined. In discussing the steady
conduction of heat through a gas Clausius assumes that, in addition to the velocity attributed to
the molecule according to Maxwell's theory, we must also suppose a velocity normal to the stratum
and depending on the temperature of the stratum. On this assumption the factor H is investigated
along with other modifications, and an expression for the assumed velocity is determined from the
consideration that when the flow of heat is steady there is no movement of the mass. Clausius
combining his own results with those of Maxwell points out that the expression contained in (28)
of the paper involves as a result the motion of the gas. He also disputes the accuracy of ex-
pression (59) for the Conduction of Heat In the introduction to the memoir published in the
PhiL Trans., 1866, it will be found that Maxwell expresses dissatisfaction with his former theory
of the Diffusion of Gases, and admits the force of the objections made by Clausius to his expression
for the Conduction of Heat. Ed.]

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 393

by the collisions of the opposite stream. When the densities are equal, the
diffusions due to these two causes respectively are as 2 to 3.

PROP. XIV. In a system of particles whose density, velocity, &c. are
functions of x, to find the quantity of matter transferred across the plane of yz,
due to the motion of agitation alone.

If the number of particles, their velocity, or their length of path is greater
on one side of this plane than on the other, then more particles will cross the
plane in one direction than in the other ; and there will be a transference of
matter across the plane, the amount of which may be calculated.

Let there be taken a stratum whose thickness is dx, and
area unity, at a distance x from the origin. The nuniber of
collisions taking place in this stratum in unit of time will be

NV7dx. **

•

The proportion of these which reach a distance between nl and (n + dn)l before

they strike another particle is

e~ndn.

The proportion of these which pass through the plane yz is

— =- when x is between — nl and 0,
Ml

nl ^~ x •

and -- — y- when x is between 0 and + nl ;

the sign being negative in the latter case, because the particles cross the plane
in the negative direction. The mass of each particle is M ; so that the quantity
of matter which is projected from the stratum dx, crosses the plane yz in a
positive direction, and strikes other particles at distances between nl and
(n + dn) I is

where x must be between ±nl, and the upper or lower sign is to be taken
according as x is positive or negative.

In integrating this expression, we must remember that N, v, and I are
functions of x, not vanishing with x, and of which the variations ate very
small between the limits x= —nl and x= +nl. ,

VOL. i. 50

394 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

As we may have occasion to perform similar integrations, we may state
here, to save trouble, that if U and r are functions of x not vanishing with x,
whose variations are very small between the limits x = + r and x = —r,

<27>-

When m is an odd number, the upper sign only is to be considered ;
when m is even or zero, the upper sign is to be taken with positive values
of x, and the lower with negative values. Applying this to the case before us,

f+* MNv
- ~2T

We have now to integrate

r-i

n being taken from 0 to <x> . We thus find for the quantity of matter trans-
ferred across unit of area by the motion of agitation in unit of time,

where p = MN is the density, v the mean velocity of agitation, and I the mean
length of path.

PROP. XV. The quantity transferred, in consequence of a mean motion of
translation V, would obviously be

Q=Vp ......................... ........... (29).

PROP. XVI. To find the resultant dynamical effect of all the collisions
which take place in a given stratum.

Suppose the density and velocity of the particles to be functions of x,
then more particles will be thrown into the given stratum from that side
on which the density is greatest ; and those particles which have greatest
velocity will have the greatest effect, so that the stratum will not be generally

.ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 395

in equilibrium, and the dynamical measure of the force exerted on the stratum
will be the resultant momentum of all the particles which lodge in it during
unit of time. We shall first take the case in which there is no mean motion
of translation, and then consider the eifect of such motion separately.

Let a stratum whose thickness is a (a small quantity
compared with I), and area unity, be taken at the origin,
perpendicular to the axis of x ; and let another stratum, of
thickness dx, and area unity, be taken at a distance x from
the first.

If MI be the mass of a particle, N the number in unit of volume, v the
velocity of agitation, I the mean length of path, then the number of collisions
which take place in the stratum dx is

N j dx. , ',

The proportion of these which reach a distance between nZ and (n + dn)l is

e~ndn.

The proportion of these which have the extremities of their paths in the
stratum a is

a

Znl'

The velocity of these particles, resolved in the direction of x, is

vx
~nl'

and the mass is M ; so that multiplying all these terms together, we get

NMtfax . ., , , .

2nV. e "dxdn (30)

for the momentum of the particles fulfilling the above conditions.

To get the whole momentum, we must first integrate with respect to x
from x= —nl to x = +nl, remembering that I may be a function of x, and is a
very small quantity. The result is

d /NMv*\ , ,
di(-3-)ane dn-

50-2

396 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

Integrating with respect to n from n = 0 to n = oo , the result is

d

as the whole resultant force on the stratum a arising from these collisions.

Now - - =p by Prop. XII., and therefore we may write the equation
3

(32),

WWf

the ordinary hydrodynamical equation.

PROP. XVII. To jind the resultant effect of the collisions upon each of
several different systems of particles mixed togetJier.

Let MI, Mt, &c. be the masses of the different kinds of particles, Nlt
Nt, &c. the number of each kind in unit of volume, vlt vt, &c. their velocities
of agitation, ?,, I, their mean paths, plt pt, &c. the pressures due to each
system of particles ; then

1

,(33).

The number of collisions of the first kind of particles with each other in unit
of time will be

The number of collisions between particles of the first and second kinds will be

Nj.\Bptt or NjojCplt because v'B = v*C.

The number of collisions between particles of the second kind will be
NtvtDpn and so on, if there are more kinds of particles.

Let us now consider a thin stratum of the mixture whose volume is unity.

The resultant momentum of the particles of the first kind which lodge in
it during unit of time is

_dp1

dx '

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 397

The proportion of these which strike particles of the first kind is

The whole momentum of these remains among the particles of the first kind.
The proportion which strike particles of the second kind is

The momentum of these is divided between the striking particles in the ratio

M

of their masses ; so that -., ',, of the whole goes to particles of the first

kind, and -^f — ^> to particles of the second kind.

The effect of these collisions is therefore to produce a force

on particles of the first system, and

on particles of the second system.

The effect of the collisions of those particles of the second system which
strike into the stratum, is to produce a force

on the first system, and

-t

on the second.

The whole effect of these collisions is therefore to produce a resultant force

on the first system,

(35)

I-

Ml + Mi dx \ ' M! + M3

on the second, and so on.

398 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

PROP. XVIII. To find the -mechanical effect of a difference in tJie mean
velocity of translation of two systems of moving particles.

Let F,, Vt be the mean velocities of translation of the two systems
respectively, then ., ' 2, (Vl— F,) is the mean momentum lost by a particle

Jjl j T~ Ixl g

of the first, and gained by a particle of the second at collision. The number
of such collisions in unit of volume is

elt or tpa;
therefore the whole effect of the collisions is to produce a force

on

<>>ft>

the first system, and an equal and opposite force

on unit of volume of the second system.

PROP. XIX. To find the law of diffusion in the case of two gases diffusing
into each other through a plug made of a porous material, as in the case of
the experiments of Graham.

The pressure on each side of the plug being equal, it was found by Graham
that the quantities of the gases which passed in opposite directions through the
plug in the same time were directly as the square roots of their specific gravities.

We may suppose the action of the porous material to be similar to that
of a number of particles fixed in space, and obstructing the motion of the
particles of the moving systems. If .£, is the mean distance a particle of the
first kind would have to go before striking a fixed particle, and Z, the distance
for a particle of the second kind, then the mean paths of particles of each
kind will be given by the equations

, j 1 , ................... (38).

The mechanical effect upon the plug of the pressures of the gases on each side,
and of the percolation of the gases through it, may be found by Props. XVII.
and XVIII. to be

MJTJQ.V. dPl I, dp, I,
~ZT " ^ A ~ dx L, ~

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 399

and this must be zero, if the pressures are equal on each side of the plug.

Now if Qlf Qt be the quantities transferred through the plug by the mean
motion of translation, Ql = plV1 = M,N1V1; and since by Graham's law

Qt~

we shall have

MjN^ F, = — 3fjJV,v, F, = U suppose ;

and since the pressures on the two sides are equal, -r-2= ~~J^> an(i the

way in which the equation of equilibrium of the plug can generally subsist is

when Z, = Lt and /, = ?,. This implies that A = C and B?= D. Now we know

^

that v,'B = v?C. Let K= 3 — ; , then we shall have

v,

...................... (40),

and - = = K (v1pl + v,p,) + - ........................... (41).

The diffusion is due partly to the motion of translation, and partly to that of
agitation. Let us find the part due to the motion of translation.

The equation of motion of one of the gases through the plug is found by
adding the forces due to pressures to those due to resistances, and equating
these to the moving force, which in the case of slow motions may be neglected
altogether. The result for the first is

+N.BP.V, -fc (v,- r,)+-=o ...... (42).

Making use of the simplifications we have just discovered, this becomes

whence U= -df _ .................. (44).

dx v ''

400 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

whence the rate of diffusion due to the motion of translation may be found ; for

<?, = ^, and <?,= -^ ......................... (45).

"i "i

To find the diffusion due to the motion of agitation, we must find the
value of qt.

L d

v, dx

Similarly, ?,= + {l +ATr, (ft+jv)} .................. (47).

The whole diffusions are <?, + <?! and Qt + qv The values of qt and qt have a

term not following Graham's law of the square roots of the specific gravities,

but following the law of equal volumes. The closer the material of the plug,
the less will this term affect the result.

Our assumptions that the porous plug acts like a system of fixed particles,
and that Graham's law is fulfilled more accurately the more compact the
material of the plug, are scarcely sufficiently well verified for the foundation of
a theory of gases ; and even if we admit the original assumption that they are
systems of moving elastic particles, we have not very good evidence as yet for
the relation among the quantities A, B, C, and D.

PROP. XX. To find the rate of diffusion between two vessels connected by a
tube.

When diffusion takes place through a large opening, such as a tube con-
necting two vessels, the question is simplified by the absence of the porous
diffusion plug; and since the pressure is constant throughout the apparatus, the
volumes of the two gases passing opposite ways through the tube at the same
time must be equal Now the quantity of gas which passes through the tube
is due partly to the motion of agitation as in Prop. XIV., and partly to the,
mean motion of translation as in Prop. XV.

ILLUSTRATIONS OP THE DYNAMICAL THEORY OF GASES. 401

Let us suppose the volumes of the two vessels to be a and b, and the
length of the tube between them c, and its trans-
verse section s. Let « be filled with the first gas,
and b with the second at the commencement of
the experiment, and let the pressure throughout
the apparatus be P.

Let a volume y of the first gas pass from a to b, and a volume y' of the
second pass from b to a ; then if pl and j>, represent the pressures in a due
to the first and second kinds of gas, and p\ and p\ the same in the vessel b,

_- p - P n'~.p y> -

Pl~ ~a ' Pt~ a ' Pl~b ' 2'~ b
Since there is still equilibrium,

which gives y = y and pl +pt = P =p\ +p'.} ..................... (49).

The rate of diffusion will be + -^ for the one gas, and — -~- for the other,
measured in volume of gas at pressure P.

Now the rate of diffusion of the first gas will be

and that of the second,

JL A

_dy_ *v'dx

We have also the equation, derived from Props. XVI. and XVII.,

wMW-V^O ..... (52).

From these three equations we can eliminate Vl and V,, and find -:- in
terms of p and -~- , so that we may write

-

at \I' dx

VOL. I. 51

402 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

Since the capacity of the tube is small compared with that of the vessels,
we may consider -3- constant through the whole length of the tube. We may

then solve the differential equation in p and x ; and then making p =p, when
x = 0, ahd p=p\ when x = c, and substituting for pt and p\ their values in
terms of y, we shall have a differential equation in y and t, which being solved,
will give the amount of gas diffused in a given time.

The solution of these equations would be difficult unless we assume rela-
tions among the quantities A, B, C, D, which are not yet sufficiently estab-
lished in the case of gases of different density. Let us suppose that in a
particular case the two gases have the same density, and that the four quan-
tities A, B, C, D are all equal.

The volume diffused, owing to the motion of agitation of the particles, is

then

.? dp ,

-*pdxvl>

and that due to the motion of translation, or the interpenetration of the two

gases in opposite streams, is

s dp kl
~Pdxv'

The values of v are distributed according to the law of Prop. IV., so that

the mean value of v is - , and that of is -p-- , that of k being ia*. The

Jir V Jva

diffusions due to these two causes are therefore in the ratio of 2 to 3, and
their sum is

dy_ 4 /2k si dp . }

dt~ s/s/TPdx"
If we suppose -j- constant throughout the tube, or, in other words, if we

Ct7)

regard the motion as steady for a short time, then -/ will be constant and
equal to — ' — — '; or substituting from (48),

whence - y.-i-e-*') ........................... (56).

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 403

By choosing pairs of gases of equal density, and ascertaining the amount
of diffusion in a given time, we might determine the value of I in this expres-
sion. The diffusion of nitrogen into carbonic oxide or of deutoxide of nitrogen
into carbonic acid, would be suitable cases for experiment. The only existing
experiment which approximately fulfils the conditions is one by Graham, quoted
by Herapath from Brande's Quarterly Journal of Science, Vol. xvm. p. 76.

A tube 9 inches long and 0'9 inch diameter, communicated with the
atmosphere by a tube 2 inches long and 0'12 inch diameter; 152 parts of
olefiant gas being placed in the tube, the quantity remaining after four hours

was 99 parts.

/
In this case there is not much difference of specific gravity between the

gases, and we have a = 9x(0'9)2- cubic inches, 6 = 00, c = 2 inches, and

« = (0'12)4 — square inches;

a

= 0-00000256 inch =^^Vo7 inch .................. (58).

PROP. XXI. To find the amount of energy which crosses unit of area in
unit of time when the velocity of agitation is greater on one side of the area
tlian on the other.

The energy of a single particle is composed of two parts, — the vis viva
of the centre of gravity, and the vis viva of the various motions of rotation
round that centre, or, if the particle be capable of internal motions, the vis
viva of these. We shall suppose that the whole vis viva bears a constant
proportion to that due to the motion of the centre of gravity, or

where /8 is a coefficient, the experimental value of which is T634. Substituting
E for M in Prop. XIV., we get for the transference of energy across unit
of area in unit of time,

51—2

404 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

where J is the mechanical equivalent of heat in foot-pounds, and q is the
transfer of heat in thermal units.

Now MN=p, and l — ~r- . so that

Also, if T is the absolute temperature,

1 dT2

(60),
where jp must be measured in dynamical units of force.

Let «/=772 foot-pounds, j> = 2116 pounds to square foot, ^
v=1505 feet per second, 7*= 522 or 62° Fahrenheit; then

where 5 is the flow of heat in thermal units per square foot of area ; and T'
and T are the temperatures at the two sides of a stratum of air x inches thick.

In Prof. Eankine's work on the Steam-engine, p. 259, values of the therum/
resistance, or the reciprocal of the conductivity, are given for various substances
as computed from a Table of conductivities deduced by M. Peclet from experi-
ments by M. Despretz : —

Resistance.

Gold, Platinum, Silver .......... ,.0'0036

Copper ................................ Q'0040

Iron .................................... O'OOOG

Lead .................................. 0*0198

Brick .................................. 0-3306

Air by our calculation ............ 40000

It appears, therefore, that the resistance of a stratum of air to the con-
duction of heat is about 10,000,000 times greater than that of a stratum of

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 405

copper of equal thickness. It would be almost impossible to establish the value
of the conductivity of a gas by direct experiment, as the heat radiated from the
sides of the vessel would be far greater than the heat conducted through the
air, even if currents could be entirely prevented*.

PART III.

ON THE COLLISION OF PERFECTLY ELASTIC BODIES OF ANY FORM.

When two perfectly smooth spheres strike each other, the force which acts
between them always passes through their centres of gravity ; and therefore their
motions of rotation, if they have any, are not affected by the collision, and
do not enter into our calculations. But, when the bodies are not spherical,
the force of compact will not, in general, be in the line joining their centres
of gravity ; and therefore the force of impact will depend both on the motion
of the centres and the motions of rotation before impact, and it will affect
both these motions after impact.

In this way the velocities of the centres and the velocities of rotation
will act and react on each other, so that finally there will be some relation
established between them ; and since the rotations of the particles about their
three axes are quantities related to each other in the same way as the three
velocities of their centres, the reasoning of Prop. IV. will apply to rotation as
well as velocity, and both will be distributed according to the law

dN A7 1 -•-,

-j- =N — 7=6 •'.
ax av/7r

* [Clausius, in the memoir cited in the last foot-note, lias pointed out two oversights in this
calculation. In the first place the numbers have not been properly reduced to English measure,
and have still to be multiplied by '4356, the ratio of the English pound to the kilogramme. The
numbers have, further, been calculated with one hour as the unit of time, whereas Maxwell has
used them as if a second had been the unit. Taking account of these circumstances and using his
own expression for the conduction which differs from (59) only in having ^ in place of J.on the
right-hand side, Clausius finds that the resistance of a stratum of air to the conduction of heat is
1400 times greater than that of a stratum of lead of the same thickness, or about 7000 times greater
than that of copper. Ed.]

406

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES.

Also, by Prop. V., if a: be the average velocity of one set of particles, and y
that of another, then the average value of the sum or difference of the velocities is

from which it is easy to see that, if in each individual case

u = ax + by + cz,

where x, y, z are independent quantities distributed according to the law above
stated, then the average values of these quantities will be connected by the
equation

PROP. XXII. Tivo perfectly elastic bodies of any form strike each other :
their motions before impact, and the line of impact, to find their motions
• if/i'r impact.

Let 37, and 37, be the centres of gravity of the two bodies. 3/,A',, 37,}",,
and 37,Z, the principal axes of the first ; and 37^T,,
37, y, and 37,Z, those of the second. Let 7 be the
point of impact, and 7^,772, the line of impact.

Let the co-ordinates of 7 with respect to 37, be
.ruv,s,, and with respect to 37, let them be x^f,t.

Let the direction-cosines of the line of impact
72,7/2, be /,w,n, with respect to 37,, and Ijnajn^ with
respect to 37,.

Let 37, and 37, be the masses, and ^4,7^,C, and
inertia of the bodies about their principal axes.

Let the velocities of the centres of gravity, resolved in the direction of
the principal axes of each body, be

Ult Vlt Wlt and Ut, Vn Wn before impact,
and Vu Vu W\, and Vu Vu W. after impact.

Let the angular velocities round the same axes be

pt. qa ru and pn q0 rn before impact,
and p'it q\, r\, and p'n q'K r'a after impact.

the moments of

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 407

Let R be the impulsive force between the bodies, measured by the momentum
it produces in each.

Then, for the velocities of the centres of gravity, we have the following
equations :

Z7>ffi + j|, U',= UZ-^ ...................... (62),

with two other pairs of equations in V and W.

The equations for the angular velocities are

R 7?

........... (63),

with two other pairs of equations for q and r.

The condition of perfect elasticity is that the whole vis viva shall be the
same after impact as before, which gives the equation

Ml ( U'\ - U\) + M, ( U'\ - U\) + A, (p'\ -p\) + A, (p'\ -p\] + &c. = 0. . . . (64).

The terms relating to the axis of x are here given ; those relating to y and
z may be easily written down.

Substituting the values of these terms, as given by equations (62) and (63),
and dividing by R, we find

* (U\+ U^-lt(U\ + Ut) + (y1nl-z1m1)(p'1+p1)-(y3n1-z,m2) (p's-hpa) + &c. =0...(65).

Now if vl be the velocity of the striking-point of the first body before
impact, resolved along the line of impact,

vt = 1^ + (y^ - 2,771.) p, + &c. ;

and if we put v, for the velocity of the other striking-point resolved along the
same line, and v\ and v\ the same quantities after impact, we may write,
equation (65),

v1 + v'l-va-'z/2 = 0 .............................. (66),

or vl — v, = v't — v\ ................................. (67),

which shows that the velocity of separation of the striking-points resolved in
the line of impact is equal to that of approach.

408 ILLUSTRATIONS OF THE DYNAMICAL THEORY OF OASES.

Substituting the values of the accented quantities in equation (65) by means
of equations (63) and (64), and transposing terms in It, we find

2 {UJ, - UJ,

(68),

the other terms being related to y and z as these are to x. From this equation
we may find the value of R ; and by substituting this in equations (63), (64),
we may obtain the values of all the velocities after impact.

We may, for example, find the value of U\ from the equation

(69).

TT* I1' *• (y.n.-s.m.)* (y.n. - 2,™,)' } M,
J*\ff+Jl* ~AT ~£T *J /,

= u [ !'- 1 l* i (yini-z>m>)\(y*n>-z>my | jfcc] 3/i

1 \ Jtf, iVf, 4, 4t ' '/ /,

+ 2UJ,- 2p, (y.n, - Z.TO,) + 2p3 (y,nt - z.w,) - &c.

PROP. XXIII. To ^wcZ <Ae relations between the average velocities of trans-
lation and rotation after many collisions among many bodies.

Taking equation (69), which applies to an individual collision, we see that
£T, is expressed as a linear function of Uu Un pt) pu &c., all of which are
quantities of which the values are distributed among the different particles
according to the law of Prop. IV. It follows from Prop. V., that if we square
every term of the equation, we shall have a new equation between the average
itdues of the different quantities. It is plain that, as soon as the required
relations have been established, they will remain the same after collision, so that
we may put U?= U? in the equation of averages. The equation between the
average values may then be written

(. .If. U* - M. Uf) + ( Jf, L? - A jfl (y^ ' + ( Jf. U! - Aff) (y'n* - + &c. = 0.

Now since there are collisions in every possible way, so that the values of
/, m, n, &c. and x, y, z, &c. are infinitely varied, this equation cannot subsist
unless

MM = M,U; = A,P? = A,p? = Ac.

The final state, therefore, of any number of systems of moving particles of
any form is that in which the average vis viva of translation along each of the

ILLUSTRATIONS OF THE DYNAMICAL THEORY OF GASES. 409

three axes is the same in all the systems, and equal to the average vis viva
of rotation about each of the three principal axes of each particle.

Adding the -vires vivce with respect to the other axes, we find that the
whole vis viva of translation is equal to that of rotation in each system of
particles, and is also the same for different systems, as was proved in Prop. VI.

This result (which is true, however nearly the bodies approach the spherical
form, provided the motion of rotation is at all affected by the collisions) seems
decisive against the unqualified acceptation of the hypothesis that gases are such
systems of hard elastic particles. For the ascertained fact that y, the ratio of
the specific heat at constant pressure to that at constant volume, is equal to
T408, requires that the ratio of the whole vis viva to the vis viva of translation
should be

whereas, according to our hypothesis, ^8 = 2.

We have now followed the mathematical theory of the collisions of hard
elastic particles through various cases, in which there seems to be an analogy
with the phenomena of gases. We have deduced, as others have done already,
the relations of pressure, temperature, and density of a single gas. We have
also proved that when two different gases act freely on each other (that is, when
at the same temperature), the mass of the single particles of each is inversely
proportional to the square of the molecular velocity ; and therefore, at equal
temperature and pressure, the number of particles in unit of volume is the same.

We then offered an explanation of the internal friction of gases, and deduced
from experiments a value of the mean length of path of a particle between
successive collisions.

We have applied the theory to the law of diffusion of gases, and, from an
experiment on olefiant gas, we have deduced a value of the length of path not
very different from that deduced from experiments on friction.

Using this value of the length of path between collisions, we found that the
resistance of air to the conduction of heat is 10,000,000 times that of copper, a
result in accordance with experience.

Finally, by establishing a necessary relation between the motions of trans-
lation and rotation of all particles not spherical, we proved that a system of
such particles could not possibly satisfy the known relation between the two
specific heats of all gases.

VOL. I. 52

[From the Philosophical Trantaotioru, MDCCCLX.]

XXI. On the Theory of Compound Colours, and the Relations of the Colours
of the Spectrum. Communicated by Professor STOKES, Sec. RS.

(Received January 5, — Read March 22, 1860.)

§ I. Introduction.

ACCORDING to Newton's analysis of light*, every colour in nature is pro-
duced by the mixture, in various proportions, of the different kinds of light
into which white light is divided by refraction. By means of a prism we may
analyse any coloured light, and determine the proportions in which the different
homogeneous rays enter into it; and by means of a lens we may recombine
these rays, and reproduce the original coloured light.

Newton has also shewnt how to combine the different rays of the spectrum
so as to form a single beam of light, and how to alter the proportions of the
different colours so as to exhibit the result of combining them in any arbitrary
manner.

The number of different kinds of homogeneous light being infinite, and the
proportion in which each may be combined being also variable indefinitely, the
results of such combinations could not be appreciated by the eye, unless the
chromatic effect of every mixture, however complicated, could be expressed in
some simpler form. Colours, as seen by the human eye of the normal type, can
all be reduced to a few classes, and expressed by a few well-known names; and
even those colours which have different names have obvious relations among them-
selves. Every colour, except purple, is similar to some colour of the spectrum \,

* Optici, Book L Part 2, Prop. 7.

t Lectioneg Opticce, Part 2, § 1, pp. 100 to 105; and Optics, Book i. Part 2, Prop. 11.

I Optict, Book i. Part 2, Prop. 4.

ON THE THEORY OF COMPOUND COLOURS. 411

although less intense ; and all purples may be compounded of blue and red,
and diluted with white to any required tint. Brown colours, which at first
sight seem different, are merely red, orange or yellow of feeble intensity, more
or less diluted with white.

It appears therefore that the result of any mixture of colours, however
complicated, may be defined by its relation to a certain small number of
well-known colours. Having selected our standard colours, and determined the
relations of a given colour to these, we have defined that colour completely as
to its appearance. Any colour which has the same relation to the standard
colours, will be identical in appearance, though its optical constitution, as
revealed by the prism, may be very different.

We may express this by saying that two compound' colours may be chro-
matically identical, but optically different. The optical properties of light are
those which have reference to its origin and propagation through media, till it
falls on the sensitive organ of vision ; the chromatical properties of light are
those which have reference to its power of exciting certain sensations of colour,
perceived through the organ of vision.

The investigation of the chromatic relations of the rays of the spectrum
must therefore be -founded upon observations of the apparent identity of com-
pound colours, as seen by an eye either of the normal or of some abnormal
type ; and the results to which the investigation leads must be regarded as
partaking of a physiological, as well as of a physical character, and as indicating
certain laws of sensation, depending on the constitution of the organ of vision,
which may be different in different individuals. We have to determine the
laws of the composition of colours in general, to reduce the number of standard
colours to the smallest possible, to discover, if we can, what they are, and to
ascertain the relation which the homogeneous light of different parts of the
spectrum bears to the standard colours.

§ II. History of the Theory of Compound Colours.

The foundation of the theory of the composition of colours was laid by
Newton*. He first shews that, by the mixture of homogeneal light, colours
may be produced which are "like to the colours of homogeneal light as to
the appearance of colour, but not as to the immutability of colour and consti-

* Optics, Book i. Part 2, Props. 4, 5, 6.

52—2

412 ox THE THEORY OF COMPOUND COLOURS.

tution of light- lied and yellow give an orange colour, which is chromatically
similar to the orange of the spectrum, but optically different, because it is
resolved into its component colours by a prism, while the orange of the spectrum
remains unchanged. When the colours to be mixed lie at a distance from one
another in the spectrum, the resultant appears paler than that intermediate
colour of the spectrum which it most resembles ; and when several are mixed,
the resultant may appear white. Newton* is always careful, however, not to
call any mixture white, unless it agrees with common white light in its optical
as well as its chromatical properties, and is a mixture of all the homogeneal
colours. The theory of compound colours is first presented in a mathematical
form in Prop. 6, "In a mii-tm-^ of prinuiry colours, tin- </iinntity and <jt«iJiti/
of each being given, to know the colour of the corn-pan ml." He divides the
circumference of a circle into seven parts, proportional to the seven musical
intervals, in accordance with his opinion about the proportions of the colours
in the spectrum. At the centre of gravity of each of these arcs he places a
little circle, whose area is proportional to the number of rays of the corre-
sponding colour which enter into the given mixture. The position of the centre
of gravity of all these circles indicates the nature of the resultant colour. A
radius drawn through it points out that colour of the spectrum which it most
resembles, and the distance from the centre determines the fulness of its colour.

With respect to this construction, Newton says, " This rule I conceive
accurate enough for practice, though not mathematically accurate." He gives no
reasons for the different parts of his rule, but we shall find that his method
of finding the centre of gravity of the component colours is completely con-
firmed by my observations, and that it involves mathematically the theory of three
elements of colour ; but that the disposition of the colours on the circumference
of a circle was only a provisional arrangement, and that the true relations of
the colours of the spectrum can only be determined by direct observation.

Young t appears to have originated the theory, that the three elements of
colour are determined as much by the constitution of the sense of sight as by
anything external to us. He conceives that three different sensations may be
excited by light, but that the proportion in which each of the three is excited
depends on the nature of the light. He conjectures that these primary sensa-

• 7th and 8th Letters to Oldenburg.

f Young'* Lectures on Natural Philosophy, Kelland's Edition, )>. 345, or Quarto, 1807, Vol. i.
p. 441 ; see alio Young in Philosophical Transactions, 1801, or Works in Quarto, VoL II. p. 617.

ON THE THEORY OF COMPOUND COLOURS. 413

tions correspond to red, green, and violet. A blue ray, for example, though
homogeneous in itself, he conceives capable of exciting both the green and the
violet sensation, and therefore he would call blue a compound colour, though
the colour of a simple kind of light. The quality of any colour depends,
according to this theory, on the ratios of the intensities of the three sensations
which it excites, and its brightness depends on the sum of these three intensities.
Sir David Brewster, in his paper entitled " On a New Analysis of Solar
Light, indicating three Primary Colours, forming Coincident Spectra of equal
length*," regards the actual colours of the spectrum as arising from the inter-
mixture, in various proportions, of three primary kinds of light, red, yellow,
and blue, each of which is variable in intensity, but uniform in colour, from
one end of the spectrum to the other ; so that every colour in the spectrum
is really compound, and might be shewn to be so if we had the means of
separating its elements.

Sir David Brewster, in his researches, employed coloured media, which,
according to him, absorb the three elements of a single prismatic colour in
different degrees, and change their proportions, so as to alter the colour of the
light, without altering its refrangibility.

In this paper I shall not enter into the very important questions affecting
the physical theory of light, which can only be settled by a careful inquiry
into the phenomena of absorption. The physiological facts, that we have a
threefold sensation of colour, and that the three elements of this sensation are
affected in different proportions by light of different refrangibilities, are equally
true, whether we adopt the physical theory that there are three kinds of light
corresponding to these three colour-sensations, or whether we regard light of
definite refrangibility as an undulation of known length, and therefore variable
only in intensity, but capable of producing different chemical actions on different
substances, of being absorbed in different degrees by different media, and of
exciting in different degrees the three different colour-sensations of the human
eye.

Sir David Brewster has given a diagram of three curves, in which the
base-line represents the length of the spectrum, and the ordinates of the curves
represent, by estimation, the intensities of the three kinds of light at each point
of the spectrum. I have employed a diagram of the same kind to express the

* Transactions of the Royal Society of Edinburgh, Vol. xn. p. 1 23.

414 ON THE THEORY OF COMPOUND COLOURS.

results arrived at in this paper, the ordinutes being made to represent the
intensities of each of the three elements of colour, as calculated from the
experiments.

The most complete series of experiments on the mixture of the colours of
the spectrum, is that of Professor Helmholtz*, of Konigsberg. By using two
slits at right angles to one another, he formed two pure spectra, the fixed
lines of which were seen crossing one another when viewed in the ordinary
way by means of a telescope. The colours of these spectra were thus combined
in every possible way, and the effect of the combination of any two could be
seen separately by drawing the eye back from the eye-piece of the telescope,
when the compound colour was seen by itself at the eye-hole. The proportion
of the components was altered by turning the combined slits round in their
own plane.

One result of these experiments was, that a colour, chromatically identical
with white, could be formed by combining yellow with indigo. M. Helmholtz
was not then able to produce white with any other pair of simple colours, and
considered that three simple colours were required in general to produce white,
one from each of the three portions into which the spectrum is divided by
the yellow and indigo.

Professor Grassmannf shewed that Newton's theory of compound colours
implies that there are an infinite number of pairs of complementary colours in
the spectrum, and pointed out the means of finding them. He also shewed
how colours may be represented by lines, and combined by the method of the
parallelogram.

In a second memoirj, M. Helmholtz describes his method of ascertaining
these pairs of complementary colours. He formed a pure spectrum by means
of a slit, a prism, and a lens ; and in this spectrum he placed an apparatus
having two parallel slits which were capable of adjustment both in position
and breadth, so as to let through any two portions of the spectrum, in any
proportions. Behind this slit, these rays were united in an image of the prism,
which was received on paper. By arranging the slits, the colour of this image
may be reduced to white, and made identical with that of paper illuminated with
white light. The wave-lengths of the component colours were then measured by
observing the angle of diffraction through a grating. It was found that the

* PoggendorfTs Annalen, Band LXXXVIL (Philosophical Magazine, 1852, December).

t Ibid. Band LXXXIX. (Philosophical Magazine, 1854, April). J Ibid. Band xciv.

ON THE THEORY OF COMPOUND COLOURS. 415

colours from red to green-yellow (X = 2082) were complementary to colours ranging
from green-blue (X=1818) to violet, and that the colours between green-yellow
and green-blue have no homogeneous complementaries, but must be neutralized
by mixtures of red and violet.

M. Helmholtz also gives a provisional diagram of the curve formed by the
spectrum on Newton's diagram, for which his experiments did not furnish him
with the complete data.

Accounts of experiments by myself on the mixture of artificial colours by
rapid rotation, may be found in the Transactions of the Royal Society of
Edinburgh, Vol. xxi. Pt. 2 (1855); in an appendix to Professor George Wilson's
work on Colour-Blindness ; in the Report of the British Association for 1856,
p. 12; and in the Philosophical Magazine, July 1857, p. 40' These experiments
shew that, for the normal eye, there are three, and only three, elements of
colour, and that in the colour-blind one of these is absent. They also prove
that chromatic observations may be made, both by normal and abnormal eyes,
with such accuracy, as to warrant the employment of the results in the calcu-
lation of colour-equations, and in laying down colour-diagrams by Newton's rule.

The first instrument which I made (in 1852) to examine the mixtures of
the colours of the spectrum was similar to that which I now use, but smaller,
and it had no constant light for a term of comparison. The second was 6^ feet
long, made in 1855, and shewed two combinations of colour side by side. I have
now succeeded in making the mixture much more perfect, and the comparisons
more exact, by using white reflected light, instead of the second compound
colour. An apparatus in which the light passes through the prisms, and is
reflected back again in nearly the same path by a concave mirror, was shewn
by me to the British Association in 1856. It has the advantage of being
portable, and need not be more than half the length of the other, in order
to produce a spectrum of eqTial length. I am so well satisfied with the working
of this form of the instrument, that I intend to make use of it in obtaining
equations from a greater variety of observers than I could meet with when I
was obliged to use the more bulky instrument. It is difficult at first to get
the observer to believe that the compound light can ever be so adjusted as to
appear to his eyes identical with the white light in contact with it. He has to
learn what adjustments are necessary to produce the requisite alteration under
all circumstances, and he must never be satisfied till the two parts of the
field are identical in colour and illumination. To do this thoroughly, implies

416 "N THE THhXWY OK COMPOUND COLOURS.

not merely good eyes, but a power of judging as to the exact nature of the
difference between two very pale and nearly identical tints, whether they differ
in the amount of red, green, or blue, or in brightness of illumination.

In the following paper I shall first lay down the mathematical theory of
Newton's diagram, with its relation to Young's theory of the colour-sensation.
I shall then describe the experimental method of mixing the colours of the
spectrum, and determining the wave-lengths of the colours mixed. The results
of my experiments will then be given, and the chromatic relations of the
spectrum exhibited in a system of colour-equations, in Newton's diagram, and
in three curves of intensity, as in Brewster's diagram. The differences between
the results of two observers will then be discussed, shewing on what they
depend, and in what way such differences may affect the vision of persons
otherwise free from defects of sight.

§ III. Mathematical Tlieory of Newton's Diagram of Colours.

Newton's diagram is a plane figure, designed to exhibit the relations of
colours to each other.

Every point in the diagram represents a colour, simple or compound, and
we may conceive the diagram itself so painted, that every colour is found at
its corresponding point. Any colour, differing only in quantity of illumination
from one of the colours of the diagram, is referred to it as a unit, and is
measured by the ratio of the illumination of the given colour to that of the
corresponding colour in the diagram. In this way the quantity of a colour is
estimated. The resultant of mixing any two colours of the diagram is found
by dividing the line joining them inversely as the quantity of each ; then, if
the sum of these quantities is unity, the resultant will have the illumination
as well as the colour of the point so found ; but if the sum of the components
is different from unity, the (fuantity of the resultant will be measured by the
sum of the components.

This method of determining the position of the resultant colour is mathe-
matically identical with that of finding the centre of gravity of two weights,
and placing a weight equal to their sum at the point so found. We shall
therefore speak of the resultant tint an the sum of its components placed at
their centre of gravity.

ON THE THEORY OF COMPOUND COLOURS. 417

By compounding this resultant tint with some other colour, we may find the
position of a mixture of three colours, at the centre of gravity of its components ;
and by taking these components in different proportions, we may obtain colours
corresponding to every part of the triangle of which they are the angular points.
In this way, by taking any three colours we should be able to construct a
triangular portion of Newton's diagram by painting it with mixtures of the three
colours. Of course these mixtures must be made to correspond with optical
mixtures of light, not with mechanical mixtures of pigments.

Let us now take any colour belonging to a point of the diagram outside
this triangle. To make the centre of gravity of the three weights coincide with
this point, one or more of the weights must be made negative: This, though
following from mathematical principles, is not capable of direct physical inter-
pretation, as we cannot exhibit a negative colour.

The equation between the three selected colours, x, y, z, and the new colour
u, may in the first case be written

u = x+y + z (1),

x, y, z being the quantities of colour required to produce u. In the second case
suppose that z must be made negative,

u = x + y — z (2).

As we cannot realize the term —z as a negative colour, we transpose it to the
other side of the equation, which then becomes

u + z = x + y (3),

which may be interpreted to mean, that the resultant tint, u + z, is identical
with the resultant, x + y. "We thus find a mixture of the new colour with one
of the selected colours, which is chromatically equivalent to a mixture of the
other two selected colours.

When the equation takes the form

u = x-y-z (4),

two of the components being negative, we must transpose them thus,

u + y + z = x (5),

which means that a mixture of certain proportions of the new colour and two

of the three selected, is chromatically equivalent to the third. We may thus in

all cases find the relation between any three colours and a fourth, and exhibit

VOL. I. 53

418 ON THE THEORY OF COMPOUND COLOURS.

this relation in a form capable of experimental verification ; and by proceeding
in this way we may map out the positions of all colours upon Newton's diagram.
Every colour in nature will then be defined by the position of the corresponding
colour in the diagram, and by the ratio of its illumination to that of the
colour in the diagram.

§ IV. Method of representing Colours by Straight Lines drawn from a Point.

To extend our ideas of the relations of colours, let us form a new geome-
trical conception by the aid of solid geometry.

Let us take as origin any point not in the plane of the diagram, and let
us draw lines through this point to the different points of the diagram ; then
the direction of any of these lines will depend upon the position of the point
of the diagram through which it passes, so that we may take this line as the
representative of the corresponding colour on the diagram.

In order to indicate the quantity of this colour, let it be produced beyond
the plane of the diagram in the same ratio as the given colour exceeds in
illumination the colour on the diagram. In this way every colour in nature will
be represented by a line drawn through the origin, whose direction indicates
the quality of the colour, while its length indicates its quantity.

Let us find the resultant of two colours by this
method. Let 0 be the origin and AB be a section
of the plane of the diagram by that of the paper.
Let OP, OQ be lines representing colours, A, B the

OP
corresponding points in the diagram ; then the quantity of P will be J^A =P>

and that of Q will be 775 = 2- The resultant of these will be represented in

the diagram by the point C, where AC: CB :: q : p, and the quantity of the
resultant will be p + q, so that if we produce OC to R, so that OR = (p + q)OC,
the line OR will represent the resultant of OP and OQ in direction and
magnitude. It is easy to prove, from this construction, that OR is the diagonal
of the parallelogram of which OP and OQ are two sides. It appears therefore
that if colours are represented in quantity and quality by the magnitude and
direction of straight lines, the rule for the composition of colours is identical

ON THE THEORY OF COMPOUND COLOURS. 419

with that for the composition of forces in mechanics. This analogy has been
well brought out by Professor Grassmann in Poggendorff's Annalen, Bd. LXXXIX.

We may conceive an arrangement of actual colours in space founded upon
this construction. Suppose each of these radiating lines representing a given
colour to be itself illuminated with that colour, the brightness increasing from
zero at the origin to unity, where it cuts the plane of the diagram, and
becoming continually more intense in proportion to the distance from the origin.
In this way every colour in nature may be matched, both in quality and
quantity, by some point in this coloured space.

If we take any three lines through the origin as axes, we may, by co-ordi-
nates parallel to these lines, express the position of any point in space. That
point will correspond to a colour which is the resultant of the three colours
represented by the three co-ordinates.

This system of co-ordinates is an illustration of the resolution of a colour
into three components. According to the theory of Young, the human eye is
capable of three distinct primitive sensations of colour, which by their composition
in various proportions, produce the sensations of actual colour in all their varieties.
Whether any kinds of light have the power of exciting these primitive sensations
separately, has not yet been determined.

If colours corresponding to the three primitive sensations can be exhibited,
then all colours, whether produced by light, disease, or imagination, are com-
pounded of these, and have their places within the triangle formed by joining
the three primaries. If the colours of the pure spectrum, as laid down on the
diagram, form a triangle, the colours at the angles may correspond to the primitive
sensations. If the curve of the spectrum does not reach the angles of the circum-
scribing triangle, then no colour in the spectrum, and therefore no colour in
nature, corresponds to any of the three primary sensations.

The only data at present existing for determining the primary colours, are
derived from the comparison of observations of colour-equations by colour-blind,
and by normal eyes. The colour-blind equations differ from the others by the
non-existence of one of the elements of colour, the relation of which to known
colours can be ascertained. It appears, from observations made for me by two
colour-blind persons*, that the elementary sensation which they do not possess
is a red approaching to crimson, lying beyond both vermilion and carmine. These

* Transactions of the Royal Society of Edinburgh, VoL XXL Pt 2, p. 286.

53—2

420 ON THE THEORY OF COMPOUND COLOURS.

observations are confirmed by those of Mr Pole, and by others which I have
obtained since. I have hopes of being able to procure a set of colour-blind
equations between the colours of the spectrum, which will indicate the missing
primary in a more exact manner.

The experiments which I am going to describe have for their object the
determination of the position of the colours of the spectrum upon Newton's
diagram, from actual observations of the mixtures of those colours. They were
conducted in such a way, that in every observation the judgment of the observer
was exercised upon two parts of an illuminated field, one of which was so
adjusted as to be chromatically identical with the other, which, during the whole
series of observations, remained of one constant intensity of white. In this way
the effects of subjective colours were entirely got rid of, and all the observa-
tions were of the same kind, and therefore may claim to be equally accurate ;
which is not the case when comparisons are made between bright colours of
different kinds.

The chart of the spectrum, deduced from these observations, exhibits the
colours arranged very exactly along two sides of a triangle, the extreme red and
violet forming doubtful portions of the third side. This result greatly simplifies
the theory of colour, if it does not actually point out the three primary colours
themselves.

§ V. Description of an • Instrument for making definite Mixtures of the

Colours of the Spectrum.

The experimental method which I have used consists in forming a combi-
nation of three colours belonging to different portions of the spectrum, the quantity
of each being so adjusted that the mixture shall be white, and equal in intensity
to a given white. Fig. 1, Plate VI. p. 444, represents the instrument for
making the observations. It consists of two tubes, or long boxes, of deal, of
rectangular section, joined together at an angle of about 100°.

The part AK is about five feet long, seven inches broad, and four deep ;
KN is about two feet long, five inches broad, and four deep ; BD is a partition
parallel to the side of the long box. The whole of the inside of the instrument
is painted black, and the only openings are at the end AC, and at E. At the
angle there is a lid, which is opened when the optical parts have to be adjusted
or cleaned.

ON THE THEOKY OF COMPOUND COLOURS. 421

At E is a fine vertical slit ; L is a lens ; at P there are two equilateral
prisms. The slit E, the lens L, and the prisms P are so adjusted, that when
light is admitted at E a pure spectrum is formed at AB, the extremity of the
long box. A mirror at M is also adjusted so as to reflect the light from E
along the narrow compartment of the long box to EG. See Fig. 3.

At AB is placed the contrivance shewn in Fig. 2, Plate I. A'B' is a rect-
angular frame of brass, having a rectangular aperture of 6 x 1 inches. On this
frame are placed six brass sliders, X, Y, Z. Each of these carries a knife-edge
of brass in the plane of the surface of the frame.

These six moveable knife-edges form three slits, X, Y, Z, which may be
so adjusted as to coincide with any three portions of the pure spectrum formed
by light from E. The intervals behind the sliders are closdd by hinged shutters,
which allow the sliders to move without letting light pass between them.

The inner edge of the brass frame is graduated to twentieths of an inch,
so that the position of any slit can be read off. The breadth of the slit is
ascertained by means of a wedge-shaped piece of metal, six inches long, and
tapering to a point from a breadth of half an inch. This is gently inserted into
each slit, and the breadth is determined by the distance to which it enters, the
divisions on the wedge corresponding to the 200th of an inch difference in
breadth, so that the unit of breadth is '005 inch.

Now suppose light to enter at E, to pass through the lens, and to be
refracted by the two prisms at P ; a pure spectrum, shewing Fraunhofer's lines,
is formed at AB, but only that part is allowed to pass which falls on the three
slits X, Y, Z. The rest is stopped by the shutters. Suppose that the portion
falling on X belongs to the red part of the spectrum ; then, of the white light
entering at E, only the red will come through the slit X. If we were to admit
red light at X it would be refracted to E, by the principle in Optics, that the
course of any ray may be reversed. If, instead of red light, we were to admit
white light at X, still only red light would come to E ; for all other light
would be either more or less refracted, and would not reach the slit at E.
Applying the eye at the slit E, we should see the prism P uniformly illuminated
with red light, of the kind corresponding to the part of the spectrum which
falls on the slit X when light is admitted at E.

Let the slit Y correspond to another portion of the spectrum, say the green ;
then, if white light is admitted at Y, the prism, as seen by an eye at E, will
be uniformly illuminated with green light ; and if white light be admitted at X

422 ON THE THEORY OF COMPOUND COLOURS.

and Y simultaneously, the colour seen at E will be a compound of red and green,
the proportions depending on the breadth of the slits and the intensity of the
light which enters them. The third slit Z, enables us to combine any three kinds
of light in any given proportions, so that an eye at E shall see the face of the
prism at P uniformly illuminated with the colour resulting from the combination
of the three. The position of these three rays in the spectrum is found by
admitting the light at E, and comparing the position of the slits with the
position of the principal fixed lines ; and the breadth of the slits is determined
by means of the wedge.

At the same time white light is admitted through BC to the mirror of black
glass at M, whence it is reflected to E, past the edge of the prism at P, so that
the eye at E sees through the lens a field consisting of two portions, separated
by the edge of the prism; that on the left hand being compounded of three
colours of the spectrum refracted by the prism, while that on the right hand is
white light reflected from the mirror. By adjusting the slits properly, these two
portions of the field may be made equal, both in colour and brightness, so that
the edge of the prism becomes almost invisible.

In making experiments, the instrument was placed on a table in a room
moderately lighted, with the end AB turned towards a large board covered with
white paper, and placed in the open air, so as to be uniformly illuminated by
the sun. In this way the three slits and the mirror M were all illuminated
with white light of the same intensity, and all were affected in the same ratio
by any change of illumination ; so that if the two halves of the field were
rendered equal when the sun was under a cloud, they were found nearly correct
when the sun again appeared. No experiments, however, were considered good
unless the sun remained uniformly bright during the whole series of experiments.

After each set of experiments light was admitted at E, and the position of
the fixed lines D and F of the spectrum was read off on the scale at AB. It
was found that after the instrument had been some time in use these positions
were invariable, shewing that the eye-hole, the prisms, and the scale might be
considered as rigidly connected.

ON THE THEORY OF COMPOUND COLOURS. 423

§ VI. Method of determining the Wave-length corresponding to any point
of the Spectrum on the Scale AB.

Two plane surfaces of glass were kept apart by two parallel strips of gold-
beaters' leaf, so as to enclose a stratum of air of nearly uniform thickness. Light
reflected from this stratum of air was admitted at E, and the spectrun formed
by it was examined at AB by means of a lens. This spectrum consists of a
large number of bright bands, separated by dark spaces at nearly uniform intervals,
these intervals, however, being considerably larger, as we approach the violet end
of the spectrum.

The reason of these alternations of brightness is easily explained. By the
theory of Newton's rings, the light reflected from a stratum of air consists of
two parts, one of which has traversed a path longer than that of the other, by
an interval depending on the thickness of the stratum and the angle of incidence.
Whenever the interval of retardation is an exact multiple of a wave-length, these
two portions of light destroy each other by interference ; and when the interval
is an odd number of half wave-lengths, the resultant light is a maximum.

In the ordinary case of Newton's rings, these alternations depend upon the
varying thickness of the stratum ; while in this case a pencil of rays of different
wave-lengths, but all experiencing the same retardation, is analysed into a spectrum,
in which the rays are arranged in order of their respective wave-lengths. Every
ray whose wave-length is an exact submultiple of the retardation will be destroyed
by interference, and its place will appear dark in the spectrum ; and there will
be as many dark bands seen as there are rays whose wave-lengths fulfil this
condition.

If, then, we observe the positions of the dark bands on the scale AB,
the wave-lengths corresponding to these positions will be a series of submultiples
of the retardation.

Let us call the first dark band visible on the red side of the spectrum zero,
and let us number them in order 1, 2, 3, &c. towards the violet end. Let N
be the number of undulations corresponding to the band zero which are con-
tained in the retardation R; then if n be the number of any other band, N+n
will be the number of the corresponding wave-lengths in the retardation, or in
symbols,

R = (N+n)\ (6).

424

ON THE THEORY OF COMPOUND COLOURS.

Now observe the position of two of Fraunhofer's fixed lines with respect to
the dark bands, and let n,, n, be their positions expressed in the number of
bands, whole or fractional, reckoning from zero. Let X,, X, be the wave-lengths
of these fixed lines as determined by Fraunhofer, then

R = (N+nl)\ = (N+nt)^) (7);

whence ^^^i^^^^-n, (8),

A! — A, Aj — A,

and . JR = ^X1AS (9).

Having thus found N and R, we may find the wave-length corresponding to
the dark band n from the formula

In my experiments the line D corresponded with the seventh dark band, and
F was between the 15th and 16th, so that n,= 157. Here then for D,

n, = 7, X, = 2175l . ,, , „ , . .

•t P r, 'in rraunhoiers measure (H),

and for F, w,= 157, X,= 1794J

whence we find N=U, 72=89175 (12).

There were 22 bands visible, corresponding to 22 different positions on the
scale AS, as determined 4th August, 1859.

n =

Band. Scale.
17
19
21*

2
3
4
6
6
7
8

26
28}
31
33}

TABLE I.

Band. Scale.
n= 9 36
10 39
42
45
48

11
12
13
14
15

51

54

Band. Scale.

n= 16
17
18
19
20
21
22

57
61
65
69
73
77
82

Sixteen equidistant points on the scale were chosen for standard colours
in the experiments to be described. The following Table gives the reading on
the scale AB, the value of N+n, and the calculated wave-length for each of
these: —

ON THE THEORY OF COMPOUND COLOURS. 425

TABLE II.

Scale. (N+n). Wave-length. Colour.

20 36-4 2450 Red.

24 38-3 2328 Scarlet

28 39-8 2240 Orange.

32 41-4 2154 Yellow.

36 42-9 2078 Yellow-Green.

40 44-3 2013 Green.

44 45-7 1951 Green.

48 47 O 1879 Bluish green.

52 48-3 1846 Blue-green.

66 49-6 1797 Greenish blue.

60 50-8 1755 Blua '

64 51-8 1721 Blue.

68 52-8 1688 Blue.

72 53-7 1660 Indigo.

76 54-7 1630 Indigo.

80 55-6 1604 Indigo.

Having thus selected sixteen distinct points of the spectrum on which to
operate, and determined their wave-lengths and apparent colours, I proceeded
to ascertain the mathematical relations between these colours in order to lay
them down on Newton's diagram. For this purpose I selected three of these
as points of reference, namely, those at 24, 44, and 68 of the scale. I chose
these points because they are well separated from each other on the scale, and
because the colour of the spectrum at these points does not appear to the eye
to vary very rapidly, either in hue or brightness, in passing from one point to
another. Hence a small error of position will not make so serious an alteration
of colour at these points, as if we had taken them at places of rapid variation ;
and we may regard the amount of the illumination produced by the light
entering through the slits in these positions as sensibly proportional to the
breadth of the slits.

(24) corresponds to a bright scarlet about one-third of the distance from
C to D; (44) is a green very near the line E ; and (68) is a blue about one-
third of the distance from F to G.

VOL. i. 54

ON THE THEORY OF COMPOUND COLOURS.

§ VII. Method of Observation.

The instrument is turned with the end AB towards a board, covered with
white paper, and illuminated by sunlight. The operator sits at the end AB, to
move the sliders, and adjust the slits ; and the observer sits at the end E,
which is shaded from any bright light. The operator then places the slits so
that their centres correspond to the three standard colours, and adjusts their
breadths till the observer sees the prism illuminated with pure white light of
the same intensity with that reflected by the mirror M. In order to do this,
the observer must tell the operator what difference he observes in the two halves
of the illuminated field, and the operator must alter the breadth of the slits
accordingly, always keeping the centre of each slit at the proper point of the
scale. The observer may call for more or less red, blue or green ; and then
the operator must increase or diminish the width of the slits X, Y, and Z
respectively. If the variable field is darker or lighter than the constant field,
the operator must widen or narrow all the slits in the same proportion. When
the variable part of the field is nearly adjusted, it often happens that the
constant white light from the mirror appears tinged with the complementary
colour. This is an indication of what is required to make the resemblance of
the two parts of the field of view perfect. When no difference can be detected
between the two parts of the field, either in colour or in brightness, the observer
must look away for some time, to relieve the strain on the eye, and then look
again. If the eye thus refreshed still judges the two parts of the field to be
equal, the observation may be considered complete, and the operator must measure
the breadth of each slit by means of the wedge, as before described, and write
down the result as a colour-equation, thus —

Oct. 18, J. 18'5(24) + 27(44) + 37(68) = W* (13).

This equation means that on the 18th of October the observer J. (myself) made
an observation in which the breadth of the slit X was 18 '5, as measured by
the wedge, while its centre was at the division (24) of the scale ; that the breadths
of }' and Z were 27 and 37, and their positions (44) and (68) ; and that the
illumination produced by these slits was exactly equal, in my estimation as an
observer, to the constant white W.

ON THE THEOBY OF COMPOUND COLOURS. 427

The position of the slit X was then shifted from (24) to (28), and when
the proper adjustments were made, I found a second colour-equation of this form —

Oct. 18, J. 16 (28) + 21 (44) + 37 (68) = W (14).

Subtracting one equation from the other and remembering that the figures in
brackets are merely symbols of position, not of magnitude, we find

16(28) = 18'5 (24) + 6 (44) (15),

shewing that (28) can be made up of (24) and (44), in the proportion of 18'5
to 6.

In this way, by combining each colour with two standard colours, we may
produce a white equal to the constant white. The red and yellow colours from
(20) to (32) must be combined with green and blue, the greens from (36) to (52)
with red and blue, and the blues from (56) to (80) with red and green.

The following is a specimen of an actual series of observations made in this
way by another observer (K.) : —

TABLE III.

Oct 13, 1859. Observer (K.).

(A"> (Y) (Z)

18J(24) + 32i(44) + 32 (68) = W*.

17J(24) + 32J(44) + 63 (80) = W.

18 (24) + 32J(44) + 35 (72) = \V.

19 (24) + 32 (44) + 31£(68) = W*.

19 (24) + 30J(44) + 35 (64) = W.

20 (24) + 23 (44) + 39 (60) = W.

21 (24) + 14 (44) + 58 (56) = W.

22 (24) + 62 (52) + 11 (68) -W.

22 (24) + 42 (48)+29J(68) = W.

19 (24) + 31$(44) + 33 (68) = W*.

16 (24) + 28 (40) + 32£(68) = W.
6 (24) + 27 (36) + 32J(68) = W.

23 (32)+llJ(44) + 32J(68) = W.

17 (28) + 26 (44) + 32J(68) = W.

20 (24) + 33J(44) + 32J(68) = W*.
46 (20) + 33 (44) + 30 (68) = W.

The equations marked with an asterisk (*) are those which involve the
three standard colours, and since every other equation must be compared with
them, they must be often repeated.

54—2

428 ON THE THEORY OF COMPOUND COLOURS.

The following Table contains the means of four sets of observations by the
same observer (K.) : —

TABLE IV. (K.)

44-3 (20) + 31 -0(44) + 27-7 (68) = W.
16-1 (28) + 25-6 (44) + 30-6 (68) = W.
220 (32) + 12-1 (44) + 30-6 (68) = W.
6-4 (24) + 25-2 (36) + 31-3 (68) = W.
15-3 (24) + 26 -0(40) + 30 -7 (68) = W.
19-8 (24) + 350 (46) + 30-2 (68) = W.
21 -2 (24) + 41 -4 (48) + 27O (68) = W.
220 (24) + 620 (52) + 130 (68) = W.
21-7 (24) + 10-4 (44)+ 61 -7(56) = W.
20-5 (24) + 23-7 (44) + 40-5 (60) = W.
19-7 (24) + 30-3 (44) + 33-7 (64) = W.
180 (24) + 31 -2 (44) + 32-3 (72) = W.
17-5 (24) + 30-7 (44) + 440 (76) = W.
1 8 -3 (24) + 33 -2 (44) + 63 7 (80) = W.

§ VIII. Determination of the Average Error in Observations of different kinds.

In order to estimate the degree of accuracy of these observations, I have
taken the differences between the values of the three standard colours as
originally observed, and their means as given by the above Table. The sum
of all the errors of the red (24) from the means, was 31'1, and the number
of observations was 42, which gives the average error 74.

The sum of errors in green (44) was 48 '0, and the number of observa-
tions 31, giving a mean error 1'55.

The sum of the errors in blue (68) was 46'9, and the number of observa-
tions 35, giving a mean error T16.

It appears therefore that in the observations generally, the average error
does not exceed T5 ; and therefore the results, if confirmed by several obser-
vations, may safely be trusted to that degree of accuracy.

The equation between the three standard colours was repeatedly observed,
in order to detect any alteration in the character of the light, or any other
change. of condition which would prevent the observations from being comparable
with one another ; and also because this equation is used in the reduction of

ON THE THEORY OF COMPOUND COLOURS.

429

all the others, and therefore requires to be carefully observed. There are twenty
observations of this equation, the mean of which gives

18-6 (24) + 31-4 (44) + 30-5 (68) = W* (16)

as the standard equation.

We may use the twenty observations of this equation as a means of
determining the relations between the errors in the different colours, and thus
of estimating the accuracy of the observer in distinguishing colours.

The following Table gives the result of these operations, where R stands
for (24), G for (44), and B for (68):—

TABLE V. — Mean Errors in the Standard Equation.

(G-B)=-99
(B-R)=-85
(R-G)=-86

(G + B) = 2-31

(B+R) = l-59

R + G) = l-57

(R)= -54
(G) = l-22
(B) = M5

The first column gives the mean difference between the observed value of
each of the colours and the mean of all the observations. The second column
shews the average error of the observed differences between the values of the
standards, from the mean value of those differences. The third column shews
the average error of the sums of two standards, from the mean of such sums.
The fourth column gives the square root of the sum of the squares of the
quantities in the first column. I have also given the average error of the

sum of R, G and B, from its mean value, and the value of ./R'

It appears from the first column that the red is more accurately observed
than the green and blue.

§ IX. Relative Accuracy in Observations of Colour and of Brightness.

If the errors in the different colours occurred perfectly independent of each
other, then the probable mean error in the sum or difference of any two colours
would be the square root of the sum of their squares, as given in the fourth
column. It will be seen, however, that the number in the second column is
always less, and that in the third always greater, than that in the fourth ;
shewing that the errors are not independent of each other, but that positive
errors in any colour coincide more often with positive than with negative errors

430 ON THE THEORY OF COMPOUND COLOURS.

in another colour. Now the hue of the resultant depends on the ratios of the
components, while its brightness depends on their sum. Since, therefore, the
difference of two colours is always more accurately observed than their sum,
variations of colour are more easily detected than variations in brightness, and
the eye appears to be a more accurate judge of the identity of colour of the
two parts of the field than of their equal illumination. The same conclusion may
be drawn from the value of the mean error of the sum of the three standards,
which is 2 '6 7, while the square root of the sum of the squares of the errors
is 176.

§ X. Reduction of the Observations.

By eliminating W from the equations of page 428 by means of the standard
equation, we obtain equations involving each of the fourteen selected colours of
the spectrum, along with the three standard colours; and by transposing the
selected colour to one side of the equation, we obtain its value in terms of
the three standards. If any of the terms of these equations are negative, the
equation has no physical interpretation as it stands, but by transposing the
negative term to the other side it becomes positive, and then the equation may
be verified.

The following Table contains the values of the fourteen selected tints in
terms of the standards. To avoid repetition, the symbols of the standard colours
are placed at the head of each column.

TABLE VI.

Observer (K.). (24.)

(44.)

(68.)

44-3(20)= 18-6

+ 0-4

+ 2-8

16-1(28)= 18-6

+ 5-8

- 0-1

22-0(32)= 18-6

+ 19-3

- 0-1

25-2(36)= 12-2

+ 31-4

- 0-8

26K)(40) = 3-3

+ 31-4

- 0-2

35O(46) = - 1-2

+ 31-4

+ 0-3

41-4(48)=- 2-6

+ 31-4

+ 3-5

620(52)= - 3-4

+ 31-4

+ 17-5

61-7(56)= - 3-1

+ 21-0

+ 30-5

40-5(60)-- 1-9

+ 77

+ 30-5

337(64)=- 1-1

+ 1-1

+ 30-5

32-3(72)= + 0-6

+ 0-2

+ 30-5

44-0(76)= + 1-1

+ 0-7

+ 30-5

63-7(80)= + 0-3

- 1-8

+ 30-5

ON THE THEORY OF COMPOUND COLOURS. 431

From these equations we may lay down a chart of the spectrum on Newton's
diagram by the following method : — Take any three points, A, B, C, and let A
represent the standard colour (24), B (44), and C (68). Then, to find the position
of any other colour, say (20), divide AC in P so that (18 '6) AP=(2-8)PC, and
then divide BP in Q so that (l8-6 + 2'8)P# = (0-4) QB. At the point Q the
colour corresponding to (20) must be placed. In this way the diagram of fig. 4,
Plate VI., p. 444, has been constructed from the observations of all the colours.

§ XL TJie Spectrum as laid down on Newton's Diagram.

The curve on which these points lie has this striking feature, that two
portions of it are nearly, if not quite, straight lines. One of these portions
extends from (24) to (46), and the other from (48) to (64). The colour (20)
and those beyond (64), are not far from the line joining (24) and (68). The
spectrum, therefore, as exhibited in Newton's diagram, forms two sides of a
triangle, with doubtful fragments of the third side. Now if three colours in
Newton's diagram lie in a straight line, the middle one is a compound of the
two others. Hence all the colours of the spectrum may be compounded of
those which lie at the angles of this triangle. These correspond to the following
colours : —

TABLE VII.

Index Wave-length

in water.

1-747
1-435
1-282

All the other colours of the spectrum may be produced by combinations of
these; and since all natural colours are compounded of the colours of the spec-
trum, they may be compounded of these three primary colours. I have strong
reason to believe that these are the three primary colours corresponding to three
modes of sensation in the organ of vision, on which the whole system of colour,
as seen by the normal eye, depends.

§ XII. Results found by a second Observer.

We may now consider the results of three series of observations made by
myself (J.) as observer, in order to determine the relation of one observer to

Scale.

Wave-length.

in watei

R Scarlet

. . 24

2328

1-332

G Green .
B Blue .

. . 46f
. . 64J

1914
1717

1-334
1-339

432 ON THE THEORY OF COMPOUND COLOURS.

another in the perception of colour. The standard colours are connected by the
following equation, as determined by six observations : —

18-l(24) + 27'5(44) + 37(68) = W* (17).

The average errors in these observations were —

TABLE VIII.

R, -28
G, -83
B, -16

G + B, -83
B + R, -42
R + G, -95

G - B, -83
B-R, -28
R-G, -72

R + G + B, -95

shewing that in this case, also, the power of distinguishing colour is more to be
depended on than that of distinguishing degrees of illumination.

The average error in the other observations from the means was "64 for red,
76 for green, and 1'02 for blue.

TABLE IX.
Observations by J., October 1859.

(24.)

(44.)

(68.)

44-3(20)= 18-1

- 2-5

+ 2-3

16-0(28)= 18-1

+ 6-2

- 0-7

21-5(32)= 18-1

+ 25-2

- 0-7

19-3(36)= 8-1

+ 27-5

- 0-3

20-7(40)= 2-1

+ 27-5

- 0-5

52-3(48) = - 1-4

+ 27-5

+ 10-7

95-0(52) = - 2-4

+ 27-5

+ 37-0

51-7(56) = - 2-2

+ 4-8

+ 37-0

37-2(60) = - 1-2

+ 0-8

+ 37-0

36-7(64) = - 0-2

+ 0-8

+ 37-0

35-0(72) = + 0-6

- 0-2

+ 37-0

40-0(76) = + 0-9

+ 0-5

+ 37-0

51-0(80) = + 1-1

+ 0-5

+ 37-0

§ XIII. Comparison of ResuUs by Newton's Diagram.

The relations of the colours, as given by these observations, are laid down
in fig. 5, Plate VI., p. 444. It appears from this diagram, that the positions of
the colours lie nearly in a straight line from (24) to (44), and from (48) to (60).
The colours beyond (60) are crowded together, as in the other diagram, and
the observations are not yet sufficiently accurate to distinguish their relative
positions accurately. The colour (20) at the red end of the spectrum is further

ON THE THEORY OF COMPOUND COLOURS. 433

from the line joining (24) and (68) than in the other diagram, but I have not
obtained satisfactory observations of these extreme colours. It will be observed
that (32), (36), and (40) are placed further to the right in fig. 5 than in fig. 4,
shewing that the second observer (J.) sees more green in these colours than
the first (K.), also that (48), (52), (56), and (60) are much further up in fig. 5,
shewing that to the second observer they appear more blue and less green.
These differences were well seen in making an observation. When the instru-
ment was adjusted to suit the first observer (K.), then, if the selected colour
were (32), (36), or (40), the second (J.), on looking into the instrument, saw it
too green ; but if (48), (52), (56), or (60) were the selected colour, then, if right
to the first observer, it appeared too blue to the second. If the instrument
were adjusted to suit the second observer, then, in the first case, the other saw
red, and in the second green ; shewing that there was a real difference in the
eyes of these two individuals, producing constant and measurable differences in
the apparent colour of objects.

§ XIV. Comparison by Curves of Intensity of the Primaries.

Figs. 6 and 7, Plate VI. p. 444, are intended to indicate the intensities of
the three standard colours at different points of the spectrum. The curve marked
(R) indicates the intensity of the red or (24), (G) that of green or (44), and (B)
that of blue or (68). The curve marked (S) has its ordinates equal to the
sum of the ordinates of the other three curves. The intensities are found by
dividing every colour-equation by the coefficient of the colour on the left-hand
side. Fig. 6 represents the results of observations by K., and fig. 7 represents
those of J. It will be observed that the ordinates in fig. 7 are smaller between
(48) and (56) than in fig. 6. This indicates the feeble intensity of certain kinds
of light as seen by the eyes of J., which made it impossible to get observations
of the colour (52) at all without making the slit so wide as to include all
between (48) and (56).

This blindness of my eyes to the parts of the spectrum between the fixed
lines E and F appears to be confined to the region surrounding the axis of
vision, as the field of view, when adjusted for my eyes looking directly at the
colour, is decidedly out of adjustment when I view it by indirect vision, turning
the axis of my eye towards some other point. The prism then appears greener

VOL. I. 55

434 ON THE THEORY OF COMPOUND COLOURS.

and brighter than the mirror, shewing that the parts of my eye at a distance
from the axis are more sensitive to this blue-green light than the parts close
to the axis.

It is to be noticed that this insensibility is not to all light of a green
or blue colour, but to light of a definite refrangibility. If I had a species of
colour-blindness rendering me totally or partially insensible to that element of
colour which most nearly corresponds with the light in question, then the light
from the mirror, as well as that from the prism, would appear to me deficient
in that colour, and I should still consider them chromatically identical ; or if
there were any difference, it would be the same for all colours nearly the same
in appearance, such as those just beyond the line F, which appear to me quite
bright.

We must also observe that the peculiarity is confined to a certain portion
of the retina, which is known to be of a yellow colour, and which is the seat
of several ocular phenomena observed by Purkinje and Wheatstone, and of the
sheaf or brushes seen by Haidinger in polarized light ; and also that though,
of the two observers whose results are given here, one is much more affected
with this peculiarity than the other, both are less sensible to the light between
E and F than to that on either side ; and other observers, whose results are
not here given, confirm this.

§ XV. Explanation of the Differences between the two Observer*.

I think, therefore, that the yellow spot at the foramen centrale of Soemmering
will be found to be the cause of this phenomenon, and that it absorbs the rays
between E and F, and would, if placed in the path of the incident light,
produce a corresponding dark band in the spectrum formed by a prism.

The reason why white light does not appear yellow in consequence, is that
this absorbing action is constant, and we reckon as white the mean of all the
colours we are accustomed to see. This may be proved by wearing spectacles
of any strong colour for some time, when we shall find that we judge white
objects to be white, in spite of the rays which enter the eye being coloured.

Now ordinary white light is a mixture of all kinds of light, including that
between E and F, which is partially absorbed. If, therefore, we compound an
artificial white containing the absorbed ray as one of its three components, it

ON THE THEORY OF COMPOUND COLOURS. 435

will be much more altered by the absorption than the ordinary light, which
contains many rays of nearly the same colour, which are not absorbed. On the
other hand, if the artificial light do not contain the absorbed ray, it will be
less altered than the ordinary light which contains it. Hence the greater the
absorption the less green will those colours appear which are near the absorbed
part, such as (48), (52), (56), and the more green will the colours appear which
are not near it, such as (32), (36), (40). And these are the chief differences
between fig. 4 and fig. 5.

I first observed this peculiarity of my eyes when observing the spectrum
formed by a very long vertical slit. I saw an elongated dark spot running up
and down in the blue, as if confined in a groove, and following the motion
of the eye as it moved up or down the spectrum, but, refusing to pass out
of the blue into other colours. By increasing the breadth of the spectrum, the
dark portion was found to correspond to the foramen centrale, and to be visible
only when the eye is turned towards the blue-green between E and F. The
spot may be well seen by first looking at a yellow paper, and then at a blue
one, when the spot will be distinctly seen for a short time, but it soon dis-
appears when the eye gets accustomed to the blue*.

I have been the more careful in stating this peculiarity of my eyes, as I
have reason to believe that it affects most persons, especially those who can see
Haidinger's brushes easily. Such persons, in comparing their vision with that
of others, may be led to think themselves affected with partial colour-blindness,
whereas their colour-vision may be of the ordinary kind, but the rays which
reach their sense of sight may be more Or less altered in their proportions by
passing through the media of the eye. The existence of real, though partial
colour-blindness will make itself apparent, in a series of observations, by the
discrepancy between the observed values and the means being greater in certain
colours than in others.

§ XVI. General Conclusions.

Neither of the observers whose results are given here shew any indications
of colour-blindness, and when the differences arising from the absorption of the
rays between E and F are put out of account, they agree in proving that there
are three colours in the spectrum, red, green, and blue, by the mixtures of

* See the Report of tlie British Association for 1856, p. 12.

55—2

436 ON THE THEORY OF COMPOUND COLOURS.

which colours chromatically identical with the other colours of the spectrum
may be produced. The exact position of the red and blue is not yet ascer-
tained ; that of the green is £ from E towards F.

The orange and yellow of the spectrum are chromatically equivalent to
mixtures of red and green. They are neither richer nor paler than the corre-
sponding mixtures, and the only difference is that the mixture may be resolved
by a prism, whereas the colour in the spectrum cannot be so resolved. This
result seems to put an end to the pretension of yellow to be considered a
primary element of colour.

In the same way the colours from the primary green to blue are chro-
matically identical with mixtures of these ; and the extreme ends of the spectrum
are probably equivalent to mixtures of red and blue, but they are so feeble
in illumination that experiments on the same plan with the rest can give no
result, but they must be examined by some special method. When observations
have been obtained from a greater number of individuals, including those whose
vision is dichromatic, the chart of the spectrum may be laid down independently
of accidental differences, and a more complete discussion of the laws of the
sensation of colour attempted.

POSTSCRIPT.
[Received May 8,— Read May 24, I860.]

Since sending the above paper to the Royal Society, I have obtained
some observations of the colour of the spectrum by persons whose vision is
" dichromic," and who are therefore said to be " colour-blind."

The instrument used in making these observations was similar in principle
to that formerly described, except that, in order to render it portable, the rays
are reflected back through the prisms, nearly in their original direction ; thus
rendering one of the limbs of the instrument unnecessary, and allowing the
other to be shortened considerably on account of the greater angular dispersion.
The principle of reflecting light, so as to pass twice through the same prism,
was employed by me in an instrument for combining colours made in 1856,
and a reflecting instrument for observing the spectrum has been constructed
independently by M. Porro.

ON THE THEORY OF COMPOUND COLOURS. 437

Light from a sheet of paper illuminated by sunlight is admitted at the slits
X, Y, Z (fig. 8, Plate VII. p. 444), falls on the prisms P and Pf (angles = 45°),
then on a concave silvered glass, S, radius 34 inches. The light, after reflexion,
passes again through the prisms Pt and P, and is reflected by a small mirror,
e, to the slit E, where the eye is placed to receive the light compounded of
the colours corresponding to the positions and breadths of the slits X, Y, and Z.

At the same time, another portion of the light from the illuminated paper
enters the instrument at BC, is reflected at the mirror M, passes through the
lens L, is reflected at the mirror M', passes close to the edge of the prism P,
and is reflected along with the coloured light at e, to the eye-slit at E.

In this way the compound colour is compared with a constant white light
in optical juxtaposition with it. The mirror M is made xof silvered glass, that
at M' is made of glass roughened and blackened at the back, to reduce the
intensity of the constant light to a convenient value for the experiments.

This instrument gives a spectrum in which the lines are very distinct,
and the length of the spectrum from A to H is 3 '6 baches. The outside
measure of the box is 3 feet 6 inches, by 11 inches by 4 inches, and it can
be carried about, and set up in any position, without readjustment. It was
made by Messrs Smith and Ramage of Aberdeen.

In obtaining observations from colour-blind persons, two slits only are
required to produce a mixture chromatically equivalent to white ; and at one
point of the spectrum the colour of the pure rays appears identical with white.
This point is near the line F, a little on the less refrangible side. From this
point to the more refrangible end of the spectrum appears to them " blue."
The colours on the less refrangible side appear to them all of the same quality,
but of different degrees of brightness ; and when any of them are made
sufficiently bright, they are called "yellow." It is convenient to use the term
"yellow" in speaking of the colours from red to green inclusive, since it will
be found that a dichromic person in speaking of red, green, orange, and brown,
refers to different degrees of brightness or purity of a single colour, and not
to different colours perceived by him. This colour we may agree to call
"yellow," though it is not probable that the sensation of it is like that of
yellow as perceived by us.

Of the three standard colours which I formerly assumed, the red appears
to them "yellow," but so feeble that there is not enough in the whole red
division of the spectrum to form an equivalent to make up the standard white.

438 ON THE THEORY OF COMPOUND COLOURS.

The green at E appears a good "yellow," and the blue at f from F towards
G appears a good "blue." I have therefore taken these as standard colours for
reducing dichromic observations. The three standard colours will be referred to
as (104), (88), and (68), these being the positions of the red, green, and blue on
the scale of the new instrument.

Mr James Simpson, formerly student of Natural Philosophy in my class, has
furnished me with thirty-three observations taken in good sunlight. Ten of
these were between the two standard colours, and give the following result : —

337 (88) + 33-1 (68) = W (1).

The mean errors of these observations were as follows :—
Error of (88) = 2'5 ; of (68) = 2'3 ; of (88) + (68) = 4'8 ; of (88) - (68) = 1 "3.

The fact that the mean error of the sum was so much greater than the mean
error of the difference indicates that in this case, as in all others that I have
examined, observations of equality of tint can be depended on much more than
observations of equality of illumination or brightness.

From six observations of my own, made at the same time, I have deduced
the " trichromic " equation

22-6 (104) + 26 (88) + 37-4 (68) = W (2).

If we suppose that the light which reached the organ of vision was the
same in both cases, we may combine these equations by subtraction, and so find

22-6 (104) -77 (88) + 4'3 (68) = D (3),

where D is that colour, the absence of the sensation of which constitutes the
defect of the dichromic eye. The sensation which I have in addition to
those of the dichromic eye is therefore similar to the full red (104), but
different from it, in that the red (104) has 77 of green (88) in it which must
be removed, and 4'3 of blue (68) substituted. This agrees pretty well with the
colour which Mr Pole* describes as neutral to him, though crimson to others.
It must be remembered, however, that different persons of ordinary vision require
different proportions of the standard colours, probably owing to differences in the
absorptive powers of the media of the eye, and that the above equation (2), if
observed by K., would have been

23 (104) + 32 (88) + 31 (68) = W (4).

* Philosophical Transactions, 1859, Part I. p. 329.

ON THE THEORY OF COMPOUND COLOURS. 439

and the value of D, as deduced from these observers, would have been

23 (104) -17 (88) -I'l (68) =D (5),

in which the defective sensation is much nearer to the red of the spectrum. It
is probably a colour to which the extreme red of the spectrum tends, and
which differs from the extreme red only in not containing that small proportion
of " yellow " light which renders it visible to the colour-blind.

From other observations by Mr Simpson the following results have been
deduced : —

TABLE a.

(88.) (68.)

(99-2 + )= 33-7 1-9

31-3(96)= 33-7 2-1

28 (92)= 33-7 1-4

33-7(88)= 33-7 0

54-7(84)= 33-7 6-1

71 (82)= 33-7 15-1

99 (80)= 33-7 33-1

70 (78)= 15-7 33-1

56 (76)= 5-7 33-1

36 (72)= - 0-3 33-1

33-1(68)= 0 33-1

40 (64)= 0-2 33-1

55-5(60)= 1-7 33-1

(57 - ) = - 0-3 33-1

'' (88.) (68.)

100(96)= 108 7

100(92)= 120 5

100(88)= 100 0

100(84)= 61 11

100(82)= 47 21

100(80)= 34 33

100(78)= 22 47

100(76)= 10 59

100(72)= 1 92

100(68)= 0 100

100(64)= 0 83

100(60)= 3 60

In the Table on the left side (99'2 + ) means the whole of the spectrum beyond
(99'2) on the scale, and (57 — ) means the whole beyond (57) on the scale. The
position of the fixed lines with reference to the scale was as follows : —

A, 116; a, 112; B, 110; C, 106; D, 98'3 ; E, 88; F, 79; G, 61; H, 44.

The values of the standard colours in different parts of the spectrum are given
on the right side of the above Table, and are represented by the curves of
fig. 9, Plate VII. p. 444, where the left-hand curve represents the intensity
of the " yellow " element, and the right-hand curve that of the " blue " element
of colour as it appears to the colour-blind.

The appearance of the spectrum to the colour-blind is as follows : —
From A to E the colour is pure "yellow" very faint up to D, and
reaching a maximum between D and E. From E to one-third beyond F towards

440 ON THE THEORY OF COMPOUND COLOURS.

G the colour is mixed, varying from "yellow" to "blue," and becoming neutral
or "white" at a point near F. In this part of the spectrum, the total inten-
sity, as given by the dotted line, is decidedly less than on either side of it, and
near the line F, the retina close to the "yellow spot" is less sensible to light
than the parts further from the axis of the eye. This peculiarity of the light
near F is even more marked in the colour-blind than in the ordinary eye.
Beyond F the " blue " element comes to a maximum between F and G, and
then diminishes towards H ; the spectrum from this maximum to the end being
pure " blue."

In fig. 10, Plate VII. p. 444, these results are represented in a different
manner. The point D, corresponding to the sensation wanting in the colour-blind,
is taken as the origin of coordinates, the "yellow" element of colour is represented
by distances measured horizontally to the right from D, and the " blue " element
by distances measured vertically from the horizontal line through D. The
numerals indicate the different colours of the spectrum according to the scale
shewn in fig. 9, and the coordinates of each point indicate the composition of
the corresponding colour. The triangle of colours is reduced, in the case of
dichromic vision, to a straight line "B" "Y," and the proportions of "blue"
and "yellow" in each colour are indicated by the ratios in which this line is
cut by the line from D passing through the position of that colour.

The results given above were all obtained with the light of white paper,
placed in clear sunshine. I have obtained similar results, when the sun was
hidden, by using the light of uniformly illuminated clouds, but I do not consider
these observations sufficiently free from disturbing circumstances to be employed
in calculation. It is easy, however, by means of such observations, to verify the
most remarkable phenomena of colour-blindness, as for instance, that the colours
from red to green appear to differ only in brightness, and that the brightness
may be made identical by changing the width of the slit ; that the colour
near F is a neutral tint, and that the eye in viewing it sees a dark spot in
the direction of the axis of vision ; that the colours beyond are all blue of
different intensities, and that any "blue" may be combined with any "yellow"
hi such proportions as to form " white." These results I have verified by the
observations of another colour-blind gentleman, who did not obtain sunlight for
his observations ; and as I have now the means of carrying the requisite
apparatus easily, I hope to meet with other colour-blind observers, and to obtain
their observations under more favourable circumstances.

ON THE THEORY OF COMPOUND COLOURS. 441

On the Comparison of Colour-blind with ordinary Vision by means of Observations

with Coloured Papers.

In March 1859 I obtained a set of observations by Mr Simpson, of the
relations between six coloured papers as seen by him. The experiments were
made with the colour-top in the manner described in my paper in the Trans-
actions of the Royal Society of Edinburgh, Vol. xxi. pt. 2, p. 286 ; and the
colour-equations were arranged so as to be equated to zero, as in those given
in the Philosophical Magazine, July, 1857. The colours were — Vermilion (V),
ultramarine (U), emerald-green (G), ivory-black (B), snow-white (W), and pale
chrome-yellow (Y). These six colours afford fifteen colour-blind equations, since
four colours enter into each equation. Fourteen of these were observed by
Mr Simpson, and from these I deduced three equations, giving the relation of
the three standards (V), (U), (G) to the other colours, according to his kind of
vision. From these three equations I then deduced fifteen equations, admitting
of comparison with the observed equations, and necessarily consistent in
themselves.

The comparison of these equations furnishes a test of the truth of the theory
that the colour-blind see by means of two colour-sensations, and that therefore
every colour may be expressed in terms of two given colours, just as in ordinary
vision it may be expressed in terms of three given colours. The one set of
equations are each the result of a single observation ; the other set are deduced
from three equations in accordance with this theory, and the two sets agree to
within an average error = 2' 1.

TABLE b.

2.
3.
4.
5.
VOL. I. 56

V.

U.

G.

B.

W.

T.

Observed . . .

0

0

-100

+ 45

+ 22

+ 33

= 0.

Calculated . .

0

0

-iod

+ 37-5

+ 26-5

+ 36

= 0.

Observed . . .

0

+ 58

0

-69

-31

-42

= 0.

Calculated . .

0

+ 58-3

0

-67-3

-32-7

+ 41-7

= 0.

Observed . . .

0

+ 32

-100

0

+ 12

+ 56

= 0.

Calculated . .

0

+ 32-3

-100

0

+ 8-3

+ 59-4

= 0.'

Observed . . .

0

+ 38

- 89

-11

0

+ 62

= 0.

Calculated . .

0

+ 40

- 85

-15

0

+ 60

= 0.

Observed . . .

0

+ 32

+ 68

-60

-40

0

= 0.

Calculated . .

0

+ 34

+ 66

-63-5

-36-5

0

= 0.

V.

n.

G.

B.

W.

Y.

Observed .

..-100

0

0

+ 82

+ 5

+ 13

= 0.

Calculated

. .-100

0

0

+ 83-9

+ 4-5

+ R-6

= 0.

Observed .

..+ 47

0

-100

0

+ 22

+ 31

= 0.

Calculated

..+ 44-7

0

-100

0

+ 24-5

+ 30-8

= 0.

Observed .

..-100

0

+ 20

+ 77

0

+ 3

= 0.

Calculated

..-100

0

+ 17

+ 77-5

0

+ 5-5

= 0.

442 ON THE THEORY OP COMPOUND COLOURS.

TABLE b (continued).
v.
6.

7.

8.

9. Not Observed
10.
11.
12.
13.
14.
15.

But, according to our theory, colour-blind vision is not only dichromic, but
the two elements of colour are identical with two of the three elements of
colour as seen by the ordinary eye ; so that it differs from ordinary vision
only in not perceiving a particular colour, the relation of which to known colours
may be numerically defined. This colour may be expressed under the form

aV + &U + cG = D (16),

where V, U, and G are the standard colours used in the experiments, and D is
the colour which is visible to the ordinary eye, but invisible to the colour-
blind. If we know the value of D, we may always change an ordinary colour-
equation into a colour-blind equation by subtracting from it «D (n being chosen
so that one of the standard colours is eliminated), and adding n of black.

In September 1856 I deduced, from thirty-six observations of my own, the
chromatic relations of the same set of six coloured papers. These observations,
with a comparison of them with the trichromic theory of vision, are to be
found in the Philosophical Magazine for July 1857. The relations of the

Calculated

..+ 96

0

- 31

-69

+ 4

0

= 0.

Observed .

..-70

+ 53

0

0

-30

+ 47

= 0.

Calculated

..- 73-5

+ 53

0

0

-26-5

+ 47

= 0.

Observed .

..-100

+ 8

0

+ 71

0

+ 21

= 0.

Calculated

. .-100

+ 8

0

+ 74-5

0

+ 17-5

= 0.

Observed .

. .+ 85

+ 15

0

-88

-12

0

= 0.

Calculated

..+ 86

+ 14

0

-88-5

-11-5

0

= 0.

Observed .

..-20

+ 39

- 80

0

0

+ 61

= 0.

Calculated

..-19

+ 40

- 81

0

0

+ 60

= 0.

Observed .

..-66

+ 30

+ 70

0

-34

0

= 0.

Calculated

..-70

+ 27

+ 73

0

-30

0

= 0.

Observed .

. . + 100

- 2

- 27

-71

0

0

= 0.

Calculated

. .+ 96

+ 4

- 24

-76

0

0

= 0.

ON THE THEORY OF COMPOUND COLOURS. 443

six colours may be deduced from two equations, of which the most convenient

form is

V. U. G. B. W. Y.

•+397 +26-6 +337 -227 -77'3 0 =0 (17).

-62-4 +18-6 -37'6 0 +457 +357 = 0 (18).

The value of D, as deduced from a comparison of these equations with the
colour-blind equations, is

M98V + 0-078U-0-276G = D (19).

By making D the same thing as black (B), and eliminating W and Y
respectively from the two ordinary colour-equations by means of D, we obtain
three colour-blind equations, calculated from the ordinary equations and con-
sistent with them, supposing that the colour (D) is black to the colour-blind.

The following Table is a comparison of the colour-blind equations deduced
from Mr Simpson's observations alone, with those deduced from my observations
and the value of D.

TABLE c.

V.

U.

Q.

B.

W.

T.

(15) Calculated . .

+ 96

+ 4

-24

-76

0

0

By (19)

+ 93-9

+ 6-1

-21-7

-78-3

0

0

(14) Calculated . .

-70

+ 27

+ 73

0

-30

0

By (17) and (19) . .

-70

+ 27-2

-72-8

0

-30

0

(13) Calculated . .

-19

+ 40

-81

0

0

+ 60

By (18) and (19) . . -13'6 +38-5 -86-4 0 0 +61-5

The average error here is 1'9, smaller than the average error of the indi-
vidual colour-blind observations, shewing that the theory of colour-blindness being
the want of a certain colour-sensation which is one of the three ordinary colour-
sensations, agrees with observation to within the limits of error.

In fig. 11, Plate VII. p. 444, I have laid down the chromatic relations of these
colours according to Newton's method. V (vermilion), U (ultramarine), and G
(emerald-green) are assumed as standard colours, and placed at the angles of
an equilateral triangle. The position of W (white) and Y (pale chrome-yellow)
with respect to these are laid down from equations (17) and (18), deduced
from my own observations. The positions of the defective colour, of white, and
of yellow, as deduced from Mr Simpson's equations alone, are given at " d,"
"w," and "y." The positions of these points, as deduced from a combination

56—2

444 ON THE THEORY OP COMPOUND COLOURS.

of these equations with my own, are given at "D," "W," and "Y." The
difference of these positions from those of "d," "w," and "y," shews the amount
of discrepancy between observation and theory.

It will be observed that D is situated near V (vermilion), but that a line
from D to W cuts UV at C near to V. D is therefore a red colour, not
scarlet, but further from yellow. It may be called crimson, and may be imitated
by a mixture of 86 vermilion and 14 ultramarine. This compound colour will be
of the same hue as D ; but since C lies between D and W, C must be
regarded as D diluted with a certain amount of white ; and therefore D must
be imagined to be like C in hue, but without the intermixture of white which
is unavoidable in actual pigments, and which reduces the purity of the tint.

Lines drawn from D through "W" and "Y," the colour-blind positions of
white and yellow, pass through W and Y, their positions in ordinary vision.
The reason why they do not coincide with W and Y, is that the white and
yellow papers are much brighter than the colours corresponding to the points
W and Y of the triangle V, U, G; and therefore lines from D, which represent
them in intensity as well as in quality, must be longer than DW and DY in
the proportion of their brightness.

VOL. I. I' LATE VI. (i)

On Compound/ fbloars.
Pig. 1.

20 M 2S 32~36 40 4f

32 36 W'U'W&'WGO & 68 72 76 SO

pfir

Cambridge University Pro*

VOL. I. PLATE VII. (ii)

fig. 8.

9t> 02 M 84 SO 76 7Z 68 61 60 56 St 48

To face paye 444

Cambridge University Prem

[Lecture at the Royal Institution of Great Britain. May 17, 1861.]

XXII. On the Theory of Three Primary Colours.

THE speaker commenced by shewing that our power of vision depends
entirely on our being able to distinguish the intensity and quality of colours.
The forms of visible objects are indicated to us only by differences in colour
or brightness between them and surrounding objects. To classify and arrange
these colours, to ascertain the physical conditions on which the differences of
coloured rays depend, and to trace, as far as we are able, the physiological
process by which these different rays excite in us various sensations of colour,
we must avail ourselves of the united experience of painters, opticians, and
physiologists. The speaker then proceeded to state the results obtained by these
three classes of inquirers, to explain their apparent inconsistency by means of
Young's Theory of Primary Colours, and to describe the tests to which he had
subjected that theory.

Painters have studied the relations of colours, in order to imitate them by
means of pigments. As there are only a limited number of coloured substances
adapted for painting, while the number of tints in nature is infinite, painters
are obliged to produce the tints they require by mixing their pigments in
proper proportions. This leads them to regard these tints as actually com-
pounded of other colours, corresponding to the pure pigments in the mixture.
It is found, that by using three pigments only, we can produce all colours
lying within certain limits of intensity and purity. For instance, if we take
carmine (red), chrome yellow, and ultramarine (blue), we get by mixing the
carmine and the chrome, all varieties of orange, passing through scarlet to
crimson on the one side, and to yellow on the other ; by mixing chrome and
ultramarine we get all hues of green; and by mixing ultramarine with carmine,
we get all hues of purple, from violet to mauve and crimson. Now these are
all the strong colours that we ever see or can imagine : all others are like

446 OX THE TllEORY OF THREE PRIMARY COLOURS.

these, only less pure in tint. Our three colours can be mixed so as to form
a neutral grey ; and if this grey be mixed with any of the hues produced by
mixing two colours only, all the tints of that hue will be exhibited, from the
pure colour to neutral grey. If we could assume that the colour of a mixture
of different kinds of paint is a true mixture of the colours of the pigments,
and in the same proportion, then an analysis of colour might be made with
the same ease as a chemical analysis of a mixture of substances.

The colour of a mixture of pigments, however, is often very different from
a true mixture of the colours of the pure pigments. It is found to depend on
the size of the particles, a finely ground pigment producing more effect than
one coarsely ground. It has also been shewn by Professor Helmholtz, that when
light falls on a mixture of pigments, part of it is acted on by one pigment
only, and part of it by another ; while a third portion is acted on by both pig-
ments in succession before it is sent back to the eye. The two parts reflected
directly from the pure pigments enter the eye together, and form a true mixture
of colours ; but the third portion, which has suffered absorption from both
pigments, is often so considerable as to give its own character to the resulting
tint. This is the explanation of the green tint produced by mixing most blue
and yellow pigments.

In studying the mixture of colours, we must avoid these sources of error,
either by mixing the rays of light themselves, or by combining the impressions
of colours within the eye by the rotation of coloured papers on a disc.

The speaker then stated what the opticians had discovered about colour.
White light, according to Newton, consists of a great number of different kinds
of coloured light which can be separated by a prism. Newton divided these
into seven classes, but we now recognize many thousand distinct kinds of light
in the spectrum, none of which can be shewn to be a compound of more
elementary rays. If we accept the theory that light is an undulation, then,
as there are undulations of every different period from the one end of the
spectrum to the other, there are an infinite number of possible kinds of light,
no one of which can be regarded as compounded of any others.

Physical optics does not lead us to any theory of three primary colours,
but leaves us in possession of an infinite number of pure rays with an infinitely
more infinite number of compound beams of light, each containing any propor-
tions of any number of the pure rays.

These beams of light, passing through the transparent parts of the eye, fall

ON THE THEORY OF THREE PRIMARY COLOURS. 447

on a sensitive membrane, and we become aware of various colours. We know
that the colour we see depends on the nature of the light ; but the opticians
say there are an infinite number of kinds of light ; while the painters, and all
who pay attention to what they see, tell us that they can account for all
actual colours by supposing them mixtures of three primary colours.

The speaker then next drew attention to the physiological difficulties in
accounting for the perception of colour. Some have supposed that the different
kinds of light are distinguished by the time of their vibration. There are
about 447 billions of vibrations of red light in a second ; and 577 billions of
vibrations of green light in the same time. It is certainly not by any mental
process of which we are conscious that we distinguish between these infini-
tesimal portions of time, and it is difficult to conceive any mechanism by which
the vibrations could be counted so that we should become conscious of the
results, especially when many rays of different periods of vibration act on the
same part of the eye at once.

Besides, all the evidence we have on the nature of nervous action goes
to prove that whatever be the nature of the agent which excites a nerve, the
sensation will differ only in being more or less acute. By acting on a nerve
in various ways, we may produce the faintest sensation or the most violent
pain ; but if the intensity of the sensation is the same, its quality must be
the same.

Now, we may perceive by our eyes a faint red light which may be made
stronger and stronger till our eyes are dazzled. We may then perform the
same experiment with a green light or a blue light. We shall thus see that
our sensation of colour may differ in other ways, besides in being stronger or
fainter. The sensation of colour, therefore, cannot be due to one nerve only.

The speaker then proceeded to state the theory of Dr Thomas Young, as
the only theory which completely reconciles these difficulties in accounting for
the perception of colour.

Young supposes that the eye is provided with three distinct sets of nervous
fibres, each set extending over the whole sensitive surface of the eye. Each
of these three systems of nerves, when excited, gives us a different sensation.
One of them, which gives us the sensation we call red, is excited most by
the red rays, but also by the orange and yellow, and slightly by the violet ;
another is acted on by the green rays, but also by the orange and yellow and
part of the blue; while the third is acted on by the blue and violet rays.

448 ON THE THEORY OF THREE PRIMARY COLOURS.

If we could excite one of these sets of nerves without acting on the
others, we should have the pure sensation corresponding to that set of nerves.
This would be truly a primary colour, whether the nerve were excited by pure
or by compound light, or even by the action of pressure or disease.

If such experiments could be made, we should be able to see the primary
colours separately, and to describe their appearance by reference to the scale
of colours in the spectrum.

But we have no direct consciousness of the contrivances of our own bodies,
and we never feel any sensation which is not infinitely complex, so that we
can never know directly how many sensations are combined when we see a
colour. Still less can we isolate one or more sensations by artificial means, so
that in general when a ray enters the eye, though it should be one of the
pure rays of the spectrum, it may excite more than one of the three sets of
nerves, and thus produce a compound sensation.

The terms simple and compound, therefore, as applied to colour-sensation,
have by no means the same meaning as they have when applied to a ray of
light.

The speaker then stated some of the consequences of Young's theory, and
described the tests to which he had subjected it: —

1st. There are three primary colours.

2nd. Every colour is either a primary colour, or a mixture of primary
colours.

3rd. Four colours may always be arranged in one of two ways. Either
one of them is a mixture of the other three, or a mixture of two of them
can be found, identical with a mixture of the other two.

4th. These results may be stated in the form of colour-equations, giving
the numerical value of the amount of each colour entering into any mixture.
By means of the Colour Top*, such equations can be obtained for coloured
papers, and they may be obtained with a degree of accuracy shewing that the
colour-judgment of the eye may be rendered very perfect.

The speaker had tested in this way more than 100 different pigments and
mixtures, and had found the results agree with the theory of three primaries

* Described in the Trant. of the Royal Society of Edinburgh, Vol. XXL, and in the Phil. Mag.

ON THE THEORY OF THREE PRIMARY COLOURS. 449

in every case. He had also examined all the colours of the spectrum with
the same result.

The experiments with pigments do not indicate what colours are to be
considered as primary ; but experiments on the prismatic spectrum shew that
all the colours of the spectrum, and therefore all the colours in nature, are
equivalent to mixtures of three colours of the spectrum itself, namely, red,
green (near the line E], and blue (near the line (7). Yellow was found to be
a mixture of red and green.

The speaker, assuming red, green, and blue as primary colours, then exhi-
bited them on a screen by means of three magic lanterns, before which were
placed glass troughs containing respectively sulphocyanide of iron, chloride of
copper, and ammoniated copper.

A triangle was thus illuminated, so that the pure colours appeared at its
angles, while the rest of the triangle contained the various mixtures of the
colours as in Young's triangle of colour.

The graduated intensity of the primary colours in different parts of the
spectrum was exhibited by three coloured images, which, when superposed on
the screen, gave an artificial representation of the spectrum.

Three photographs of a coloured ribbon taken through the three coloured
solutions respectively, were introduced into the camera, giving images represent-
ing the red, the green, and the blue parts separately, as they would be seen
by each of Young's three sets of nerves separately. When these were super-
posed, a coloured image was seen, which, if the red and green images had
been as fully photographed as the blue, would have been a truly-coloured image
of the ribbon. By finding photographic materials more sensitive to the less
refrangible rays, the representation of the colours of objects might be greatly
improved.

The speaker then proceeded to exhibit mixtures of the colours of the pure
spectrum. Light from the electric lamp was passed through a narrow slit, a
lens and a prism, so as to throw a pure spectrum on a screen containing three
moveable slits, through which three distinct portions of the spectrum were
suffered to pass. These portions were concentrated by a lens on a screen at
a distance, forming a large, uniformly coloured image of the prism.

When the whole spectrum was allowed to pass, this image was white, as
in Newton's experiment of combining the rays of the spectrum. When portions
of the spectrum were allowed to pass through the moveable slits, the image was

VOL. I. 57

450 ON THE THEORY OF THREE PRIMARY COLOURS.

uniformly illuminated with a mixture of the corresponding colours. In order
to see these colours separately, another lens was placed between the moveable
slits and the screen. A magnified image of the slits was thus thrown on the
screen, each slit shewing, by its colour and its breadth, the quality and quantity
of the colour which it suffered to pass. Several colours were thus exhibited,
first separately, and then in combination. Red and blue, for instance, produced
purple ; red and green produced yellow ; blue and yellow produced a pale pink ;
red, blue, and green produced white ; and red and a bluish green near the
line F produced a colour which appears very different to different eyes.

The speaker concluded by stating the peculiarities of colour-blind vision,
and by shewing that the investigation into the theory of colour is truly a
physiological inquiry, and that it requires the observations and testimony of
persons of every kind in order to discover and explain the various peculiarities
of vision.

[From the Philosophical Magazine, Vol. xxi.]

XXIII. On Physical Lines of Force.

PAET I.
THE THEORY OP MOLECULAR VORTICES APPLIED TO MAGNETIC PHENOMENA.

IN all phenomena involving attractions or repulsions, or any forces depend-
ing on the relative position of bodies, we have to determine the magnitude and
direction of the force which would act on a given body, if placed in a given
position.

In the case of a body acted on by the gravitation of a sphere, this force
is inversely as the square of the distance, and in a straight line to the centre
of the sphere. In the case of two attracting spheres, or of a body not spherical,
the magnitude and direction of the force vary according to more complicated
laws. In electric and magnetic phenomena, the magnitude and direction of the
resultant force at any point is the main subject of investigation. Suppose that
the direction of the force at any point is known, then, if we draw a line so
that in every part of its course it coincides in direction with the force at that
point, this line may be called a line of force, since it indicates the direction
of the force in every part of its course.

By drawing a sufficient number of lines of force, we may indicate the
direction of the force in every part of the space in which it acts.

Thus if we strew iron filings on paper near a magnet, each filing will be
magnetized by induction, and the consecutive filings will unite by their opposite
poles, so as to form fibres, and these fibres will indicate the direction of the lines
of force. The beautiful illustration of the presence of magnetic force afforded
by this experiment, naturally tends to make us think of the lines of force as
something real, and as indicating something more than the mere resultant of
two forces, whose seat of action is at a distance, and which do not exist there

57—2

452 ON PHYSICAL LINES OF FORCE.

at all until a magnet is placed in that part of the field. We are dissatisfied
with the explanation founded on the hypothesis of attractive and repellent
forces directed towards the magnetic poles, even though we may have satisfied
ourselves that the phenomenon is in strict accordance with that hypothesis, and
we cannot help thinking that in every place where we find these lines of force,
some physical state or action must exist in sufficient energy to produce the
actual phenomena.

My object in this paper is to clear the way for speculation in this direction,
by investigating the mechanical results of certain states of tension and motion
in a medium, and comparing these with the observed phenomena of magnetism
and electricity. By pointing out the mechanical consequences of such hypotheses,
I hope to be of some use to those who consider the phenomena as due to the
action of a medium, but are in doubt as to the relation of this hypothesis to
the experimental laws already established, which have generally been expressed
in the language of other hypotheses.

I have in a former paper* endeavoured to lay before the mind of the
geometer a clear conception of the relation of the lines of force to the space
in which they are traced. By making use of the conception of currents in a
fluid, I shewed how to draw lines of force, which should indicate by their
number the amount of force, so that each line may be called a unit-line of
force (see Faraday's Researches, 3122) ; and I have investigated the path of
the lines where they pass from one medium to another.

In the same paper I have found the geometrical significance of the " Elec-
trotonic State," and have shewn how to deduce the mathematical relations
between the electrotonic state, magnetism, electric currents, and the electromotive
force, using mechanical illustrations to assist the imagination, but not to account
for the phenomena.

I propose now to examine magnetic phenomena from a mechanical point of
view, and to determine what tensions in, or motions of, a medium are capable
of producing the mechanical phenomena observed. If, by the same hypothesis,
we can connect the phenomena of magnetic attraction with electromagnetic phe-
nomena and with those of induced currents, we shall have found a theory
which, if not true, can only be proved to be erroneous by experiments which
will greatly enlarge our knowledge of this part of physics.

* See a paper " On Faraday's Lanes of Force," Cambridge Philosophical Transactions, Vol. \. Part i.
Page 155 of this volume.

ON PHYSICAL LINES OF FORCE. 453

The mechanical conditions of a medium under magnetic influence have been
variously conceived of, as currents, undulations, or states of displacement or
strain, or of pressure or stress.

Currents, issuing from the north pole and entering the south pole of a
magnet, or circulating round an electric current, have the advantage of repre-
senting correctly the geometrical arrangement of the lines of force, if we could
account on mechanical principles for the phenomena of attraction, or for the
currents themselves, or explain their continued existence

Undulations issuing from a centre would, according to the calculations of
Professor Challis, produce an effect similar to attraction in the direction of the
centre ; but admitting this to be true, we know that two /series of undulations
traversing the same space do not combine into one resultant as two attractions
do, but produce an effect depending on relations of phase as well as intensity,
and if allowed to proceed, they diverge from each other without any mutual
action. In fact the mathematical laws of attractions are not analogous in any
respect to those of undulations, while they have remarkable analogies with those
of currents, of the conduction of heat and electricity, and of elastic bodies.

In the Cambridge and Dublin Mathematical Journal for January 1847,
Professor William Thomson has given a " Mechanical Representation of Electric,
Magnetic, and Galvanic Forces," by means of the displacements of the particles of
an elastic solid in a state of strain. In this representation we must make the
angular displacement at every point of the solid proportional to the magnetic
force at the corresponding point of the magnetic field, the direction of the axis
of rotation of the displacement corresponding to the direction of the magnetic
force. The absolute displacement of any particle will then correspond in magni-
tude and direction to that which I have identified with the electrotonic state ;
and the relative displacement of any particle, considered with reference to the
particle in its immediate neighbourhood, will correspond in magnitude and direc-
tion to the quantity of electric current passing through the corresponding point
of the magneto-electric field. The author of this method of representation does
not attempt to explain the origin of the observed forces by the effects due to
these strains in the elastic solid, but makes use of the mathematical analogies
of the two problems to assist the imagination in the study of both.

We come now to consider the magnetic influence as existing in the form of
some kind of pressure or tension, or, more generally, of stress in the medium.

Stress is action and reaction between the consecutive parts of a body, and

454 OX PHYSICAL LINES OF FORCE.

consists in general of pressures or tensions different in different directions at
th" same point of the medium.

The necessary relations among these forces have been investigated by mathe-
maticians ; and it has been shewn that the most general type of a stress
consists of a combination of three principal pressures or tensions, in directions
at right angles to each other.

When two of the principal pressures are equal, the third becomes an axis
of symmetry, either of greatest or least pressure, the pressures at right angles
to this axis being all equal.

When the three principal pressures are equal, the pressure is equal in every
direction, and there results a stress having no determinate axis of direction, of
which we have an example in simple hydrostatic pressure.

The general type of a stress is not suitable as a representation of a mag-
netic force, because a line of magnetic force has direction and intensity, but
has no third quality indicating any difference between the sides of the line,
which would be analogous to that observed in the case of polarized light*.

We must therefore represent the magnetic force at a point by a stress
having a single axis of greatest or least pressure, and all the pressures at right
angles to this axis equal. It may be objected that it is inconsistent to represent
a line of force, which is essentially dipolar, by an axis of stress, which is
necessarily isotropic ; but we know that every phenomenon of action and reaction
is isotropic in its results, because the effects of the force on the bodies between
which it acts are equal and opposite, while the nature and origin of the force
may be dipolar, as in the attraction between a north and a south pole.

Let us next consider the mechanical effect of a state of stress symmetrical
about an axis. We may resolve it, in all cases, into a simple hydrostatic
pressure, combined with a simple pressure or tension along the axis. When the
axis is that of greatest pressure, the force along the axis will be a pressure.
When the axis is that of least pressure, the force along the axis will be a
tension.

If we observe the lines of force between two magnets, as indicated by iron
tilings, we shall see that whenever the lines of force pass from one pole to
another, there is attraction between those poles; and where the lines of force
from the poles avoid each other and are dispersed into space, the poles repel

* See Faraday's Researches, 3252.

ON PHYSICAL LINES OF FORCE. 455

each, other, so that in both cases they are drawn in the direction of the
resultant of the lines of force.

It appears therefore that the stress in the axis of a line of magnetic force
is a tension, like that of a rope.

If we calculate the lines of force in the neighbourhood of two gravitating
bodies, we shall find them the same in direction as those near two magnetic
poles of the same name ; but we know that the mechanical effect is that of
attraction instead of repulsion. The lines of force in this case do not run
between the bodies, but avoid each other, and are dispersed over space. In
order to produce the effect of attraction, the stress along the lines of gravi-
tating force must be a pressure.

Let us now suppose that the phenomena of magnetism depend on the
existence of a tension in the direction of the lines of force, combined with a
hydrostatic pressure ; or in other words, a pressure greater in the equatorial
than in the axial direction : the next question is, what mechanical explanation
can we give of this inequality of pressures in a fluid or mobile medium ? The
explanation which most readily occurs to the mind is that the excess of pres-
sure in the equatorial direction arises from the centrifugal force of vortices or
eddies in the medium having their axes in directions parallel to the lines of force.

This explanation of the cause of the inequality of pressures at once suggests
the means of representing the dipolar character of the line of force. Every
vortex is essentially dipolar, the two extremities of its axis being distinguished
by the direction of its revolution as observed from those points.

We also know that when electricity circulates in a conductor, it produces
lines of magnetic force passing through the circuit, the direction of the lines
depending on the direction of the circulation. Let us suppose that the direction
of revolution of our vortices is that in which vitreous electricity must revolve
in order to produce lines of force whose direction within the circuit is the
same as that of the given lines of force.

We shall suppose at present that all the vortices in any one part of the
field are revolving in the same direction about axes nearly parallel, but
that in passing from one part of the field to another, the direction of the
axes, the velocity of rotation, and the density of the substance of the vortices
are subject to change. We shall investigate the resultant mechanical effect upon
an element of the medium, and from the mathematical expression of this
resultant we shall deduce the physical character of its different component parts.

456 ON PHYSICAL LINES OF FORCE.

PROP. I. — If in two fluid systems geometrically similar the velocities and
densities at corresponding points are proportional, then the differences of pres-
sure at corresponding points due to the motion will vary in the duplicate ratio
of the velocities and the simple ratio of the densities.

Let I be the ratio of the linear dimensions, m that of the velocities,
n that of the densities, and p that of the pressures due to the motion. Then
the ratio of the masses of corresponding portions will be Pn, and the ratio of
the velocities acquired in traversing similar parts of the systems will be m ;
so that Fmn is the ratio of the momenta acquired by similar portions in
traversing similar parts of their paths.

The ratio of the surfaces is Z1, that of the forces acting on them is l*p,

and that of the times during which they act is — ; so that the ratio of the

m

impulse of the forces is — , and we have now

m

Pmn-fe,

TO

or m'n = p ;

that is, the ratio of the pressures due to the motion (p) is compounded of
the ratio of the densities (n) and the duplicate ratio of the velocities (m1), and
does not depend on the linear dimensions of the moving systems.

In a circular vortex, revolving with uniform angular velocity, if the
pressure at the axis is pt, that at the circumference will be p1=p0 + ^pv', where
p is the density and v the velocity at the circumference. The mean pressure
parallel to the axis will be

If a number of such vortices were placed together side by side with their
axes parallel, they would form a medium in which there would be a pressure
]>t parallel to the axes, and a pressure p1 in any perpendicular direction. If the
vortices are circular, and have uniform angular velocity and density throughout,
then

If the vortices are not circular, and if the angular velocity and the density
are not uniform, but vary according to the same law for all the vortices,

ON PHYSICAL LINES OF FORCE. 457

where p is the mean density, and C is a numerical quantity depending on the
distribution of angular velocity and density in the vortex. In future we shall

write •*- instead of Co, so that

4?r

where ft is a quantity bearing a constant ratio to the density, and v is the
linear velocity at the circumference of each vortex.

A medium of this kind, filled with molecular vortices having their axes
parallel, differs from an ordinary fluid in having different pressures in different
directions. If not prevented by properly arranged pressures, it would tend to
expand laterally. In so doing, it would allow the diameter of each vortex to
expand and its velocity to diminish in the same proportion. In order that a
medium having these inequalities of pressure in different directions should be in
equilibrium, certain conditions must be fulfilled, which we must investigate.

PROP. II. — If the direction-cosines of the axes of the vortices with respect
to the axes of x, y, and z be I, m, and n, to find the normal and tangential
stresses on the co-ordinate planes.

The actual stress may be resolved into a simple hydrostatic pressure pl acting
in all directions, and a simple tension pl— p,, or — ptf, acting along the axis

of stress.

Hence if p^, pm> and pa be the normal stresses parallel to the three axes,
considered positive when they tend to increase those axes ; and if p^, plx, and
Pxy be the tangential stresses in the three co-ordinate planes, considered positive
when they tend to increase simultaneously the symbols subscribed, then by
the resolution of stresses*,

1

Pm-fr

i

^

i

* Rankine's Applied Mechanics, Art. 106.
VOL. I. 58

458

ON PHYSICAL LINES OF FORCE.
1

If we write
then

\ = vm) and y =
l^

J_
1

(2),

PROP. III. — To find the resultant force on an element of the medium,
arising from the variation of internal stress.

We have in general, for the force in the direction of x per unit of volume
by the law of equilibrium of stresses*,

In this case the expression may be written
1 (dJM da d

da

Remembering that aj~

da

= j~ (<*' + /? + •/), this becomes

1 (dfi da\ 1 (da dy\ dp,

The expressions for the forces parallel to the axes of y and z may be written
down from analogy.

* Rankine's Applied Mechanics, Art. 116.

ON PHYSICAL LINES OF FORCE. 459

We have now to interpret the meaning of each term of this expression.

We suppose a, /8, y to be the components of the force which would act
upon that end of a unit magnetic bar which points to the north.

fi represents the magnetic inductive capacity of the medium at any point
referred to air as a standard, /xa, fj.j3, p.y represent the quantity of magnetic
induction through unit of area perpendicular to the three axes of x, y z
respectively.

The total amount of magnetic induction through a closed surface surrounding
the pole of a magnet, depends entirely on the strength of that pole ; so that
if dxdydz be an element, then

(-T-/ta + -i-/iij6 + -T- py } dxdydz=kirm dxdydz ............ (6),

-i- -T-

which represents the total amount of magnetic induction outwards through the
surface of the element dxdydz, represents the amount of "imaginary magnetic
matter" within the element, of the kind which points north.

The first term, of the value of X, therefore,

d d 0 d

-J- PP + 'I'
dyr dz

may be written

am .......................................... (8),

where a is the intensity of the magnetic force, and m is the amount of mag-
netic matter pointing north in unit of volume.

The physical interpretation of this term is, that the force urging a north pole
in the positive direction of x is the product of the intensity of the magnetic
force resolved in that direction, and the strength of the north pole of the magnet.

Let the parallel lines from left to right in fig. 1 represent a field of mag-
netic force such as that of the earth, sn being the direction from south to north.
The vortices, according to our hypothesis, will be in the direction shewn by the
arrows in fig. 3, that is, in a plane perpendicular to the lines of force, and
revolving in the direction of the hands of a watch when observed from s
looking towards n. The parts of the vortices above the plane of the paper
will be moving towards e, and the parts below that plane towards w.

58—2

460

ON PHYSICAL LINES OF FORCE.

n

Fig. 2.
1 -,

. / •

^^ Y ^

5

B KB)< P *»

* A K

We shall always mark by an arrow-head the direction in which we must
look in order to see the vortices rotating in the
direction of the hands of a watch. The arrow-head
will then indicate the northward direction in the
magnetic field, that is, the direction in which that
end of a magnet which points to the north would
set itself in the field.

Now let A be the end of a magnet which
points north. Since it repels the north ends of
other magnets, the lines of force will be directed
from A outwards in all directions. On the north
side the line AD will be in the same direction with
the lines of the magnetic field, and the velocity of
the vortices will be increased. On the south side
the line AC will be in the opposite direction, and
the velocity of the vortices will be diminished, so
that the lines of force are more powerful on the
north side of A than on the south side.

We have seen that the mechanical effect of the
vortices is to produce a tension along their axes,
so that the resultant effect on A will be to pull

it more powerfully towards D than towards C '; that is, A will tend to move
to the north.

Let B in fig. 2 represent a south pole. The lines of force belonging to B
will tend towards B, and we shall find that the lines of force are rendered
stronger towards E than towards F, so that the effect in this case is to urge B
towards the south.

It appears therefore that, on the hypothesis of molecular vortices, our first
term gives a mechanical explanation of the force acting on a north or south
pole in the magnetic field.

We now proceed to examine the second term,

Here a' + ^ + y* is the square of the intensity at any part of the field, and
is the magnetic inductive capacity at the same place. Any body therefore

ON PHYSICAL LINES OF FOECE. 461

placed in the field will be urged towards places of stronger magnetic intensity
with a force depending partly on its own capacity for magnetic induction, and
partly on the rate at which the square of the intensity increases.

If the body be placed in a fluid medium, then the medium, as well as the
body, will be urged towards places of greater intensity, so that its hydrostatic
pressure will be increased in that direction. The resultant effect on a body
placed in the medium will be the difference of the actions on the body and
on the portion of the medium which it displaces, so that the body will tend
to or from places of greatest magnetic intensity, according as it has a greater
or less capacity for magnetic induction than the surrounding medium.

In fig. 4 the lines of force are represented as converging and becoming
more powerful towards the right, so that the magnetic tension at B is stronger
than at A, and the body AB will be urged to the right. If the capacity for
magnetic induction is greater hi the body than in the surrounding medium, it
will move to the right, but if less it will move to the left.

Fig. 4. Fig. 5.

We may suppose in this case that the lines of force are converging to a
magnetic pole, either north or south, on the right hand.

In fig. 5 the lines of force are represented as vertical, and becoming more
numerous towards the right. It may be shewn that if the force increases
towards the right, the lines of force will be curved towards the right. The
effect of the magnetic tensions will then be to draw any body towards the right
with a force depending on the excess of its inductive capacity over that of the
surrounding medium.

"We may suppose that in this figure the lines of force are those surrounding
an electric current perpendicular to the plane of the paper and on the right
hand of the figure.

These two illustrations will shew the mechanical effect on a paramagnetic
or diamagnetic body placed in a field of varying magnetic force, whether the
increase of force takes place along the lines or transverse to them. The form

462 ON PHYSICAL LINES OF FORCE.

of the second term of our equation indicates the general law, which is quite
independent of the direction of the lines of force, and depends solely on the
manner in which the force vanes from one part of the field to another.

We come now to the third term of the value of X,

1 dp da\
•

Here /i/J is, as before, the quantity of magnetic induction through unit of area
perpendicular to the axis of y, and -/- — -r- is a quantity which would disap-

pear if adx + fidy + ydz were a complete differential, that is, if the force acting
on a unit north pole were subject to the condition that no work can be done
upon the pole in passing round any closed curve. The quantity represents the
work done on a north pole in travelling round unit of area in the direction
from +x to +y parallel to the plane of xy. Now if an electric current whose
strength is r is traversing the axis of z, which, we may suppose, points
vertically upwards, then, if the axis of a; is east and that of y north, a unit
north pole will be urged round the axis of z in the direction from x to y, so

that in one revolution the work done will be = 47rr. Hence — I -/- — -7- ) repre-

4?r \dx ay]

sents the strength of an electric current parallel to z through unit of area ; and
if we write

1 [dy d$\ 1 (da dy\ 1 /d/3 da\

_ I — L _ _JLT I =/n — I — — — '— 1 — n _ I — v I Q i

4»\3y dzj P' ^\dz dx)~q> 4ir\dx dy)~

then p, q, r will be the quantity of electric current per unit of area perpen-
dicular to the axes of x, y, and z respectively.

The physical interpretation of the third term of X, — /u.y8r, is that if pfi is
the quantity of magnetic induction parallel to y, and r the quantity of electricity
flowing in the direction of z, the element will be urged in the direction of —x,
transversely to the direction of the current and of the lines of force; that is,
an ascending current in a field of force magnetized towards the north would
tend to move west.

To illustrate the action of the molecular vortices, let sn be the direction
of magnetic force in the field, and let C be the section of an ascending mag-
netic current perpendicular to the paper. The lines of force due to this current

ON PHYSICAL LINES OF FORCE. 463

will be circles drawn in the opposite direction from that of the hands of a
watch ; that is, in the direction mvse. At e the lines of force Fig. 6.

will be the sum of those of the field and of the current, and
at w they will be the difference of the two sets of lines ; so

that the vortices on the east side of the current will be more
powerful than those on the west side. Both sets of vortices have

5> J

their equatorial parts turned towards C, so that they tend to
expand towards (7, but those on the east side have the greatest
effect, so that the resultant effect on the current is to urge it towards the west.

The fourth term,

may be interpreted in the same way, and indicates that a current q in the
direction of y, that is, to the north, placed in a magnetic field in which the
lines are vertically upwards in the direction of z, will be urged towards the east.

The fifth term,

-
dx"

merely implies that the element will be urged in the direction in which the
hydrostatic pressure pl diminishes.

We may now write down the expressions for the components of the resultant
force on an element of the medium per unit of volume, thus :

<13>'

The first term of each expression refers to the force acting on magnetic
poles.

The second term to the action on bodies capable of magnetism by induction.
The third and fourth terms to the force acting on electric currents.
And the fifth to the effect of simple pressure.

464 ON PHYSICAL LINES OP FORCE.

Before going further in the general investigation, we shall consider equations
(12, 13, 14), in particular cases, corresponding to those simplified cases of the
actual phenomena which we seek to obtain in order to determine their laws by
experiment.

We have found that the quantities p, q, and r represent the resolved parts
of an electric current in the three co-ordinate directions. Let us suppose in the
first instance that there is no electric current, or that p, q, and r vanish. We
have then by (9),

whence we learn that ajdx + ftdy + ydz = d<l> .............................. (16),

is an exact differential of <f>, so that

_d$ d<f> _d<j>

~dx' P-dy' y~dz-

p is proportional to the density of the vortices, and represents the " capacity
for magnetic induction" in the medium. It is equal to 1 in air, or in whatever
medium the experiments were made which determined the powers of the magnets,
the strengths of the electric currents, &c.

Let us suppose /t constant, then
d d

represents the amount of imaginary magnetic matter in unit of volume. That
there may be no resultant force on that unit of volume arising from the action
represented by the first term of equations (12, 13, 14), we must have m = 0, or

Now it may be shewn that equation (19), if true within a given space,
implies that the forces acting within that space are such as would result from
a distribution of centres of force beyond that space, attracting or repelling
inversely as the square of the distance.

Hence the lines of force in a part of space where p is uniform, and where
there are no electric currents, must be such as would result from the theory
of "imaginary matter" acting at a distance. The assumptions of that theory
are unlike those of ours, but the results are identical

ON PHYSICAL LINES OF FORCE. 465

Let us first take the case of a single magnetic pole, that is, one end of
a long magnet, so long that its other end is too far off to have a perceptible
influence on the part of the field we are considering. The conditions then are,
that equation (18) must be fulfilled at the magnetic pole, and (19) everywhere
else. The only solution under these conditions is

<j>=-™1- .................................... (20),

p, r

where r is the distance from the pole, and m the strength of the pole.
The repulsion at any point on a unit pole of the same kind is

^
dr

—

771

In the standard medium yn = 1 ; so that the repulsion is simply — in that

medium, as has been shewn by Coulomb.

In a medium having a greater value of /n, (such as oxygen, solutions of
salts of iron, &c.) the attraction, on our theory, ought to be less than in air,
and in diamagnetic media (such as water, melted bismuth, &c.) the attraction
between the same magnetic poles ought to be greater than in air.

The experiments necessary to demonstrate the difference of attraction of two
magnets according to the magnetic or diamagnetic character of the medium in
which they are placed, would require great precision, on account of the limited
range of magnetic capacity in the fluid media known to us, and the small
amount of the difference sought for as compared with the whole attraction.

Let us next take the case of an electric current whose quantity is C,
flowing through a cylindrical conductor whose radius is R, and whose length is
infinite as compared with the size of the field of force considered.

Let the axis of the cylinder be that of z, and the direction of the current
positive, then within the conductor the quantity of current per unit of area is

_C_ 1 (dft_da\ . }.

?9"£rV& Ty)~
so that within the conductor

a=-2~y, P = 2~x, y = 0 ..................... (23).

VOL. I. 59

466 ON PHYSICAL LINES OF FORCE.

Beyond the conductor, in the space round it,

<£ = 2Ctan-' %• (24),

a = ^=-2C7 y . 0 = ^ = 2(7^* ,, 7 = ^ = 0 (25).

If p^Jaf + y* is the perpendicular distance of any point from the axis of

2(7
the conductor, a unit north pole will experience a force = — , tending to move

it round the conductor in the direction of the hands of a watch, if the observer
view it in the direction of the current.

Let us now consider a current running parallel to the axis of z in the
plane of xz at a distance p. Let the quantity of the current be c', and let

the length of the part considered be I, and its section s, so that is its

S

strength per unit of section. Putting this quantity for p in equations (12, 13,
14), we find

per unit of volume ; and multiplying by Is, the volume of the conductor con-

sidered, we find

X= -c'l

(26),

shewing that the second conductor will be attracted towards the first with a
force inversely as the distance.

We find in this case also that the amount of attraction depends on the
value of p, but that it varies directly instead of inversely as /i; so that the
attraction between two conducting wires will be greater in oxygen than in air,
and greater in air than in water.

We shall next consider the nature of electric currents and electromotive
forces in connexion with the theory of molecular vortices.

ON PHYSICAL LINES OF FORCE. 467

PAET II.
THE THEORY OF MOLECULAR VORTICES APPLIED TO ELECTRIC CURRENTS.

We have already shewn that all the forces acting between magnets, sub-
stances capable of magnetic induction, and electric currents, may be mechanically
accounted for on the supposition that the surrounding medium is put into such
a state that at every point the pressures are different in different directions,
the direction of least pressure being that of the observed lines of force, and
the difference of greatest and least pressures being proportional to the square
of the intensity of the force at that point.

Such a state of stress, if assumed to exist in the medium, and to be
arranged according to the known laws regulating lines of force, will act upon
the magnets, currents, &c. in the field with precisely the same resultant forces
as those calculated on the ordinary hypothesis of direct action at a distance.
This is true independently of any particular theory as to the cause of this
state of stress, or the mode in which it can be sustained in the medium. We
have therefore a satisfactory answer to the question, " Is there any mechanical
hypothesis as to the condition of the medium indicated by lines of force, by
which the observed resultant forces may be accounted for ? " The answer is,
the lines of force indicate the direction of minimum pressure at every point of
the medium.

The second question must be, "What is the mechanical cause of this
difference of pressure in different directions ? " We have supposed, in the first
part of this paper, that this difference of pressures is caused by molecular
vortices, having their axes parallel to the lines of force.

We also assumed, perfectly arbitrarily, that the direction of these vortices
is such that, on looking along a line of force from south to north, we should
see the vortices revolving in the direction of the hands of a watch.

We found that the velocity of the circumference of each vortex must be
proportional to the intensity of the magnetic force, and that the density of
the substance of the vortex must be proportional to the capacity of the medium
for magnetic induction.

We have as yet given no answers to the questions, " How are these vortices
set in rotation ? " and " Why are they arranged according to the known laws

59—2

468 ON PHYSICAL LINES OF FORCE.

of lines of force about magnets and currents?" These questions are certainly
of a higher order of difficulty than either of the former ; and I wish to separate
the suggestions I may offer by way of provisional answer to them, from the
mechanical deductions which resolved the first question, and the hypothesis of
vortices which gave a probable answer to the second.

We have, in fact, now come to inquire into the physical connexion of these
vortices with electric currents, while we are still in doubt as to the nature of
electricity, whether it is one substance, two substances, or not a substance at
all, or in what way it differs from matter, and how it is connected with it.

We know that the lines of force are affected by electric currents, and we
know the distribution of those lines about a current ; so that from the force
we can determine the amount of the current. Assuming that our explanation
of the lines of force by molecular vortices is correct, why does a particular
distribution of vortices indicate an electric current ? A satisfactory answer to
this question would lead us a long way towards that of a very important one,
"What is an electric current?"

I have found great difficulty in conceiving of the existence of vortices in a
medium, side by side, revolving in the same direction about parallel axes. The
contiguous portions of consecutive vortices must be moving in opposite directions ;
and it is difficult to understand how the motion of one part of the medium
can coexist with, and even produce, an opposite motion of a part in contact
with it.

The only conception which has at all aided me in conceiving of this kind of
motion is that of the vortices being separated by a layer of particles, revolving
each on its own axis in the opposite direction to that of the vortices, so that
the contiguous surfaces of the particles and of the vortices have the same
motion.

In mechanism, when two wheels are intended to revolve in the same direc-
tion, a wheel is placed between them so as to be in gear with both, and this
wheel is called an "idle wheel." The hypothesis about the vortices which I
have to suggest is that a layer of particles, acting as idle wheels, is interposed
between each vortex and the next, so that each vortex has a tendency to make
the neighbouring vortices revolve in the same direction with itself.

In mechanism, the idle wheel is generally made to rotate about a fixed
axle ; but in epicyclic trains and other contrivances, as, for instance, in Siemens's

ON PHYSICAL LINES OP FOKCE. 469

governor for steam-engines *, we find idle wheels whose centres are capable of
motion. In all these cases the motion of the centre is the half sum of the
motions of the circumferences of the wheels between which it is placed. Let
us examine the relations which must subsist between the motions of our vortices
and those of the layer of particles interposed as idle wheels between them.

PROP. IV. — To determine the motion of a layer of particles separating tw®
vortices.

Let the circumferential velocity of a vortex, multiplied by the three direc-
tion-cosines of its axis respectively, be a, ^8, y, as in Prop. II. Let I, m, n be
the direction-cosines of the normal to any part of the surface of this vortex,
the outside of the surface being regarded positive. Then the components of the
velocity of the particles of the vortex at this part of its surface will be

n/3 — my parallel to x,

ly — na parallel to y,

ma — lft parallel to z.

If this portion of the surface be in contact with another vortex whose velocities
are a, ft, y, then a layer of very small particles placed between them will
have a velocity which will be the mean of the superficial velocities of the
vortices which they separate, so that if u is the velocity of the particles in
the direction of x,

u = lm(y'-y)-ln(f?-p) ........................ (27),

since the normal to the second vortex is in the opposite direction to that of
the first.

PROP. V. — To determine the whole amount of particles transferred across
unit of area in the direction of x in unit of time.

2,

Let xt, ylt z1 be the co-ordinates of the centre of the first vortex, »2, y2
those of the second, and so on. Let Vlt F2, &c. be the volumes of the first,
second, &c. vortices, and V the sum of their volumes. Let dS be an element
of the surface separating the first and second vortices, and x, y, z its co-ordinates.
Let p be the quantity of particles on every unit of surface. Then if p be the
whole quantity of particles transferred across unit of area in unit of time in

* See Goodeve's Elements of Mechanism, p. 118.

470 ON PHYSICAL LINES OF FORCE.

the direction of x, the whole momentum parallel to x of the particles within
the space whose volume is V will be Vp, and we shall have

Vp^tupdS ................................. (28),

the summation being extended to every surface separating any two vortices
within the volume V.

Let us consider the surface separating the first and second vortices. Let an
element of this surface be dS, and let its direction-cosines be /„ m,, n, with
respect to the first vortex, and Z,, m^, n, with respect to the second; then we
know that

Z, + Z, = 0, m, + OT1 = 0, ni + n, = 0 ..................... (29).

The values of a, & y vary with the position of the centre of the vortex ;
so that we may write

a , * (la , . da , «

with similar equations for /3 and y.

The value of u may be written : —

u = i {m, (x - xt) + m, (a; -a;,)}

In effecting the summation of 2,updS, we must remember that round any
closed surface ZldS and all similar terms vanish; also that terms of the form
SlydS, where / and y are measured in different directions, also vanish ; but that
terms of the form tlxdS, where I and x refer to the same axis of co-ordinates,
do not vanish, but are equal to the volume enclosed by the surface. The
result is

ON PHYSICAL LINES OF FORCE. 471

or dividing by V= F,+ F, + &c.,

If we make P = ir-

ZTT

then equation (33) will be identical with the first of equations (9), which give
the relation between the quantity of an electric current and the intensity of
the lines of force surrounding it.

It appears therefore that, according to our hypothesis, an electric current
is represented by the transference of the moveable particles interposed between
the neighbouring vortices. We may conceive that these particles are very small
compared with the size of a vortex, and that the mass of all the particles
together is inappreciable compared with that of the vortices, and that a great
many vortices, with their surrounding particles, are contained in a single complete
molecule of the medium. The particles must be conceived to roll without sliding
between the vortices which they separate, and not to touch each other, so that,
as long as they remain within the same complete molecule, there is no loss of
energy by resistance. When, however, there is a general transference of par-
ticles in one direction, they must pass from one molecule to another, and in
doing so, may experience resistance, so as to waste electrical energy and generate
heat.

Now let us suppose the vortices arranged in a medium in any arbitrary
manner. The quantities -, - — -—- , &c. will then in general have values, so that

there will at first be electrical currents in the medium. These will be opposed
by the electrical resistance of the medium ; so that, unless they are kept up
by a continuous supply of force, they will quickly disappear, and we shall then

have -?• — -7- = 0, &c. ; that is, adx + ftdy + ydz will be a complete differential

(see equations (15) and (16)); so that our hypothesis accounts for the distri-
bution of the lines of force.

In Plate VIII. p. 488, fig. 1, let the vertical circle EE represent an
electric current flowing from copper C to zinc Z through the conductor EE',
as shewn by the arrows.

472 ON PHYSICAL LINES OF FORCE.

Let the horizontal circle MM' represent a line of magnetic force embracing
the electric circuit, the north and south directions being indicated by the lines
SN and NS.

Let the vertical circles V and V represent the molecular vortices of which
the line of magnetic force is the axis. V revolves as the hands of a watch,
and V the opposite way.

It will appear from this diagram, that if V and V were contiguous vortices,
particles placed between them would move downwards ; and that if the particles
were forced downwards by any cause, they would make the vortices revolve as
in the figure. We have thus obtained a point of view from which we may
regard the relation of an electric current to its lines of force as analogous to
the relation of a toothed wheel or rack to wheels which it drives.

In the first part of the paper we investigated the relations of the statical
forces of the system. We have now considered the connexion of the motions
of the parts considered as a system of mechanism. It remains that we should
investigate the dynamics of the system, and determine the forces necessary to
produce given changes in the motions of the different parts.

PROP. VI. — To determine the actual energy of a portion of a medium due
to the motion of the vortices within it.

Let a, /3, y be the components of the circumferential velocity, as in Prop. II.,
then the actual energy of the vortices in unit of volume will be proportional
to the density and to the square of the velocity. As we do not know the
distribution of density and velocity in each vortex, we cannot determine the
numerical value of the energy directly ; but since p also bears a constant
though unknown ratio to the mean density, let us assume that the energy
in unit of volume is

where C is a constant to be determined.
Let us take the case in which

dx ' dy ' ' dz '
Let <£ = <£, + <£, (36),

ON PHYSICAL LINES OF FORCE. 473

then fa Is the potential at any point due to the magnetic system m1} and fa
that due to the distribution of magnetism represented by m,. The actual
energy of all the vortices is

E^SCpp + F + tfdV ............................. (38),

the integration being performed over all space.

This may be shewn by integration by parts (see Green's ' Essay on Elec-
tricity/ p. 10) to be equal to

E= -^(^(fam. + fam^fam^ + fam^dV ................. (39).

Or since it has been proved (Green's 'Essay,' p. 10) that

^ V,

V'. ................. (40).

Now let the magnetic system ml remain at rest, and let m, be moved
parallel to itself in the direction of x through a space So; ; then, since fa
depends on m^ only, it will remain as before, so that fan^ will be constant ;
and since fa depends on wj, only, the distribution of fa about m, will remain
the same, so that famt will be the same as before the change. The only part
of E that will be altered is that depending on 2fam,, because fa becomes

fa + -y-1 8x on account of the displacement. The variation of actual energy due
to the displacement is therefore

ax

But by equation (12) the work done by the mechanical forces on m3 during
the motion is

(42);

and since our hypothesis is a purely mechanical one, we must have by the
conservation of force,

8E + 8W=0 ............................... (43);

that is, the loss of energy of the vortices must be made up by work done in
moving magnets, so that

2 mtdv Bx + 2 ' <m3d V 8x = 0,
\ ctx / \ct/x /

c=sl ...................................... <44>-

VOL. I. 60

474 ON PHYSICAL LINES OF FORCE.

so that the energy of the vortices in unit of volume is
and that of a vortex whose volume is V is

(46).

In order to produce or destroy this energy, work must be expended on,
or received from, the vortex, either by the tangential action of the layer of
particles in contact with it, or by change of form in the vortex. We shall first
investigate the tangential action between the vortices and the layer of particles
in contact with them.

PROP. VII. — To find the energy spent upon a vortex in unit of tune by
the layer of particles which surrounds it.

Let P, Q, R be the forces acting on unity of the particles in the three
co-ordinate directions, these quantities being functions of x, y, and z. Since
each particle touches two vortices at the extremities of a diameter, the reaction
of the particle on the vortices will be equally divided, and will be

-\P, -iQ, -\R

on each vortex for unity of the particles; but since the superficial density of

the particles is - - (see equation (34) ), the forces on unit of surface of a vortex
ZTT

will be

Now let dS be an element of the surface of a vortex. Let the direction-cosines
of the normal be I, ra, n. Let the co-ordinates of the element be x, y, z. Let
the component velocities of the surface be u, v, w. Then the work expended on
that element of surface will be

(47).

Let us begin with the first term, PudS. P may be written

dP dP dP

and u = n/3 — my.

ON PHYSICAL LINES OF FORCE. 475

Remembering that the surface of the vortex is a closed one, so that

ZnxdS = ?,mxdS = ZmydS = tmzdS = 0,
and 'S,mydS=tnzdS= V,

we find tPudS=(^/3-^y) V ........................ (49),

and the whole work done on the vortex in unit of time will be

rUf i

^ = -~ 2 (Pu + Qv + Rw) dS
at 477

1 ( fdQ dR\^0fdR dP\ (dP dQ\\v .

= — - lai-r- -- i-) + P -7 --- T~ +7 j -- -rlf * ...... (50).

<iir\ \dz dyj \dx dz/ \dy / dx/j

PROP. VIII. — To find the relations between the alterations of motion of the
vortices, and the forces P, Q, R which they exert on the layer of particles
between them.

Let V be the volume of a vortex, then by (46) its energy is

V ......................... (51),

dE 1 ,,/ da Odf3 , dy\ , .

and -jr = 7-/A^ a -77 + £-£+r -57) ..................... (52).

at ±irr \ at dt ' at)

Comparing this value with that given in equation (50), we find

dR da/dR dP cZ£\ IdP , .

--~~0 ...... (53)-

This equation being true for all values of a, ft, and y, first let ft and y
vanish, and divide by a. We find

Q. ., ,
Similarly,

dQ dR da

___ ^* ^™ - ~~7 II — —

dz dy dt
dR

(54).

dP dQ dy

and -^— — y51 = IL -r-

dy ax dt

From these equations we may determine the relation between the alterations
of motion -7-, &c. and the forces exerted on the layers of particles between

60—2

476 ON PHYSICAL LINES OF FORCE.

the vortices, or, in the language of our hypothesis, the relation between changes
in the state of the magnetic field and the electromotive forces thereby brought
into play.

In a memoir "On the Dynamical Theory of Diffraction" (Cambridge Philo-
sophical Transactions, Vol. IX. Part 1, section 6), Professor Stokes has given a
method by which we may solve equations (54), and find P, Q, and R in terms
of the quantities on the right hand of those equations. I have pointed out*
the application of this method to questions in electricity and magnetism.

Let us then find three quantities F, G, H from the equations

dG_dH
dz dy

dH

(fj5\

rfo : ~ dz ~^ t • • I00/'

dF__dG
dy dx

lid . d d

with the conditions — l-j- pa + -r-pB + , p.-y}=m = Q ................ (56),

dF dG dH
and + + -

Differentiating (55) with respect to t, and comparing with (54), we find

- -

dt ' - dt' ' dt

We have thus determined three quantities, F, G, H, from which we can
find P, Q, and R by considering these latter quantities as the rates at which
the former ones vary. In the paper already referred to, I have given reasons
for considering the quantities F, G, H as the resolved parts of that which
Faraday has conjectured to exist, and has called the electrotonic state. In that
paper I have stated the mathematical relations between this electrotonic state
and the lines of magnetic force as expressed in equations (55), and also between
the electrotonic state and electromotive force as expressed in equations (58). We
must now endeavour to interpret them from a mechanical point of view in
connexion with our hypothesis.

* Cambndge Philosojrfiical Transactions, Vol. X. Part I. Art. 3. "On Faraday's Lines of Force,"
pp. 205—209 of thia vol.

ON PHYSICAL LINES OF FORCE. 477

We shall in the first place examine the process by which the lines of force
are produced by an electric current.

Let AB, Plate VIII. , p. 488, fig. 2, represent a current of electricity in the
direction from A to B. Let the large spaces above and below AB represent the
vortices, and let the small circles separating the vortices represent the layers of
particles placed between them, which in our hypothesis represent electricity.

Now let an electric current from left to right commence in AB. The
row of vortices gh above AB will be set in motion in the opposite direction
to that of a watch. (We shall call this direction +, and that of a watch — .)
We shall suppose the row of vortices kl still at rest, then the layer of particles
between these rows will be acted on by the row gh on their lower sides, and
will be at rest above. If they are free to move, tKey will rotate in the
negative direction, and will at the same time move from right to left, or in
the opposite direction from the current, and so form an induced electric current.

If this current is checked by the electrical resistance of the medium, the
rotating particles will act upon the row of vortices Id, and make them revolve
in the positive direction till they arrive at such a velocity that the motion of
the particles is reduced to that of rotation, and the induced current disappears.
If, now, the primary current AB be stopped, the vortices in the row gh will
be checked, while those of the row kl still continue in rapid motion. The
momentum of the vortices beyond the layer of particles pq will tend to move
them from left to right, that is, in the direction of the primary current ; but
if this motion is resisted by the medium, the motion of the vortices beyond pq
will be gradually destroyed.

It appears therefore that the phenomena of induced currents are part of the
process of communicating the rotatory velocity of the vortices from one part of
the field to another.

As an example of the action of the vortices in producing induced currents,
let us take the following case : — Let B, Plate VIII., p. 488, fig. 3, be a circular
ring, of uniform section, lapped uniformly with covered wire. It may be shewn
that if an electric current is passed through this wire, a magnet placed within
the coil of wire will be strongly affected, but no magnetic effect will be produced
on any external point. The effect will be that of a magnet bent round till
its two poles are in contact.

If the coil is properly made, no effect on a magnet placed outside it can

478 ON PHYSICAL LINES OF FORCE.

be discovered, whether the current is kept constant or made to vary in strength ;
but if a conducting wire C be made to embrace the ring any number of times,
an electromotive force will act on that wire whenever the current in the coil is
made to vary ; and if the circuit be closed, there will be an actual current in
the wire C.

This experiment shews that, in order to produce the electromotive force, it
is not necessary that the conducting wire should be placed in a field of magnetic
force, or that lines of magnetic force should pass through the substance of the
wire or near it. All that is required is that lines of force should pass through
the circuit of the conductor, and that these lines of force should vary in quantity
during the experiment.

In this case the vortices, of which we suppose the lines of magnetic force
to consist, are all within the hollow of the ring, and outside the ring all is at
rest. If there is no conducting circuit embracing the ring, then, when the
primary current is made or broken, there is no action outside the ring, except
an instantaneous pressure between the particles and the vortices which they
separate. If there is a continuous conducting circuit embracing the ring, then,
when the primary current is made, there will be a current in the opposite
direction through C; and when it is broken, there will be a current through C
in the same direction as the primary current.

We may now perceive that induced currents are produced when the elec-
tricity yields to the electromotive force, — this force, however, still existing
when the formation of a sensible current is prevented by the resistance of the
circuit.

The electromotive force, of which the components are P, Q, R, arises from
the action between the vortices and the interposed particles, when the velocity
of rotation is altered in any part of the field. It corresponds to the pressure
on the axle of a wheel in a machine when the velocity of the driving wheel
is increased or diminished.

The electrotonic state, whose components are F, G, II, is what the electromotive
force would be if the currents, &c. to which the lines of force are due, instead
of arriving at their actual state by degrees, had started instantaneously from
rest with their actual values. It corresponds to the impulse which would act
on the axle of a wheel in a machine if the actual velocity were suddenly given
to the driving wheel, the machine being previously at rest.

ON PHYSICAL LINES OF FORCE. 479

If the machine were suddenly stopped by stopping the driving wheel, each
wheel would receive an impulse equal and opposite to that which it received
when the machine was set in motion.

This impulse may be calculated for any part of a system of mechanism,
and may be called the reduced momentum of the machine for that point. In
the varied motion of the machine, the actual force on any part arising from
the variation of motion may be found by differentiating the reduced momentum
with respect to the time, just as we have found that the electromotive force
may be deduced from the electrotonic state by the same process.

Having found the relation between the velocities of the vortices and the
electromotive forces when the centres of the vortices are at rest, we must
extend our theory to the case of a fluid medium containing vortices, and
subject to all the varieties of fluid motion. If we fix our attention on any
one elementary portion of a fluid, we shall find that it not only travels from
one place to another, but also changes its form and position, so as to be elon-
gated in certain directions and compressed in others, and at the same time (in
the most general case) turned round by a displacement of rotation.

These changes of form and position produce changes in the velocity of the
molecular vortices, which we must now examine.

The alteration of form and position may always be reduced to three simple
extensions or compressions in the direction of three rectangular axes, together
with three angular rotations about any set of three axes. We shall first con-
sider the effect of three simple extensions or compressions.

PROP. IX. — To find the variations of a, ft, y in the parallelepiped x, y, z
when x becomes x + Bx; y, y + 8y; and z, z + Sz; the volume of the figure
remaining the same.

By Prop. II. we find for the work done by the vortices against pressure,

8 W=p$ (xyz) - £- (a'yzSx + fizxSy + -/xy$z) (59) ;

and by Prop. VI. we find for the variation of energy,

(60).

480 ON PHYSICAL LINES OF FORCE.

The sum 8W+8E must be zero by the conservation of energy, and &(xyz) = Q,
since xyz is constant; so that

(61).

In order that this should be true independently of any relations between a, /3,
and y, we must have

Sa = a^, S0 = j8^, Sy = y^ ...................... (62).

x y z

PROP. X. — To find the variations of a, ft, y due to a rotation 6l about the
axis of x from y to z, a rotation 0t about the axis of y from z to x, and a
rotation Ot about the axis of z from x to y.

The axis of ft will move away from the axis of x by an angle 0, ; so
that ft resolved in the direction of x changes from 0 to — ft0t.

The axis of y approaches that of x by an angle 63 ; so that the resolved
part of y in direction x changes from 0 to yd*.

The resolved part of a in the direction of x changes by a quantity depending
on the second power of the rotations, which may be neglected. The variations of
a, ft, y from this cause are therefore

Ba = y0t-ft0t, 8ft = a0i-y01, Sy = ^-a^ ............... (63).

The most general expressions for the distortion of an element produced by
the displacement of its different parts depend on the nine quantities

d « d « d ~ d ~ d ~ d ., eZ~ d « d -,
-j-ox, -3- ox, -j-Sx; -j-oy, -j-oy, -j-oy; -r-oz, -y-8z, -5-02;
dx dy dz dx y dy y dz y dx dy dz

and these may always be expressed in terms of nine other quantities, namely,
three simple extensions or compressions,

Sx' 8y' Sz;
x" y" z-

along three axes properly chosen, x', y", z', the nine direction-cosines of these
axes with their six connecting equations, which are equivalent to three inde-
pendent quantities, and the three rotations #„ &„ 0S about the axes of x, y, z.

Let the direction-cosines of x' with respect to x, y, z be la ra,, nlt those of
//', I., in,, na and those of z', ln m^ nt; then we find

ON PHYSICAL LINES OF FOKCE.

481

x

y

d
j-

d

Sz'
--r

dz'

(64),

with similar equations for quantities involving Sy and 8z.

Let a, ft, y be the values of a, ft, y referred to the axes x, y', z ; then

(65).

y = l,a + mj$ 4- n3y J
We shall then have 8a = ll8a+l$ft' + l38y' + y0t-ft03 (66),

-ft0, (67).

By substituting the values of a, f?, y', and comparing with equations (64), we
find

(68)

as the variation of a due to the change of form and position of the element.
The variations of ft and y have similar expressions.

PROP. XI. — To find the electromotive forces in a moving body.

The variation of the velocity of the vortices in a moving element is due to
two causes — the action of the electromotive forces, and the change of form and
position of the element. The whole variation of a is therefore

8* ............... <69)-

But since a is a function of x, y, z and t, the variation of a may be also written

Sa^Sx + ^Sy+^Sz + ^fc ..................... (70).

dx ay dz at

Equating the two values of Sa and dividing by S^, and remembering that in the
motion of an incompressible medium

a. ,_ _n

dx dt "*" dy dt + dz dt ~

VOL. I.

61

482 ON PHYSICAL LINES OF FORCE.

and that in the absence of free magnetism

we find

l(dQ_d_R\ d d_x_ d_dz

dy dx da. dz da. dy dft dx _ da _ , v

*"dz~d~t~dz'di~fydi + dydi~di =

putting a=K^"W) (74)>

aa 1 /a'Cr d1JI\ C7z\

and -7- = - I , ., — j 1. 1 v/3J>

dt fj. \dz dt dy dt]

where F, G, and // are the values of the electrotonic components for a fixed
point of space, our equation becomes

d fn dx dz dG\ d /„ <!;/ fidx dlf\

The expressions for the variations of $ and y give us two other equations
which may be written down from symmetry. The complete solution of the three
equations is

p- ty.- fi^ ^_ —
~ W ~dt~ PP dt dt dx

dz dx . dG dV , ^

Odx di/ dll dV
= --.a-- + -- —

—

The first and second terms of each equation indicate the effect of the motion
of any body in the magnetic field, the third term refers to changes in the
electrotonic state produced by alterations of position or intensity of magnets
or currents in the field, and V is a function of x, y, z, and t, which is inde-
terminate as far . as regards the solution of the original equations, but which
may always be determined in any given case from the circumstances of the
problem. The physical interpretation of ¥ is, that it is the electric tension at
each point of space.

ON PHYSICAL LINES OF FORCE. 483

The physical meaning of the terms in the expression for the electromotive
force depending on the motion of the body, may be made simpler by supposing
the field of magnetic force uniformly magnetized with intensity a in the direction
of the axis of x. Then if I, m, n be the direction-cosines of any portion of a
linear conductor, and S its length, the electromotive force resolved in the direction
of the conductor will be

+ Rn) ........................... (78),

dz

that is, the product of pa, the quantity of magnetic induction over unit of area

multiplied by S ( m " — n -3? j , the area swept out by the 'conductor S in unit of
\ *•*/

time, resolved perpendicular to the direction of the magnetic force.

The electromotive force in any part of a conductor due to its motion is
therefore measured by the number of lines of magnetic force which it crosses
in unit of time ; and the total electromotive force in a closed conductor is
measured by the change of the number of lines of force which pass through it ;
and this is true whether the change be produced by the motion of the con-
ductor or by any external cause.

In order to understand the mechanism by which the motion of a conductor
across lines of magnetic force generates an electromotive force in that conductor,
we must remember that in Prop. X. we have proved that the change of form
of a portion of the medium containing vortices produces a change of the velocity
of those vortices ; and in particular that an extension of the medium in the
direction of the axes of the vortices, combined with a contraction in all direc-
tions perpendicular to this, produces an increase of velocity of the vortices ;
while a shortening of the axis and bulging of the sides produces a diminution
of the velocity of the vortices.

Tliis change of the velocity of the vortices arises from the internal effects
of change of form, and is independent of that produced by external electro-
motive forces. If, therefore, the change of velocity be prevented or checked,
electromotive forces will arise, because each vortex will press on the surrounding
particles in the direction in which it tends to alter its motion.

Let A, fig. 4, p. 488, represent the section of a vertical wire moving in the
direction of the arrow from west to east, across a system of lines of magnetic force

61—2

484 ON PHYSICAL LINES OF FORCE.

running north and south. The curved lines in fig. 4 represent the lines of fluid
motion about the wire, the wire being regarded as stationary, and the fluid as
having a motion relative to it. It is evident that, from this figure, we can trace
the variations of form of an element of the fluid, as the form of the element
depends, not on the absolute motion of the whole system, but on the relative
motion of its parts.

In front of the wire, that is, on its east side, it will be seen that as the
wire approaches each portion of the medium, that portion is more and more
compressed in the direction from east to west, and extended in the direction
from north to south ; and since the axes of the vortices lie in the north and
south direction, their velocity will continually tend to increase by Prop. X.,
unless prevented or checked by electromotive forces acting on the circumference
of each vortex.

We shall consider an electromotive force as positive when the vortices tend
to move the interjacent particles upwards perpendicularly to the plane of the
paper.

The vortices appear to revolve as the hands of a watch when we look at
them from south to north ; so that each vortex moves upwards on its west side,
and downwards on its east side. In front of the wire, therefore, where each
vortex is striving to increase its velocity, the electromotive force upwards must
be greater on its west than on its east side. There will therefore be a con-
tinual increase of upward electromotive force from the remote east, where it is
zero, to the front of the moving wire, where the upward force will be strongest.

Behind the wire a different action takes place. As the wire moves away
from each successive portion of the medium, that portion is extended from east
to west, and compressed from north to south, so as to tend to diminish the
velocity of the vortices, and therefore to make the upward electromotive force
greater on the east than on the west side of each vortex. The upward electro-
motive force will therefore increase continually from the remote west, where it
is zero, to the back of the moving wire, where it will be strongest.

It appears, therefore, that a vertical wire moving eastwards will experience
an electromotive force tending to produce in it an upward current. If there
is no conducting circuit in connexion with the ends of the wire, no current will
be formed, and the magnetic forces will not be altered ; but if such a circuit
exists, there will be a current, and the lines of magnetic force and the velocity

ON PHYSICAL LINES OF FORCE. 485

of the vortices will be altered from their state previous to the motion of the
wire. The change in the lines of force is shewn in fig. 5. The vortices in
front of the wire, instead of merely producing pressures, actually increase in
velocity, while those behind have their velocity diminished, and those at the
sides of the wire have the direction of their axes altered; so that the final
effect is to produce a force acting on the wire as a resistance to its motion.
We may now recapitulate the assumptions we have made, and the results we
have obtained.

(1) Magneto-electric phenomena are due to the existence of matter under
certain conditions of motion or of pressure in every part of the magnetic field,
and not to direct action at a distance between the magnets or currents. The
substance producing these effects may be a certain part of ordinary matter, or
it may be an aether associated with matter. Its density is greatest in iron,
and least in diamagnetic substances ; but it must be in all cases, except that of
iron, very rare, since no other substance has a large ratio of magnetic capacity
to what we call a vacuum.

(2) The condition of any part of the field, through which lines of magnetic
force pass, is one of unequal pressure in different directions, the direction of
the lines of force being that of least pressure, so that the lines of force may
be considered lines of tension.

(3) This inequality of pressure is produced by the existence in the medium
of vortices or eddies, having their axes in the direction of the lines of force,
and having their direction of rotation determined by that of the lines of force.

We have supposed that the direction was that, of a watch to a spectator
looking from south to north. We might with equal propriety have chosen the
reverse direction, as far as known facts are concerned, by supposing resinous elec-
tricity instead of vitreous to be positive. The effect of these vortices depends
on their density, and on their velocity at the circumference, and is independent
of their diameter. The density must be proportional to the capacity of the
substance for magnetic induction, that of the vortices in air being 1. The
velocity must be very great, in order to produce so powerful effects in so rare
a medium.

The size of the vortices is indeterminate, but is probably very small as
compared with that of a complete molecule of ordinary matter*.

* The angular momentum of the system of vortices depends on their average diameter ; so that if the
diameter were sensible, we might expect that a magnet would behave as if it contained a revolving body

ON PHYSICAL LINES OF FORCE.

(4) The vortices are separated from each other by a single layer of round
particles, so that a system of cells is formed, the partitions being these layers
of particles, and the substance of each cell being capable of rotating as a vortex.

(5) The particles forming the layer are in rolling contact with both the
vortices which they separate, but do not rub against each other. They are
perfectly free to roll between the vortices and so to change their place, provided
they keep within one complete molecule of the substance ; but in passing from
one molecule to another they experience resistance, and generate irregular
motions, which constitute heat. These particles, in our theory, play the part of
electricity. Their motion of translation constitutes an electric current, their
rotation serves to transmit the motion of the vortices from one part of the
field to another, and the tangential pressures thus called into play constitute
electromotive force. The conception of a particle having its motion connected
with that of a vortex by perfect rolling contact may appear somewhat awkward.
I do not bring it forward as a mode of connexion existing in nature, or even
as that which I would willingly assent to as an electrical hypothesis. It is,
however, a mode of connexion which is mechanically conceivable, and easily
investigated, and it serves to bring out the actual mechanical connexions
between the known electro-magnetic phenomena ; so that I venture to say that
any one who understands the provisional and temporary character of this
hypothesis, will find himself rather helped than hindered by it in his search
after the true interpretation of the phenomena.

The action between the vortices and the layers of particles is in part
tangential ; so that if there were any slipping or differential motion between
the parts in contact, there would be a loss of the energy belonging to the
lines of force, and a gradual transformation of that energy into heat. Now we
know that the lines of force about a magnet are maintained for an indefinite
time without any expenditure of energy ; so that we must conclude that
wherever there is tangential action between different parts of the medium, there
is no motion of slipping between those parts. We must therefore conceive that
the vortices and particles roll together without slipping ; and that the interior
strata of each vortex receive their proper velocities from the exterior stratum
without slipping, that is, the angular velocity must be the same throughout each
vortex.

within it, and that the existence of this rotation might be detected by experiments on the free rotation of a
magnet. I have made experiments to investigate this question, but have not yet fully tried the apparatus.

ON PHYSICAL LINES OF FORCE. 487

The only process in which electro-magnetic energy is lost and transformed
into heat, is in the passage of electricity from one molecule to another. In all
other cases the energy of the vortices can only be diminished when an equivalent
quantity of mechanical work is done by magnetic action.

(6) The effect of an electric current upon the surrounding medium is to
make the vortices in contact with the current revolve so that the parts next
to the current move in the same direction as the current. The parts furthest
from the current will move in the opposite direction ; and if the medium is a
conductor of electricity, so that the particles are free to move in any direction,
the particles touching the outside of these vortices will be moved in a direction
contrary to that of the current, so that there will be an induced current in
the opposite direction to the primary one.

If there were no resistance to the motion of the particles, the induced
current would be equal and opposite to the primary one, and would continue
as long as the primary current lasted, so that it would prevent all action of
the primary current at a distance. If there is a resistance to the induced
current, its particles act upon the vortices beyond them, and transmit the motion
of rotation to them, till at last all the vortices in the medium are set in
motion with such velocities of rotation that the particles between them have no
motion except that of rotation, and do not produce currents.

In the transmission of the motion from one vortex to another, there arises a
force between the particles and the vortices, by which the particles are pressed
in one direction and the vortices in the opposite direction. We call the force
acting on the particles the electromotive force. The reaction on the vortices is
equal and opposite, so that the electromotive force cannot move any part of
the medium as a whole, it can only produce currents. When the primary
current is stopped, the electromotive forces all act in the opposite direction.

(7) When an electric current or a magnet is moved in presence of a
conductor, the velocity of rotation of the vortices in any part of the field is
altered by that motion. The force by which the proper amount of rotation is
transmitted to each vortex, constitutes in this case also an electromotive force,
and, if permitted, will produce currents.

(8) When a conductor is moved in a field of magnetic force, the vortices
in it and in its neighbourhood are moved out of their places, and are changed
in form. The force arising from these changes constitutes the electromotive

488 ON PHYSICAL LINES OF FORCE.

force on a moving conductor, and is found by calculation to correspond with
that determined by experiment.

We have now shewn in what way electro-magnetic phenomena may be
imitated by an imaginary system of molecular vortices. Those who have been
already inclined to adopt an hypothesis of -this kind, will find here the con-
ditions which must be fulfilled in order to give it mathematical coherence, and
a comparison, so far satisfactory, between its necessary results and known facts.
Those who look in a different direction for the explanation of the facts, may
be able to compare this theory with that of the existence of currents flowing
freely through bodies, and with that which supposes electricity to act at a
distance with a force depending on its velocity, and therefore not subject to
the law of conservation of energy.

The facts of electro-magnetism are so complicated and various, that the
explanation of any number of them by several different hypotheses must be
interesting, not only to physicists, but to all who desire to understand how
much evidence the explanation of phenomena lends to the credibility of a theory,
or how far we ought to regard a coincidence in the mathematical expression of
two sets of phenomena as an indication that these phenomena are of the same
kind. We know that partial coincidences of this kind have been discovered ;
and the fact that they are only partial is proved by the divergence of the
laws of the two sets of phenomena in other respects. We may chance to find,
in the higher parts of physics, instances of more complete coincidence, which
may require much investigation to detect their ultimate divergence.

NOTE.

Since the firsc part of this paper was written, I have seen in Crelle's Journal for 1859,
a paper by Prof. Helmholtz on Fluid Motion, in which he has pointed out that the lines
of fluid motion are arranged according to the same laws as the lines of magnetic force, the
path of an electric current corresponding to a line of axes of those particles of the fluid
which are in a state of rotation. This is an additional instance of a physical analogy, the
investigation of which may illustrate both electro-magnetism and hydrodynamics.

VOL. I. PLATE VIII.

To fur,- pnrjr 488

Cambridge CfauMraityr Pros*

ON PHYSICAL LINES OF FOKCE. 489

[From the Philosophical Magazine for January and February, 1862.]

PART III.

THE THEORY OF MOLECULAR VORTICES APPLIED TO STATICAL ELECTRICITY.

In the first part of this paper* I have shewn how the forces acting between
magnets, electric currents, and matter capable of magnetic induction may be
accounted for on the hypothesis of the magnetic field being occupied with
innumerable vortices of revolving matter, their axes coinciding with the direction
of the magnetic force at every point of the field.

The centrifugal force of these vortices produces pressures distributed in such
a way that the final effect is a force identical in direction and magnitude
with that which we observe.

In. the second partf I described the mechanism by which these rotations
may be made to coexist, and to be distributed according to the known laws
of magnetic lines of force.

I conceived the rotating matter to be the substance of certain cells, divided
from each other by cell- walls composed of particles which are very small com-
pared with the cells, and that it is by the motions of these particles, and their
tangential action on the substance in the cells, that the rotation is communi-
cated from one cell to another.

I have not attempted to explain this tangential action, but it is necessary
to suppose, in order to account for the transmission of rotation from the exterior
to the interior parts of each cell, that the substance in the cells possesses
elasticity of figure, similar in kind, though different in degree, to that observed
in solid bodies. The uudulatory theory of light requires us to admit this kind
of elasticity in the luminiferous medium, in order to account for transverse
vibrations. We need not then be surprised if the magneto-electric medium
possesses the same property.

* Phil. Mag. March, 1861 [pp. 451—466 of this vol.].

t Phil. Mag. April and May, 1861 [pp. 467—488 of this vol.].

VOL. I. 62

490 ON PHYSICAL LINES OF FORCE.

According to our theory, the particles which form the partitions between
the cells constitute the matter of electricity. The motion of these particles
constitutes an electric current; the tangential force with which the particles
are pressed by the matter of the cells is electromotive force, and the pressure
of the particles on each other corresponds to the tension or potential of the
electricity.

If we can now explain the condition of a body with respect to the
surrounding medium when it is said to be "charged" with electricity, and
account for the forces acting between electrified bodies, we shall have established
a connexion between all the principal phenomena of electrical science.

We know by experiment that electric tension is the same thing, whether
observed in statical or in current electricity; so that an electromotive force
produced by magnetism may be made to charge a Leyden jar, as is done by
the coil machine.

When a difference of tension exists in different parts of any body, the
electricity passes, or tends to pass, from places of greater to places of smaller
tension. If the body is a conductor, an actual passage of electricity takes
place; and if the difference of tensions is kept up, the current continues to
flow with a velocity proportional inversely to the resistance, or directly to the
conductivity of the body.

The electric resistance has a very wide range of values, that of the metals
being the smallest, and that of glass being so great that a charge of electricity
has been preserved* in a glass vessel for years without penetrating the thick-
ness of the glass.

Bodies which do not permit a current of electricity to flow through them
are called insulators. But though electricity does not flow through them,
the electrical effects are propagated through them, and the amount of these
effects differs according to the nature of the body; so that equally good insu-
lators may act differently as dielectrics f.

Here then we have two independent qualities of bodies, one by which they
allow of the passage of electricity through them, and the other by which they
allow of electrical action being transmitted through them without any electri-
city being allowed to pass. A conducting body may be compared to a porous
membrane which opposes more or less resistance to the passage of a fluid,

* By Professor W. Thomson. t Faraday, Experimental Ittxearcftes, Series xi.

ON PHYSICAL LINES OF FORCE. 491

while a dielectric is like an elastic membrane which may be impervious to the
fluid, but transmits the pressure of the fluid on one side to that on the other.

As long as electromotive force acts on a conductor, it produces a current
which, as it meets with resistance, occasions a continual transformation of
electrical energy into heat, which is incapable of being restored again as electri-
cal energy by any reversion of the process.

Electromotive force acting on a dielectric produces a state of polarization
of its parts similar in distribution to the polarity of the particles of iron under
the influence of a magnet*, and, like the magnetic polarization, capable of
being described as a state in which every particle has its poles in opposite

conditions.

/

In a dielectric under induction, we may conceive that the electricity in
each molecule is so displaced that one side is rendered positively, and the
other negatively electrical, but that the electricity remains entirely connected
with the molecule, and does not pass from one molecule to another.

The effect of this action on the whole dielectric mass is to produce a
general displacement of the electricity in a certain direction. This displace-
ment does not amount to a current, because when it has attained a certain
value it remains constant, but it is the commencement of a current, and its
variations constitute currents in the positive or negative direction, according as
the displacement is increasing or diminishing. The amount of the displacement
depends on the nature of the body, and on the electromotive force ; so that
if h is the displacement, R the electromotive force, and E a coefficient
depending on the nature of the dielectric,

R = -

and if r is the value of the electric current due to displacement,

_dh
~dt'

These relations are independent of any theory about the internal mechanism
of dielectrics ; but when we find electromotive force producing electric displace-
ment in a dielectric, and when we find the dielectric recovering from its state
of electric displacement with an equal electromotive force, we cannot help

* See Prof. Mossotti, " Discussione Analitica," Memorie della Soc. Italiana (Modena), Vol. xxiv.
Part 2, p. 49.

62—2

492 ON PHYSICAL LINES OF FORCE.

regarding the phenomena as those of an elastic body, yielding to a pressure,
and recovering its form when the pressure is removed.

According to our hypothesis, the magnetic medium is divided into cells,
separated by partitions formed of a stratum of particles which play the part
of electricity. When the electric particles are urged in any direction, they will,
by their tangential action on the elastic substance of the cells, distort each
cell, and call into play an equal and opposite force arising from the elasticity
of the cells. When the force is removed, the cells will recover their form,
and the electricity will return to its former position.

In the following investigation I have considered the relation between the
displacement and the force producing it, on the supposition that the cells are
spherical. The actual form of the cells probably does not differ from that of
a sphere sufficiently to make much difference in the numerical result.

I have deduced from this result the relation between the statical and
dynamical measures of electricity, and have shewn, by a comparison of the
electro- magnetic experiments of MM. Kohlrausch and Weber with the velocity
of light as found by M. Fizeau, that the elasticity of the magnetic medium
in air is the same as that of the luminiferous medium, if these two coex-
istent, coextensive, and equally elastic media are not rather one medium.

It appears also from Prop. XV. that the attraction between two electrified
bodies depends on the value of E*, and that therefore it would be less in
turpentine than in air, if the quantity of electricity in each body remains the
same. If, however, the potentials of the two bodies were given, the attraction
between them would vary inversely as E*, and would be greater in turpentine
than in air.

PROP. XII. To find the conditions of equilibrium of an elastic sphere
whose surface is exposed to normal and tangential forces, the tangential forces
being proportional to the sine of the distance from a given point on the sphere.

Let the axis of 2 be the axis of spherical co-ordinates.

Let £, 77, £ be the displacements of any particle of the sphere in the direc-
tions of x, y, and z.

Let PXB, pn, pa be the stresses normal to planes perpendicular to the three
axes, and let pv, pm p^ be the stresses of distortion in the planes yz, zx,
and xy.

OX PHYSICAL LIXES OF FORCE.

Let p. be the coefficient of cubic elasticity, so that if

493

Let m be the coefficient of rigidity, so that

dr)\ ,

— _-£ I j &(».

Then we have the following equations of elasticity in an isotropic medium,

, x (dj; d-n d£\ d%

ff\ — f ii ^— .A.-jyi I I —I— — — — L -- .. I J— Vfli ' — *

f xx \i o / \ /V'/* fill ft 7. I Ci 'IT

with similar equations in y and z, and also

ft.-^f^ + ^)f &c (83).

z \az ay I

In the case of the sphere, let us assume the radius = a, and

£=exz, rj = ezy, t,=f(x' + 'i/f)-\-gz1 + d (84).

Then pa = 2(fj. — §m)

(e+g)z + 2mgz

(85).

The equation of internal equilibrium with respect to z is

d d d

dx dy "* dz • \ )>

which is satisfied in this case if

m (e + 2f+2g) + 2 (/A — -Jm) (e+g) = 0 (87)-

The tangential stress on the surface of the sphere, whose radius is a at
an angular distance 0 from the axis in plane xz,

.(88) r: ,

.(89).

494 ON PHYSICAL UXES Of FORCE.

In order that T may be proportional to sin0, the first term must vanish, and
therefore

?-«+/ ............................................. (90),

T=-~(e + 2f)Bm0 ........................... (91).

The normal stress on the surface at any point is

N=p<a sin1 0 + pm cos' 0 + 2pa sin 6 cos 6
= 2 (/i - £?n) (e +#) a cos 0 + 2ma cos 0 {(e +/) sin5 Q +g cos1 0} ...... (92) ;

or by (87) and (90), N= -ma(e + 2f) cos 0 ........................... (93).

The tangential displacement of any point is

« = £cos0-£sin0 = - (a!/+c/)sin0 ..................... (94).

The normal displacement is

7i = £sin0 + £cos0 = {a'(e+/) + d}cos0 .................. (95).

If we make a'(e+f) + d = 0 ................................. (96),

there will be no normal displacement, and the displacement will be entirely
tangential, and we shall have

t = a'esm0 .................................. (97).

The whole work done by the superficial forces is

the summation being extended over the surface of the sphere.
The energy of elasticity in the substance of the sphere is

the summation beuig extended to the whole contents of the sphere.

We find, as we ought, that these quantities have the same value, namely

Z7= -fwoi-me(e + 2/) .............................. (98).

We may now suppose that the tangential action on the surface arises from a
layer of particles in contact with it, the particles being acted on by their own
mutual pressure, and acting on the surfaces of the two cells with which they
are in contact.

ON PHYSICAL LINES OF FOECE. 495

We assume the axis of z to be in the direction of maximum variation of
the pressure among the particles, and we have to determine the relation
between an electromotive force R acting on the particles in that direction, and
the electric displacement h which accompanies it.

PKOP. XIII. — To find the relation between electromotive force and electric
displacement when a uniform electromotive force R acts parallel to the axis of z.

Take any element 8S of the surface, covered with a stratum whose density
is p, and having its normal inclined 6 to the axis of z ; then the tangential

force upon it will be

PR8Ssmd=2T8S .............................. (99),

T being, as before, the tangential force on each side of 'the surface. Putting
p = -- as in equation (34)*, we find

'2.TT

R= -2irma(e+2f) ........................... (100).

The displacement of electricity due to the distortion of the sphere is

ZSS^pt sin 0 taken over the whole surface ............ (101);

and if h is the electric displacement per unit of volume, we shall have

f7ra% = faje ............................... (102),

or h= — ae ................................. (103);

27T

so that # = 47rjm— /t ........................ (104),

6

or we may write R = — ±irE2h .............................. (105),

gi ^f
provided we assume E*= — -rnn - -3L ......................... (106).

6

Finding e and / from (87) and (90), we get

1—.....1 ...................... (107).

The ratio of m to p. varies in different substances ; but in a medium whose
elasticity depends entirely upon forces acting between pairs of particles, this
ratio is that of G to 5, and in this case

JS'-vm (108).

* Phil. J%. April, 1861 [p. 471 of this vol.].

496 ON PHYSICAL LINES OF FORCE.

When the resistance to compression is infinitely greater than the resistance to
distortion, as in a liquid rendered slightly elastic by gum or jelly,

Et = Zirm ................................... (109).

The value of E1 must lie between these limits. It is probable that the substance
of our cells is of the former kind, and that we must use the first value of E1,
which is that belonging to a hypothetically "perfect" solid*, in which

5m = 6/x. ................................. (110),

so that we must use equation (108).

PROP. XIV. — To correct the equations (9)f of electric currents for the effect
due to the elasticity of the medium.

We have seen that electromotive force and electric displacement are
connected by equation (105). Differentiating this equation with respect to t, we
find

dR n, dh

shewing that when the electromotive force varies, the electric displacement also
varies. But a variation of displacement is equivalent to a current, and this
current must be taken into account in equations (9) and added to r. The three
equations then become

(112),

r =

(1

dz

1 dP\

~&~dt)

(da.

-*Y

1' ^^\

\<*y

dx

& dt)

'dfi

_da

I dR\

dx

dy

~Wdt)

where p, q, r are the electric currents in the directions of x, y, and z ; a, y8, y
are the components of magnetic intensity; and P, Q, R are the electromotive
forces. Now if e be the quantity of free electricity in unit of volume, then the
equation of continuity will be

* See Rankine "On Elasticity," Camb. and Dub. Math. Journ. 1851.
t I'hil Mag. March, 1861 [p. 462 of this vol.].

ON PHYSICAL LINES OF FORCE. 497

Differentiating (112) with respect to x, y, and z respectively, and substituting,

we find

de 1 d IdP , dQ , dR\

dt kirE* dt \dx dy dz
1 IdP , dQ ,

the constant being omitted, because e = Q when there are no electromotive forces.

PROP. XV. — To find the force acting between two electrified bodies.
The energy in the medium arising from the electric displacements is

U=-^(Pf+Qg + Rh}W. .................... (116),

/
where P, Q, R are the forces, and /, g, h the displacements. Now when there

is no motion of the bodies or alteration of forces, it appears from equations (77)*
that

*--£.<?--£.*--£ .................. <»»»•

and we know by (105) that

(119);

whence *-» + + •* ................... (120)'

Integrating by parts throughout all space, and remembering that "9 vanishes at
an infinite distance,

(121);

or by (115), U=&(Ve)8V .............................. (122).

Now let there be two electrified bodies, and let e1 be the distribution of
electricity in the first, and ¥j the electric tension due to it, and let

1 1*9 d^ d^\
+ 3^ ~ •

Let ei be the distribution of electricity in the second body, and Wa the
tension due to it ; then the whole tension at any point will be % + ¥„ and
the expansion for U will become

U=& (*& + *& + *& + *&) 8V. .................... (124).

* Phil. Mag. May, 1861 [p. 482 of this vol.].
VOL. I. 63

498 ON PHYSICAL LINES OF FORCE.

Let the body whose electricity is e, be moved in any way, the electricity
moving along with the body, then since the distribution of tension ¥, moves
with the body, the value of ¥,e, remains the same.

also remains the same ; and Green has shewn (Essay on Electricity,
p. 10) that ¥,<;, = ^Ai 8° that the work done by moving the body against
electric forces

W =817=82(^)87 ........................ (125).

And if el is confined to a small body,

or Fdr = et dr ............................ (126),

where F is the resistance and dr the motion.

If the body e, be small, then if r is the distance from e,, equation (123)
gives

¥ = J? — -
* r '

whence F=-E>e?' ............................... (127);

or the force is a repulsion varying inversely as the square of the distance.

Now let 17, and 17, be the same quantities of electricity measured stati-
cally, then we know by definition of electrical quantity

and this will be satisfied provided

t)1 = Eel and t)t = Ee, .......................... (129);

so that the quantity E previously determined in Prop. XIII. is the number by
which the electrodynamic measure of any quantity of electricity must be
multiplied to obtain its electrostatic measure.

That electric current which, circulating round a ring whose area is unity,
produces the same effect on a distant magnet as a magnet would produce
whose strength is unity and length unity placed perpendicularly to the plane
of the ring, is a unit current ; and E units of electricity, measured statically,

ON PHYSICAL LINES OF FORCE. 499

traverse the section of this current in one second, — these units being such that
any two of them, placed at unit of distance, repel each other with unit of force.

We may suppose either that E units of positive electricity move in the
positive direction through the wire, or that E units of negative electricity move
in the negative direction, or, thirdly, that \E units of positive electricity move
in the positive direction, while \E units of negative electricity move in the
negative direction at the same time.

The last is the supposition on which MM. "Weber and Kohlrausch* proceed,
who have found

}£» 155,870,000,000 (130),

/

the unit of length being the millimetre, and that of time being one second,
whence

j£ = 310,740,000,000 (131).

PROP. XVI. — To find the rate of propagation of transverse vibrations
through the elastic medium of which the cells are composed, on the suppo-
sition that its elasticity is due entirely to forces acting between pairs of particles.

By the ordinary method of investigation we know that

where m is the coefficient of transverse elasticity, and p is the density. By
referring to the equations of Part I., it will be seen that if p is the density
of the matter of the vortices, and \L is the " coefficient of magnetic induction,"

\L = TTP (133);

whence mn= Vp (134);

and by (108), E= Vjji, (135).

In air or vacuum ^=1, and therefore
V=E

= 310,740,000,000 millimetres per second

= 193,088 miles per second

* AbhamHungen der Ki/nig. Sachsischen Geselkcliaft, Vol. m. (1857), p. 260.

(136).

63—2

500 ON PHYSICAL LINES OF FORCE.

The velocity of light in air, as determined by M. Fizeau*, is 70,843 leagues
per second (25 leagues to a degree) which gives

7=314,858,000,000 millimetres
= 195,647 miles per second ...................... (137)-

The velocity of transverse undulations in our hypothetical medium, calculated
from the electro-magnetic experiments of MM. Kohlrausch and Weber, agrees so
exactly with the velocity of light calculated from the optical experiments of
M. Fizeau, that we can scarcely avoid the inference that light consists in ll<-
transverse undulations of the same medium which is the cause of electric and
magnetic phenomena.

PROP. XVII. — To find the electric capacity of a Leyden jar composed of
any given dielectric placed between two conducting surfaces.

Let the electric tensions or potentials of the two surfaces be ¥, and ¥,.
Let <S* be the area of each surface, and 6 the distance between them, and let
e and — e be the quantities of electricity on each surface ; then the capacity

Within the dielectric we have the variation of ¥ perpendicular to the surface

e

Beyond either surface this variation is zero.

Hence by (115) applied at the surface, the electricity on unit of area is

(139);

and we deduce the whole capacity of the apparatus,

8

so

that the quantity of electricity required to bring the one surface to a

* Comptes Rendus, Vol. MIX. (1849), p. 90. In Galbraith and Haughton's Manual of Astronomy,
M. Fizeau's result is stated at 169,944 geographical miles of 1000 fathoms, which gives 193,118
wtatute miles; the value deduced from aberration is 192,000 miles.

ON PHYSICAL LINES OF FORCE. 501

given tension varies directly as the surface, inversely as the thickness, and
inversely as the square of E.

Now the coefficient of induction of dielectrics is deduced from the capacity
of induction-apparatus formed of them ; so that if D is that coefficient, D varies
inversely as E1, and is unity for air. Hence

where F and F, are the velocities of light in air and in the medium. Now

y
if i is the index of refraction, -pr = i, and

1

D=T. •'• (i42);

so that the inductive power of a dielectric varies directly as the square of the
index of refraction, and inversely as the magnetic inductive power.

In dense media, however, the optical, electric, and magnetic phenomena
may be modified in different degrees by the particles of gross matter; and their
mode of arrangement may influence these phenomena differently in different
directions. The axes of optical, electric, and magnetic properties will probably
coincide ; but on account of the unknown and probably complicated nature of
the reactions of the heavy particles on the setherial medium, it may be im-
possible to discover any general numerical relations between the optical, electric,
and magnetic ratios of these axes.

It seems probable, however, that the value of E, for any given axis,
depends upon the velocity of light whose vibrations are parallel to that axis,
or whose plane of polarization is perpendicular to that axis.

In a uniaxal crystal, the axial value of E will depend on the velocity of
the extraordinary ray, and the equatorial value will depend on that of the
ordinary ray.

In " positive " crystals, the axial value of E will be the least and in
negative the greatest.

The value of Dt, which varies inversely as E*, will, cceteris paribus, be greatest
for the axial direction in positive crystals, and for the equatorial direction in
negative crystals, such as Iceland spar. If a spherical portion of a crystal,
radius =a, be suspended in a field of electric force which would act on unit of

502 ON PHYSICAL LINES OF FORCE.

electricity with force =1, and if A and A be the coefficients of dielectric
induction along the two axes in the plane of rotation, then if 0 be the incli-
nation of the axis to the electric force, the moment tending to turn the sphere
will be

I (g^f^+T) 7Vsin2* (143)>

and the axis of greatest dielectric induction (D,) will tend to become parallel to
the lines of electric force.

PART IV.

THE THEORY OF MOLECULAR VORTICES APPLIED TO THE ACTION OF MAGNETISM

ON POLARIZED LIGHT.

The connexion between the distribution of lines of magnetic force and that
of electric currents may be completely expressed by saying that the work done
on a unit of imaginary magnetic matter, when carried round any closed curve,
is proportional to the quantity of electricity which passes through the closed
curve. The mathematical form of this law may be expressed as in equations (9)*,
which I here repeat, where a, /8, y are the rectangular components of magnetic
intensity, and p, q, r are the rectangular components of steady electric currents,

(9).

-
47T \dy dz

J_ /da _ dy\
^r (dz dx)

47T \dx dy/

The same mathematical connexion is found between other sets of phenomena
in physical science.

(1) If a, ft, y represent displacements, velocities, or forces, then p, q, r
will be rotatory displacements, velocities of rotation, or moments of couples pro-
ducing rotation, in the elementary portions of the mass.

* Phil. Mag. March, 1861 [p. 462 of this voLJ

ON PHYSICAL LINES OF FORCE. 503

(2) If a, ft, y represent rotatory displacements in a uniform and continuous
substance, then p, q, r represent the relative linear displacement of a particle
with respect to those in its immediate neighbourhood. See a paper by Prof. W.
Thomson " On a Mechanical Representation of Electric, Magnetic, and Galvanic
Forces," Camb. and Dublin Math. Journal, Jan. 1847.

(3) If a, fi, y represent the rotatory velocities of vortices whose centres
are fixed, then p, q, r represent the velocities with which loose particles placed
between them would be carried along. See the second part of this paper (Phil.
Mag. April, 1861) [p. 469].

It appears from all these instances that the connexion between magnetism
and electricity has the same mathematical form as that between certain
pairs of phenomena, of which one has a linear and the other a rotatory
character. Professor Challis* conceives magnetism to consist in currents of a
fluid whose direction corresponds with that of the lines of magnetic force ; and
electric currents, on this theory, are accompanied by, if not dependent on, a
rotatory motion of the fluid about the axis of the current. Professor Helmholtzf
has investigated the motion of an incompressible fluid, and has conceived lines
drawn so as to correspond at every point with the instantaneous axis of
rotation of the fluid there. He has pointed out that the lines of fluid motion
are arranged according to the same laws with respect to the lines of rotation,
as those by which the lines of magnetic force are arranged with respect to
electric currents. On the other hand, in this paper I have regarded magnetism
as a phenomenon of rotation, and electric currents as consisting of the actual
translation of particles, thus assuming the inverse of the relation between the
two sets of phenomena.

Now it seems natural to suppose that all the direct effects of any cause
which is itself of a longitudinal character, must be themselves longitudinal, and
that the direct effects of a rotatory cause must be themselves rotatory. A
motion of translation along an axis cannot produce a rotation about that axis
unless it meets with some special mechanism, like that of a screw, which
connects a motion in a given direction along the axis with a rotation in a given
direction round it; and a motion of rotation, though it may produce tension
along the axis, cannot of itself produce a current in one direction along the axis
rather than the other.

* Phil. Mag. December, 1860, January and February, 1861.
t Crelle, Journal, VoL LV. (1858), p. 25.

504 OX PHYSICAL LINES OF FORCE.

Electric currents are known to produce effects of transference in the direc-
tion of the current. They transfer the electrical state from one body to another,
and .they transfer the elements of electrolytes in opposite directions, but they
do not* cause the plane of polarization of light to rotate when the light tra-
verses the axis of the current.

On the other hand, the magnetic state is not characterized by any strictly
longitudinal phenomenon. The north and south poles differ only in their names,
and these names might be exchanged without altering the statement of any
magnetic phenomenon ; whereas the positive and negative poles of a battery are
completely distinguished by the different elements of water which are evolved
there. The magnetic state, however, is characterized by a well-marked rotatory
phenomenon discovered by Faraday f — the rotation of the plane of polarized light
when transmitted along the lines of magnetic force.

When a transparent diamagnetic substance has a ray of plane-polarized light
passed through it, and if lines of magnetic force are then produced in the
substance by the action of a magnet or of an electric current, the plane of
polarization of the transmitted light is found to be changed, and to be turned
through an angle depending on the intensity of the magnetizing force within
the substance.

The direction of this rotation in diamagnetic substances is the same as that
in which positive electricity must circulate round the substance in order to
produce the actual magnetizing force within it ; or if we suppose the horizontal
part of terrestrial magnetism to be the magnetizing force acting on the sub-
stance, the plane of polarization would be turned in the direction of the earth's
true rotation, that is, from west upwards to east.

In paramagnetic substances, M. VerdetJ has found that the plane of polari-
zation is turned in the opposite direction, that is, in the direction in which
negative electricity would flow if the magnetization were effected by a helix
surrounding the substance.

In both cases the absolute direction of the rotation is the same, whether
the light passes from north to south or from south to north, — a fact which dis-
tinguishes this phenomenon from the rotation produced by quartz, turpentine, &c.,

* Faraday, Experimental Researches, 951—954, and 2216—2220.

t Ibid., Series xix.

J Camples Rendus, Vol. XLIII. p. 529; Vol. iuv. p. 1209.

ON PHYSICAL LINES OP FORCE. 505

in which the absolute direction of rotation is reversed when that of the light
is reversed. The rotation in the latter case, whether related to an axis, as in
quartz, or not so related, as in fluids, indicates a relation between the direction
of the ray and the direction of rotation, which is similar in its formal expression
to that between the longitudinal and rotatory motions of a right-handed or a
left-handed screw ; and it indicates some property of the substance the mathe-
matical form of which exhibits right-handed or left-handed relations, such as are
known to appear in the external forms of crystals having these properties. In
the magnetic rotation no such relation appears, but the direction of rotation is
directly connected with that of the magnetic lines, in a way which seems to
indicate that magnetism is really a phenomenon of rotation.

The transference of electrolytes in fixed directions by the electric current,
and the rotation of polarized light in fixed directions by- magnetic force, are
the facts the consideration of which has induced me to regard magnetism as a
phenomenon of rotation, and electric currents as phenomena of translation, instead
of following out the analogy pointed out by Helmholtz, or adopting the theory
propounded by Professor Challis.

The theory that electric currents are linear, and magnetic forces rotatory
phenomena, agrees so far with that of Ampere and Weber ; and the hypothesis
that the magnetic rotations exist wherever magnetic force extends, that the
centrifugal force of these rotations accounts for magnetic attractions, and that
the inertia of the vortices accounts for induced currents, is supported by the
opinion of Professor W. Thomson*. In fact the whole theory of molecular vor-
tices developed in this paper has been suggested to me by observing the
direction in which those investigators who study .the action of media are looking
for the explanation of electro-magnetic phenomena.

Professor Thomson has pointed out that the cause of the magnetic action
on light must be a real rotation going on in the magnetic field. A right-handed
circularly polarized ray of light is found to travel with a different velocity
according as it passes from north to south, or from south to north, along a
line of magnetic force. Now, whatever theory we adopt about the direction of
vibrations in plane-polarized light, the geometrical arrangement of the parts of
the medium during the passage of a right-handed circularly polarized ray is
exactly the same whether the ray is moving north or south. The only difference

* See NichoPs Cyclopaedia, art "Magnetism, Dynamical Relations of," edition 1860; Proceedings
of Royal Society, June 1856 and June 1861 ; and Phil. Mag. 1857.

VOL. I. 64

506 ON PHYSICAL LINES OF FORCE.

is, that the particles describe their circles in opposite directions. Since, therefore,
the configuration is the same in the two cases, the forces acting between par-
ticles must be the same in both, and the motions due to these forces must be
equal in velocity if the medium was originally at rest ; but if the medium be
in a state of rotation, either as a whole or in molecular vortices, the circular
vibrations of light may differ in velocity according as their direction is similar
or contrary to that of the vortices.

We have now to investigate whether the hypothesis developed in this
paper — that magnetic force is due to the centrifugal force of small vortices, and
that these vortices consist of the same matter the vibrations of which constitute
light — leads to any conclusions as to the effect of magnetism on polarized light.
We suppose transverse vibrations to be transmitted through a magnetized
medium. How will the propagation of these vibrations be affected by the
circumstance that portions of that medium are in a state of rotation ?

In the following investigation, I have found that the only effect which the
rotation of the vortices will have on the light will be to make the plane of
polarization rotate in the same direction as the vortices, through an angle
proportional —

(A) to the thickness of the substance,

(B) to the resolved part of the magnetic force parallel to the ray,

(C) to the index of refraction of the ray,

(D) inversely to the square of the wave-length in air,

(E) to the mean radius of the vortices,

(F) to the capacity for magnetic induction.

A and B have been fully investigated by M. Verdet*, who has shewn that
the rotation is strictly proportional to the thickness and to the magnetizing
force, and that, when the ray is inclined to the magnetizing force, the rotation
is as the cosine of that inclination. D has been supposed to give the true
relation between the rotation of different rays ; but it is probable that C must
be taken into account in an accurate statement of the phenomena. The rotation
varies, not exactly inversely as the square of the wave length, but a little faster;
so that for the highly refrangible rays the rotation is greater than that given
by this law, but more nearly as the index of refraction divided by the square
of the wave-length.

* Annales de Chimie et de Physique, s6r. 3, Vol. ILL p. 370; Vol. XLIII. p. 37.

ON PHYSICAL LINES OF FORCE. 507

The relation (E) between the amount of rotation and the size of the
vortices shews that different substances may differ in rotating power inde-
pendently of any observable difference in other respects. We know nothing
of the absolute size of the vortices ; and on our hypothesis the optical phenomena
are probably the only data for determining their relative size in different sub-
stances.

On our theory, the direction of the rotation of the plane of polarization
depends on that of the mean moment of momenta, or angular momentum, of the
molecular vortices ; and since M. Verdet has discovered that magnetic substances
have an effect on light opposite to that of diamagnetic substances, it follows that
the molecular rotation must be opposite in the two classes of substances.

We can no longer, therefore, consider diamagnetic /bodies as being those
whose coefficient of magnetic induction is less than that of space empty of
gross matter. We must admit the diamagnetic state to be the opposite of the
paramagnetic ; and that the vortices, or at least the influential majority of them,
in diamagnetic substances, revolve in the direction in which positive electricity
revolves in the magnetizing bobbin, while in paramagnetic substances they
revolve in the opposite direction.

This result agrees so far with that part of the theory of M. Weber*
which refers to the paramagnetic and diamagnetic conditions. M. Weber sup-
poses the electricity in paramagnetic bodies to revolve the same way as the
surrounding helix, while in diamagnetic bodies it revolves the opposite way.
Now if we regard negative or resinous electricity as a substance the absence
of which constitutes positive or vitreous electricity, the results will be those
actually observed. This will be true independently of any other hypothesis
than that of M. Weber about magnetism and diamagnetism, and does not
require us to admit either M. Weber's theory of the mutual action of electric
particles in motion, or our theory of cells and cell-walls.

I am inclined to believe that iron differs from other substances in the
manner of its action as well as in the intensity of its magnetism ; and I think
its behaviour may be explained on our hypothesis of molecular vortices, by
supposing that the particles of the iron itself are set in rotation by the tan-
gential action of the vortices, in an opposite direction to their own. These
large heavy particles would thus be revolving exactly as we have supposed the

* Taylor's Scientific Memoirs, Vol. v. p. 477.

64—2

508 ON PHYSICAL LINES OF FORCE.

infinitely small particles constituting electricity to revolve, but without being
free like them to change their place and form currents.

The whole energy of rotation of the magnetized field would thus be greatly
increased, as we know it to be ; but the angular momentum of the iron
particles would be opposite to that of the eethereal cells and immensely greater,
so that the total angular momentum of the substance will be in the direction
of rotation of the iron, or the reverse of that of the vortices. Since, however,
the angular momentum depends on the absolute size of the revolving portions
of the substance, it may depend on the state of aggregation or chemical
arrangement of the elements, as well as on the ultimate nature of the com-
ponents of the substance. Other phenomena in nature seem to lead to the
conclusion that all substances are made up of a number of parts, finite in size,
the particles composing these parts being themselves capable of internal motion.

PROP. XVIII. — To find the angular momentum of a vortex.

The angular momentum of any material system about an axis is the sum
of -the products of the mass, dm, of each particle multiplied by twice the area
it describes about that axis in unit of time ; or if A is the angular momentum
about the axis of x,

As we do not know the distribution of density within the vortex, we shall
determine the relation between the angular momentum and the energy of the
vortex which was found in Prop. VI.

Since the time of revolution is the same throughout the vortex, the mean

angular velocity at will be uniform and =-, where a is the velocity at the

circumference, and r the radius. Then

A =
and the energy E

= -?-fta'Fby Prop. VI.*

OTT

whence A = — p.raV ..................................... (144)

• PhiL Mag. April 1861 [p. 472 of this vol.].

ON PHYSICAL LINES OF FORCE. 509

for the axis of x, with similar expressions for the other axes, V being the
volume, and r the radius of the vortex.

PROP. XIX. — To determine the conditions of undulatory motion in a medium
containing vortices, the vibrations being perpendicular to the direction of pro-
pagation.

Let the waves be plane-waves propagated in the direction of z, and let
the axis of x and y be taken hi the directions of greatest and least elasticity
in the plane xy. Let x and y represent the displacement parallel to these axes,
which will be the same throughout the same wave-surface, and therefore we
shall have x and y functions of z and t only.

Let X be the tangential stress on unit of area parallel to xy, tending to
move the part next the origin hi the direction of x.

Let Y be the corresponding tangential stress in the
direction of y.

Let &, and &, be the coefficients of elasticity with respect
to these two kinds of tangential stress ; then, if the medium
is at rest,

X=kldz' Y=k*dz-

Now let us suppose vortices in the medium whose velocities are represented
as usual by the symbols a, /8, y, and let us suppose that the value of a is

increasing at the rate -j- , on account of the action of the tangential stresses

alone, there being no electromotive force in the field. The angular momentum
in the stratum whose area is unity, and thickness dz, is therefore increasing

at the rate — pr -j- dz; and if the part of the force Y which produces this effect

is Y, then the moment of Y is — Y'dz, so that Y = — — u.r -=- .

4?r at

The complete value of Y when the vortices are in a state of varied motion is

,, , dy 1 da

Y — L J_ . _ iii» -

JL — A/- j Utl 7

dz 47r at

„. ., v , dx I dp

Similarly, X = *, -j- + — fir -j-

510

ON PHYSICAL LINES OF FORCE.

The whole force acting upon a stratum whose thickness is dz and area
unity, is r- dz in the direction of x, and —r- dz in direction of y. The mass

of the stratum is pdz, so that we have as the equations of motion,

<Px_dX_, d*x d 1 dB'

^ dt* dz dz* dz 4;r dt i-\Ar\

d*y dY , d*y d 1 da

Now the changes of velocity -3- and -y- are produced by the motion of

Gtt ' ' '

the medium containing the vortices, which distorts and twists every element
of its mass; so that we must refer to Prop. X.* to determine these quantities
in terms of the motion. We find there at equation (68),

d d , d

da = a -j— ox-}- B -=— ox + y -5- ox (68).

dx dy dz

Since 8x and 8y are functions of z and t only, we may write this equation

da _ d*x

f~7^ (147),

and in like manner, -° = y-r-

dt ' dzdt J

tut*

so that if we now put k, = a?p. k. = b*p, and - — y = c", we may write the

47T p '

equations of motion

dt*- l dz*~ S dz'dt (14g)

~dt*~ ~* ~dz*~ ' dz*dt .

These equations may be satisfied by the values

x = A cos (nt — mz + a)l
y = B sin (nt — mz + a)/'
provided ,, i »\ ,< _ > »w>

. \ (150).

and (n — mb) a*-~*

* Phil. Mag. May 1861 [p. 481 of this vol.].

ON PHYSICAL LINES OF FORCE.

Multiplying the last two equations together, we find

511

(151)

an equation quadratic with respect to ms, the solution of which is

9/W3

m' =

<

These values of m' being put in the equations (150) will each give a ratio
of A and B,

A _ a* — 62 T v(a2 — ft2)2 + 4w2c4
]B= 2nc*

which being substituted in equations (149), will satisfy the original equations
(148). The most general undulation of such a medium is therefore compounded
of two elliptic undulations of different eccentricities travelling with different
velocities and rotating in opposite directions. The results may be more easily
explained in the case in which a = 6 ; then

and A= ^-B (153).

Let us suppose that the value of A is unity for both vibrations, then we
shall have

/ nz \ I nz \ '

x = cos nt — , I + cos I nt — . }

\ Va2 — ncv \ v« +nc2/

I nz \ , . / . nz N

y = — sin ( nt — -,-=== j + sin I nt — -j=f=

The first terms of x and y represent a circular vibration in the negative
direction, and the second term a circular vibration in the positive direction,
the positive having the greatest velocity of propagation. Combining the terms,
we may write

x = 2 cos (nt —pz) cos qz\

y = 2 cos (nt —pz) sin qz]

n n

.(156).

n n

and

512 ON PHYSICAL LINES OF FORCE.

These are the equations of an undulation consisting of a plane vibration

O_ n—

whose periodic time is : - , and wave-length : - = X, propagated in the direction

of z with a velocity - = v, while the plane of the vibration revolves about the

2?r
axis of z in the positive direction so as to complete a revolution when z = — .

Now let us suppose c1 small, then we may write

ra nV

1 T

and remembering that c1 = - — py, we find

Here r is the radius of the vortices, an unknown quantity, p is the density
of the luminiferous medium in the body, which is also unknown ; but if we
adopt the theory of Fresnel, and make s the density in space devoid of gross

matter, then

/> = «* .................................... (159),

where i is the index of refraction.

On the theory of MacCullagh and Neumann,

p = s .................................... (160)

in all bodies.

\L is the coefficient of magnetic induction, which is unity in empty space
or in air.

y is the velocity of the vortices at their circumference estimated in the
ordinary units. Its value is unknown, but it is proportional to the intensity of
the magnetic force.

Let Z be the magnetic intensity of the field, measured as in the case of
terrestrial magnetism, then the intrinsic energy in air per unit of volume is

where s is the density of the magnetic medium in air, which we have reason
to believe the same as that of the luminiferous medium. We therefore put

y=j= z

•Jirs

ON PHYSICAL LINES OF FORCE.

513

X is the wave-length of the undulation in the substance. Now if A be the
wave-length for the same ray in air, and i the index of refraction of that ray in
the body,

* = J (162).

Also v, the velocity of light in the substance, is related to V, the velocity of
light in air, by the equation

o = J (163).

Hence if z be the thickness of the substance through which the ray passes, the
angle through which the plane of polarization will be turned will be in degrees,

a 180
0 = qz

TT

(164);

or, by what we have now calculated,

(165).

In this expression all the quantities are known by experiment except r, the
radius of the vortices in the body, and s, the density of the luminiferous
medium in air.

The experiments of M. Verdet* supply all that is wanted except the deter-
mination of Z in absolute measure ; and this would also be known for all his
experiments, if the value of the galvanometer deflection for a semi-rotation of
the testing bobbin in a known magnetic field, such as that due to terrestrial
magnetism at Paris, were once for all determined.

* Annales de Chimie et de P/tyrique, s6r. 3, Vol. XLI. p. 370.

VOL. I.

65

[From the London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science.

Vol. xxvii. Fourth Series.]

XXIV. On Reciprocal Figures and Diagrams of Forces.

RECIPROCAL figures are such that the properties of the first relative to the
second are the same as those of the second relative to the first. Thus inverse
figurea and polar reciprocals are instances of two different kinds of reciprocity.

The kind of reciprocity which we have here to do with has reference to
figures consisting of straight lines joining a system of points, and forming
closed rectilinear figures ; and it consists in the directions of all lines in the
one figure having a constant relation to those of the lines in the other figure
which correspond to them.

In plane figures, corresponding lines may be either parallel, perpendicular,
or at any constant angle. Lines meeting in a point in one figure form a
closed polygon in the other.

In figures in space, the lines in one figure are perpendicular to planes in
the other, and the planes corresponding to lines which meet in a point form
a closed polyhedron.

The conditions of reciprocity may be considered from a purely geometrical
point of view; but their chief importance arises from the fact that either of
the figures being considered as a system of points acted on by forces along
the lines of connexion, the other figure is a diagram of forces, in which these
forces are represented in plane figures by lines, and in solid figures by the
areas of planes.

The properties of the "triangle" and "polygon" of forces have been long
known, and the "diagram" of forces has been used in the case of the funicular
polygon ; but I am not aware of any more general statement of the method

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES. 515

of drawing diagrams of forces before Professor Eankine applied it to frames,
roofs, &c. in his Applied Mechanics, p. 137, &c. The "polyhedron of forces,"
or the equilibrium of forces perpendicular and proportional to the areas of the
faces of a polyhedron, has, I believe, been enunciated independently at various
times; but the application to a "frame" is given by Professor Rankine in the
Philosophical Magazine, February, 1864.

I propose to treat the question geometrically, as reciprocal figures are
subject to certain conditions besides those belonging to diagrams of forces.

On Reciprocal Plane Figures.

Definition. — Two plane figures are reciprocal when they consist of an equal

number of lines, so that corresponding lines in the two figures are parallel,

and corresponding lines which converge to a point in one figure form a closed
polygon in the other.

Note, — If corresponding lines in the two figures, instead of being parallel
are at right angles or any other angle, they may be made parallel by turning
one of the figures round in its own plane.

Since every polygon in one figure has three or more sides, every point in
the other figure must have three or more lines converging to it ; and since
every line in the one figure has two and only two extremities to which lines
converge, every line in the other figure must belong to two, and only two
closed polygons. The simplest plane figure fulfilling these conditions is that
formed by the six lines which join four points in pairs. The reciprocal figure
consists of six lines parallel respectively to these, the points in the one figure
corresponding to triangles in the other.

•

General Relation between the Numbers of Points, Lines, and Polygons in

Reciprocal Figures.

The effect of drawing a line, one of whose extremities is a point connected
with the system of lines already drawn, is either to introduce one new point
into the system, or to complete one new polygon, or to divide a polygon into
two parts, according as it is drawn to an isolated point, or a point already
connected with the system. Hence the sum of points and polygons in the

65—2

516 ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

system is increased by one for every new line. But the simplest figure consists
of four points, four polygons, and six lines. Hence the sum of the points and
polygons must always exceed the number of lines by two.

Note. — This is the same relation which connects the numbers of summits,
faces, and edges of polyhedra.

Conditions of indeterminateness and impossibility in drawing reciprocal Diagrams.

Taking any line parallel to one of the lines of the figure for a base,
every new point is to be determined by the intersection of two new lines.
Calling s the number of points or summits, e the number of lines or edges,
and f the number of polygons or faces, the assumption of the first line deter-
mines two points, and the remaining s — 2 points are determined by 2 (s — 2)

lines. Hence if

e = 2s-3,

every point may be determined. If e be less, the form of the figure will be
in some respects indeterminate ; and if e be greater, the construction of the
figure will be impossible, unless certain conditions among the directions of the
lines are fulfilled.

These are the conditions of drawing any diagram in which the directions
of the lines are arbitrarily given ; but when one diagram is already drawn in
which e is greater than 2* — 3, the directions of the lines will not be altogether
arbitrary, but will be subject to e — (2s — 3) conditions.

Now if e', s', f be the values of e, s, and f in the reciprocal diagram

e = e', *=/', f=s',
e = s+f-2, e' = s'+f'-2.

Hence if s =f, e = 2i — 2; and there will be one condition connecting the
directions of the lines of the original diagram, and this condition will ensure
the possibility of constructing the reciprocal diagram. If

s >/, e > 2s - 2, and e' < 2s' - 2 ;

so that the construction of the reciprocal diagram will be possible, but inde-
terminate to the extent of s —f variables.

If s<f, the construction of the reciprocal diagram will be impossible unless
(a—/) conditions be fulfilled in the original diagram.

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

517

If any number of the points of the figure are so connected among them-
selves as to form an equal number of closed polygons, the conditions of
constructing the reciprocal figure must be found by considering these points
separately, and then examining their connexion with the rest.

Let us now consider a few cases of reciprocal figures in detail. The
simplest case is that of the figure formed by the six lines connecting four
points in a plane. If we now draw the six lines con-
necting the centres of the four circles which pass through
three out of the four points, we shall have a reciprocal
figure, the corresponding lines hi the two figures being
at right angles.

The reciprocal figure formed in this way is definite
in size and position ; but any figure similar to it and
placed in any position is still reciprocal to the original
figure. If the reciprocal figures are lettered as in fig. 1,
we shall have the relation

Fig. 1.

AP
ap

BQ CR

cr

In figures 2 and II. we have a pair of reciprocal figures in which the
lines are more numerous, but the construction very easy. There are seven
points in each figure corresponding to seven polygons in the other.

Fig. 2.

Fig. II.

The four points of triple concourse of lines ABC, BDE, HIL, LJK
correspond to four triangles, abc, bde, hil, Ijk.

The three points of quadruple concourse ADFH, CEGK, IFGJ correspond
to three quadrilaterals, adfh, cegk, ifgj.

The five triangles ADB, EBC, GJK, IJL, HIF correspond to five points
of triple concourse, adb, ebc, gjk, ijl, hif.

518 ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

The quadrilateral DEGF corresponds to the point of quadruple concourse

The pentagon ACKLH corresponds to the meeting of the five lines acfdh.

In drawing the reciprocal of fig. 2, it is best to begin with a point of triple
concourse. The reciprocal triangle of this point being drawn, determines three
lines of the new figure. If the other extremities of any of the lines meeting
in this point are points of triple concourse, we may in the same way deter-
mine more lines, two at a time. In drawing these lines, we have only to
remember that those lines which in the first figure form a polygon, start from
one point in the reciprocal figure. In this way we may proceed as long as
we can always determine all the lines except two of each successive polygon.

The case represented in figs. 3 and III. is an instance of a pair of reci-
procal figures fulfilling the conditions of possibility and determinateness, but

Fig. III.

Fig. 3.

presenting a slight difficulty in drawing by the foregoing rule. Each figure has
here eight points and eight polygons ; but after we have drawn the lines s,
n, o, k, r, we cannot proceed with the figure simply by drawing the last two
lines of polygons, because the next polygons to be drawn are quadrilaterals, and
we have only one side of each given. The easiest way to proceed is to produce
abed till they form a quadrilateral, then to draw a subsidiary figure similar to
tlmpq, with abed similarly situated, and then to reduce the latter figure to
such a scale and position that a, 6, c, d coincide in both figures.

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

519

In figures 4 and IV. the condition that the number of polygons is equal
to the number of points is not fulfilled. In fig. 4 there are five points and

Fig. IV.

Fig. 4.

six triangles ; in fig, IV. there afe six points, two triangles, and three quadri-
laterals. Hence if fig. 4 is given, fig. IV. is indeterminate to the extent of one
variable, besides the elements of scale and position. In fact when we have drawn
ABC and indicated the directions of P, Q, R, we may fix on any point of P
as one of the angles of X YZ and complete the triangle XYZ. The size of
XYZ is therefore indeterminate. Conversely, if fig. IV. is given, fig. 4 cannot
be constructed unless one condition be fulfilled. That condition is that P, Q,
and R meet in a point. When' this is fulfilled, it follows by geometry that
the points of concourse of A and X, B and Y, and C and Z lie in one straight
line W, which is parallel to w in fig. 4. The condition may also be expressed
by saying that fig. IV. must be a perspective projection of a polyhedron whose
quadrilateral faces are planes. The planes of these faces intersect at the concourse
of P, Q, R, and those of the triangular faces intersect in the line W.

Figs. 5 and V. represent another case of the same kind. In fig. 5 we
have six points and eight triangles ; fig. V. is therefore capable of two degrees
of variability, and is subject to two conditions.

520

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

The conditions are that the four intersections of corresponding sides of
opposite quadrilaterals in fig. V. shall lie in one straight line, parallel to the

Fig. V. Fig. 5.

line joining the opposite points of fig. 5 which correspond to these quadrilaterals.
There are three such lines marked x, y, z, and four points of intersection lie on
each line.

We may express this condition also by saying that fig. V. must be a per-
spective projection of a plane-sided polyhedron, the intersections of opposite
planes being the lines x, y, z.

In fig. 6, let ABCDE be a portion of a polygon bounded by other polygons
of which the edges are PQRST, one or more of these edges meeting each angle
of the polygon.

In fig. VI., let abcde be lines parallel to ABCDE and meeting in a point,
and let these be terminated by the lines pqrst parallel to PQRST, one or
more of these lines completing each sector of fig. VI.

In fig. 6 draw Y through the intersections of AC and PQ, and in fig.
VL draw y through the intersections of a, p and c, q. Then the figures of
six lines ABCPQY and dbcpqy will be reciprocal, and y will be parallel to Y.
Draw X parallel to x, and through the intersections of TX and CE draw Z,
and in fig. VI. draw z through the intersections of ex and et; then CDETXZ

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES. 521

and cdetxz will be reciprocal, and Z will be parallel to z. Then through the
intersections of AE and YZ draw W, and through those of ay and ez draw
w; and since ACEYZW and aceyzw are reciprocal, W will be parallel to w.

Fig. 6.

Fig. VI.

By going round the remaining sides of the polygon ABODE in the same
way, we should find by the intersections of lines another point, the line joining
which with the intersection of AE would be parallel to w, and therefore we
should have three points in one line ; namely, the intersection of Y and Z,
the point determined by a similar process carried on on the other part of the
circumference of the polygon, and the intersection of A and E ; and we should
find similar conditions for every pair of sides of every polygon.

Now the conditions of the figure 6 being a perspective projection of a
plane-sided polyhedron are exactly the same. For A being the intersection of
the faces AP and AB, and C that of BC and QC, the intersection AC will
be a point in the intersection of the faces AP and CQ.

Similarly the intersection PQ will be another point in it, so that Y is the
line of intersection of the faces AP and CQ.

In the same way Z is the intersection of ET and CQ, so that the inter-
section of Y and Z is a point in the intersection of AP and ET.

Another such point can be determined by going round the remaining sides
of the polygon ; and these two points, together with the intersections of the
lines AE, must all be in one straight line, namely, the intersection of the faces
AP and ET.

Hence the conditions of the possibility of reciprocity in plane figures are
the same as those of each figure being the perspective projection of a plaiie-
sided polyhedron. When the number of points is in every part of the figure
equal to or less than the number of polygons, this condition is fulfilled of
itself. When the number of points exceeds the number of polygons, there will

VOL. i. 66

522 ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

be an impossible case, unless certain conditions are fulfilled so that certain sets
of intersections lie in straight lines.

Application to Statics.

The doctrine of reciprocal figures may be treated in a purely geometrical
manner, but it may be much more clearly understood by considering it as a
method of calculating the forces among a system of points in equilibrium ; for,

If forces represented in magnitude by the lines of a figure be made to act
between the extremities of the corresponding lines of the reciprocal figure, then
the points of the reciprocal figure will all be in equilibrium under the action
of these forces.

For the forces which meet in any point are parallel and proportional to
the sides of a polygon in the other figure.

If the points between which the forces are to act are known, the problem
of determining the relations among the magnitudes of the forces so as to produce
equilibrium will be indeterminate, determinate, or impossible, according as the
construction of the reciprocal figure is so.

Reciprocal figures are mechanically reciprocal ; that is, either may be taken
as representing a system of points, and the other as representing the magnitudes
of the forces acting between them.

In figures like 1, 2 and II., 3 and III., in which the equation

e = 2s-2

is true, the forces are determinate in their ratios ; so that one being given,
the rest may be found.

When e>2« — 2, as in figs. 4 and 5, the forces are indeterminate, so that
more than one must be known to determine the rest, or else certain relations
among them must be given, such as those arising from the elasticity of the
parts of a frame.

When e<2s — 2, the determination of the forces is impossible except under
certain conditions. Unless these be fulfilled, as in figs. IV. and V., no forces
along the lines of the figure can keep its points in equilibrium, and the figure,
considered as a frame, may be said to be loose.

When the conditions are fulfilled, the pieces of the frame can support forces,
but in such a way that a small disfigurement of the frame may produce in-

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

523

finitely great forces in some of the pieces, or may throw the frame into a loose
condition at once.

The conditions, however, of the possibility of determining the ratios of the
forces in a frame are not coextensive with those of finding a figure perfectly
reciprocal to the frame. The condition of determinate forces is

e = 2s-2;

the condition of reciprocal figures is that every line belongs to two polygons
only, and

In fig. 7 we have six points connected by ten lines in such a way that
the forces are all determinate ; but since the line L is a ,side of three triangles,
we cannot draw a reciprocal figure, for we should have to draw a straight line
I with three ends.

If we attempt to draw the reciprocal figure as in fig. VIL, we shall find
that, in order to represent the reciprocals of all the lines of fig. 7 and fix
their relations, we must repeat two of them, as h and e by h' and e, so as
to form a parallelogram. Fig. VII. is then a complete representation of the rela-
tions of the force which would produce equilibrium in fig. 7 ; but it is redundant
by the repetition of h and e, and the two figures are not reciprocal.

Fig. VII.

Fig. 7.

On Reciprocal Figures in three dimensions.

Definition. — Figures in three dimensions are reciprocal when they can be so
placed that every line in the one figure is perpendicular to a plane face of the
other, and every point of concourse of lines in the one figure is represented by
a closed polyhedron with plane faces.

66—2

524 ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES.

The simplest case is that of five points in space with their ten connecting
lines, forming ten triangular faces enclosing five tetrahedrons. By joining the five
points which are the centres of the spheres circumscribing these five tetrahedrons,
we have a reciprocal figure of the kind described by Professor Rankine in the
J'ln'losophical Mayazine, February 1864; and forces proportional to the areas of
the triangles of one figure, if applied along the corresponding lines of connexion
of the other figure, will keep its points in equilibrium.

In order to have perfect reciprocity between two figures, each figure must
be made up of a number of closed polyhedra having plane faces of separation,
and such that each face belongs to two and only two polyhedra, corresponding
to the extremities of the reciprocal line in the other figure. Every line in the
figure is the intersection of three or more plane faces, because the plane face in
the reciprocal figure is bounded by three or more straight lines.

Let s be the number of points or summits, e the number of lines or edges,

/ the number of faces, and c the number of polyhedra or cells. Then if about

one of the summits in which polyhedra meet, and a- edges and 17 faces, we

describe a polyhedral cell, it will have <f> faces and a- summits and rj edges,

and we shall have

s, the number of summits, will be decreased by one and increased by a- ;
c, the number of cells, will be increased by one ;
/, the number of faces, will be increased by $ ;
e, the number of edges, will be increased by 77 ;

so that e + c — (s+f) will be increased by 17+ 1 — (<r + <f> — 1), which is zero, or
this quantity is constant. Now in the figure of five points already discussed,
e=10, c = 5, *=5, f=lO; so that generally

e + c = s+f,
in figures made up of cells in the way described.

The condition of a reciprocal figure being indeterminate, determinate, or im-
possible except in particular cases, is

e ^ 3s- 5.

<

This condition is sufficient to determine the possibility of finding a system of
forces along the edges which will keep the summits in equilibrium ; but it is

ON RECIPROCAL FIGURES AND DIAGRAMS OF FORCES. 525

manifest that the mechanical problem may be solved, though the reciprocal figure
cannot be constructed owing to the condition of all the sides of a face lying
in a plane not being fulfilled, or owing to a face belonging to more than two
cells. Hence the mechanical interest of reciprocal figures in space rapidly
diminishes with their complexity.

Diagrams of forces in which the forces are represented by lines may be
always constructed in space as well as in a plane, but in general some of the
lines must be repeated.

Thus in the figure of five points, each point is the meeting place of four
lines. The forces in these lines may be represented by five gauche quadrilaterals
(that is, quadrilaterals not in one plane) ; and one of these being chosen, the
other four may be applied to its sides and to each other so as to form five
sides of a gauche hexahedron. The sixth side, that opposite the original quad-
rilateral, will be a parallelogram, the opposite sides of which are repetitions of
the same line.

We have thus a complete but redundant diagram of forces consisting of
eight points joined by twelve lines, two pairs of the lines being repetitions.
This is a more convenient though less elegant construction of a diagram of
forces, and it never becomes geometrically impossible as long as the problem is
mechanically possible, however complicated the original figure may be.

[From the Royal Society Transactions, Vol. CLV.]

XXV. A Dynamical Theory of the Electromagnetic Field.

(Received October 27,— Read December 8, 1864.)

PAET I.

INTRODUCTORY.

(1) THE most obvious mechanical phenomenon in electrical and magnetical
experiments is the mutual action by which bodies in certain states set each
other in motion while still at a sensible distance from each other. The first
step, therefore, in reducing these phenomena into scientific form, is to ascertain
the magnitude and direction of the force acting between the bodies, and when
it is found that this force depends in a certain way upon the relative position
of the bodies and on their electric or magnetic condition, it seems at first sight
natural to explain the facts by assuming the existence of something either at
rest or in motion in each body, constituting its electric or magnetic state, and
capable of acting at a distance according to mathematical laws.

In this way mathematical theories of statical electricity, of magnetism, of
the mechanical action between conductors carrying currents, and of the induction
of currents have been formed. In these theories the force acting between the
two bodies is treated with reference only to the condition of the bodies and
their relative position, and without any express consideration of the surrounding
medium.

These theories assume, more or less explicitly, the existence of substances
the particles of which have the property of acting on one another at a distance
by attraction or repulsion. The most complete development of a theory of this

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 527

kind is that of M. W. Weber*, who has made the same theory include elec-
trostatic and electromagnetic phenomena.

In doing so, however, he has found it necessary to assume that the force
between two electric particles depends on their relative velocity, as well as on
their distance.

This theory, as developed by MM. W. Weber and C. Neumann t, is ex-
ceedingly ingenious, and wonderfully comprehensive in its application to the
phenomena of statical electricity, electromagnetic attractions, induction of currents
and diamagnetic phenomena ; and it comes to us with the more authority, as
it has served to guide the speculations of one who has made so great an
advance in the practical part of electric science, both by introducing a consistent
system of units in electrical measurement, and by actually determining electrical
quantities with an accuracy hitherto unknown.

(2) The mechanical difficulties, however, which are involved in the assump-
tion of particles acting at a distance with forces which depend on their velocities
are such as to prevent me from considering this theory as an ultimate one,
though it may have been, and may yet be useful in leading to the coordina-
tion of phenomena.

I have therefore preferred to seek an explanation of the fact in another
direction, by supposing them to be produced by actions which go on in the
surrounding medium as well as in the excited bodies, and endeavouring to
explain the action between distant bodies without assuming the existence of
forces capable of acting directly at sensible distances.

(3) The theory I propose may therefore be called a theory of the Electro-
magnetic Field, because it has to do with the space in the neighbourhood of
the electric or magnetic bodies, and it may be called a Dynamical Theory,
because it assumes that in that space there is matter in motion, by which
the observed electromagnetic phenomena are produced.

(4) The electromagnetic field is that part of space which contains and
surrounds bodies in electric or magnetic conditions.

* " Electrodynamische Maassbestimmnngen." Leipzic Trans. Vol. I. 1849, and Taylor's Scientific
Memoirs, Vol. v. art. xiv.

t Explicare tenlatur quomodo fiat ut lucis planum polarizationis per vires electricas vel magneticas
declinetur. — Halis Saxonum, 1858.

528 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

It may be filled with any kind of matter, or we may endeavour to render
it empty of all gross matter, as in the case of Geissler's tubes and other so-
called vacua.

There is always, however, enough of matter left to receive and transmit
the undulations of light and heat, and it is because the transmission of these
radiations is not greatly altered when transparent bodies of measurable density
are substituted for the so-called vacuum, that we are obliged to admit that the
undulations are those of an sethereal substance, and not of the gross matter,
the presence of which merely modifies in some way the motion of the aether.

We have therefore some reason to believe, from the phenomena of light
and heat, that there is an {ethereal medium filling space and permeating bodies,
capable of being set in motion and of transmitting that motion from one part
to another, and of communicating that motion to gross matter so as to heat
it and affect it in various ways.

(5) Now the energy communicated to the body in heating it must have
formerly existed in the moving medium, for the undulations had left the source
of heat some time before they reached the body, and during that time the
energy must have been half in the form of motion of the medium and half in
the form of elastic resilience. From these considerations Professor W. Thomson
has argued*, that the medium must have a density capable of comparison with
that of gross matter, and has even assigned an inferior limit to that density.

(6) We may therefore receive, as a datum derived from a branch of science
independent of that with which we have to deal, the existence of a pervading
medium, of small but real density, capable of being set in motion, and of trans-
mitting motion from one part to another with great, but not infinite, velocity.

Hence the parts of this medium must be so connected that the motion of
one part depends in some way on the motion of the rest; and at the same
time these connexions must be capable of a certain kind of elastic yielding,
since the communication of motion is not instantaneous, but occupies time.

The medium is therefore capable of receiving and storing up two kinds of
energy, namely, the "actual" energy depending on the motions of its parts, and
"potential" energy, consisting of the work which the medium will do in recover-
ing from displacement in virtue of its elasticity.

* "On the Possible Density of the Luminiferous Medium, and on the Mechanical Value of a
t'ubic Mile of Sunlight," Transactions of tlie Royal Society of Edinburgh (1854), p. 57.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 529

The propagation of undulations consists in the continual transformation of
one of these forms of energy into the other alternately, and at any instant
the amount of energy in the whole medium is equally divided, so that half
is energy of motion, and half is elastic resilience.

(7) A medium having such a constitution may be capable of other kinds
of motion and displacement than those which produce the phenomena of light
and heat, and some of these may be of such a kind that they may be
evidenced to our senses by the phenomena they produce.

(8) Now we know that the luminiferous medium is in certain cases acted
on by magnetism ; for Faraday* discovered that when a plane polarized ray
traverses a transparent diamagnetic medium in the direction of the lines of
magnetic force produced by magnets or currents in the neighbourhood, the plane
of polarization is caused to rotate.

This rotation is always in the direction in which positive electricity must
be carried round the diamagnetic body in order to produce the actual mag-
netization of the field.

M. Verdetf has since discovered that if a paramagnetic body, such as
solution of perchloride of iron in ether, be substituted for the diamagnetic body,
the rotation is in the opposite direction.

Now Professor W. Thomson J has pointed out that no distribution of forces
acting between the parts of a medium whose only motion is that of the lumi-
nous vibrations, is sufficient to account for the phenomena, but that we must
admit the existence of a motion in the medium depending on the magnetization,
in addition to the vibratory motion which constitutes light.

It is true that the rotation by magnetism of the plane of polarization has
been observed only in media of considerable density ; but the properties of the
magnetic field are not so much altered by the substitution of one medium for
another, or for a vacuum, as to allow us to suppose that the dense medium
does anything more than merely modify the motion of the ether. We have
therefore warrantable grounds for inquiring whether there may not be a motion
of the ethereal medium going on wherever magnetic effects are observed, and

* Experimental Researclies, Series xix.

t Comptes Rendus (1856, second half year, p. 529, and 1857, first half year, p. 1209).

\ Proceedings of the Royal Society, June 1856 and June 1861.

VOL. i. 67

330 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

we have some reason to suppose that this motion is one of rotation, having
the direction of the magnetic force as its axis.

(9) We may now consider another phenomenon observed in the electro-
magnetic field. When a body is moved across the lines of magnetic force it
experiences what is called an electromotive force ; the two extremities of the
body tend to become oppositely electrified, and an electric current tends to flow
through the body. When the electromotive force is sufficiently powerful, and is
made to act on certain compound bodies, it decomposes them, and causes one
of their components to pass towards one extremity of the body, and the other
in the opposite direction.

Here we have evidence of a force causing an electric current in spite of
resistance ; electrifying the extremities of a body in opposite ways, a condition
which is sustained only by the action of the electromotive force, and which, as
soon as that force is removed, tends, with an equal and opposite force, to
produce a counter current through the body and to restore the original electrical
state of the body ; and finally, if strong enough, tearing to pieces chemical
compounds and carrying their components in opposite directions, while their
natural tendency is to combine, and to combine with a force which can generate
an electromotive force in the reverse direction.

This, then, is a force acting on a body caused by its motion through the
electromagnetic field, or by changes occurring in that field itself; and the eftect
of the force is either to produce a current and heat the body, or to decompose
the body, or, when it can do neither, to put the body in a state of electric
polarization, — a state of constraint in which opposite extremities are oppositely
electrified, and from which the body tends to relieve itself as soon as the
disturbing force is removed.

(10) According to the theory which I propose to explain, this "electro-
motive force " is the force called into play during the communication of motion
from one part of the medium to another, and it is by means of this force
tliat the motion of one part causes motion in another part. When electromotive
force acts on a conducting circuit, it produces a current, which, as it meets
with resistance, occasions a continual transformation of electrical energy into
heat, which is incapable of being restored again to the form of electrical energy
by any reversal of the process.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 531

(11) But when electromotive force acts on a dielectric it produces a state
of polarization of its parts similar in distribution to the polarity of the parts
of a mass of iron under the influence of a magnet, and like the magnetic
polarization, capable of being described as a state in which every particle has
its opposite poles in opposite conditions'*.

In a dielectric under the action of electromotive force, we may conceive
that the electricity in each molecule is so displaced that one side is rendered
positively and the other negatively electrical, but that the electricity remains
entirely connected with the molecule, and does not pass from one molecule to
another. The effect of this action on the whole dielectric mass is to produce
a general displacement of electricity in a certain direction. This displacement
does not amount to a current, because when it has attained to a certain value
it remains constant, but it is the commencement of a current, and its varia-
tions constitute currents in the positive or the negative direction ac2ording as
the displacement is increasing or decreasing. In the interior of the dielectric
there is no indication of electrification, because the electrification of the surface
of any molecule is neutralized by the opposite electrification of the surface of
the molecules in contact with it ; but at the bounding surface of the dielectric,
where the electrification is not neutralized, we find the phenomena which indicate
positive or negative electrification.

The relation between the electromotive force and the amount of electric
displacement it produces depends on the nature of the dielectric, the same
electromotive force producing generally a greater electric displacement in solid
dielectrics, such as glass or sulphur, than in air.

(12) Here, then, we perceive another effect of electromotive force, namely,
electric displacement, which according to our theory is a kind of elastic yielding
to the action of the force, similar to that which takes place in structures and
machines owing to the want of perfect rigidity of the connexions.

(13) The practical investigation of the inductive capacity of dielectrics is
rendered difficult on account of two disturbing phenomena. The first is the
conductivity of the dielectric, which, though in many cases exceedingly small,
is not altogether insensible. The second is the phenomenon called electric absorp-

* Faraday, Experimental ResearcJias, Series XI. ; Mossotti, Mem. delta Soc. Italiana (Modena),
Vol. xxiv. Part 2, p. 49.

67—2

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

tion *, in virtue of which, when the dielectric is exposed to electromotive force,
the electric displacement gradually increases, and when the electromotive force
is removed, the dielectric does not instantly return to its primitive state, but
only discharges a portion of its electrification, and when left to itself gradually
acquires electrification on its surface, as the interior gradually becomes depolarized.
Almost all solid dielectrics exhibit this phenomenon, which gives rise to the
residual charge in the Leyden jar, and to several phenomena of electric cables
described by Mr F. Jenkint.

(14) We have here two other kinds of yielding besides the yielding of
the perfect dielectric, which we have compared to a perfectly elastic body. The
yielding due to conductivity may be compared to that of a viscous fluid (that
is to say, a fluid having great internal friction), or a soft solid on which the
smallest force produces a permanent alteration of figure increasing with the
time during which the force acts. The yielding due to electric absorption may
be compared to that of a cellular elastic body containing a thick fluid in its
cavities. Such a body, when subjected to pressure, is compressed by degrees
on account of the gradual yielding of the thick fluid ; and when the pressure
is removed it does not at once recover its figure, because the elasticity of the
substance of the body has gradually to overcome the tenacity of the fluid before
it can regain complete equilibrium.

Several solid bodies in which no such structure as we have supposed can
be found, seem to possess a mechanical property of this kind J ; and it seems
probable that the same substances, if dielectrics, may possess the analogous
electrical property, and if magnetic, may have corresponding properties relating
to the acquisition, retention, and loss of magnetic polarity.

(15) It appears therefore that certain phenomena in electricity and mag-
netism lead to the same conclusion as those of optics, namely, that there is
an asthereal medium pervading all bodies, and modified only in degree by their
presence ; that the parts of this medium are capable of being set in motion
by electric currents and magnets ; that this motion is communicated from one

* Faraday, Experimental Researches, 1233—1250.

t Report* of British Association, 1859, p. 248; and Report of Committee of Board of Trade on
Submarine Cablet, pp. 136 it 464.

J As, for instance, the composition of glue, treacle, ic., of which small plastic figures are made,
which after being distorted gradually recover their bhape.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 533

part of the medium to another by forces arising from the connexions of those
parts ; that under the action of these forces there is a certain yielding depending
on the elasticity of these connexions ; and that therefore energy in two different
forms may exist in the medium, the one form being the actual energy of motion
of its parts, and the other being the potential energy stored up in the con-
nexions, in virtue of their elasticity.

(16) Thus, then, we are led to the conception of a complicated mechanism
capable of a vast variety of motion, but at the same time so connected that
the motion of one part depends, according to definite relations, on the motion
of other parts, these motions being communicated by forces arising from the
relative displacement of the connected parts, in virtue of their elasticity. Such
a mechanism must be subject to the general laws of Dynamics, and we ought
to be able to work out all the consequences of its motion, provided we know
the form of the relation between the motions of the parts.

(17) We know that when an electric current is established in a conducting
circuit, the neighbouring part of the field is characterized by certain magnetic
properties, and that if two circuits are in the field, the magnetic properties of
the field due to the two currents are combined. Thus each part of the field
is in connexion with both currents, and the two currents are put in connexion
with each other in virtue of their connexion with the magnetization of the field.
The first result of this connexion that I propose to examine, is the induction of
one current by another, and by the motion of conductors in the field.

The second result, which is deduced from this, is the mechanical action
between conductors carrying currents. The phenomenon of the induction of
currents has been deduced from their mechanical action by Helmholtz * and
Thomson t. I have followed the reverse order, and deduced the mechanical action
from the laws of induction. I have then described experimental methods of
determining the quantities L, M, N, on which these phenomena depend.

(18) I then apply the phenomena of induction and attraction of currents
to the exploration of the electromagnetic field, and the laying down systems
of lines of magnetic force which indicate its magnetic properties. By exploring

* "Conservation of Force," Physical Society of Berlin, 1847; and Taylor's Scientific Memoirs, 1853,
p. 114.

t Reports of the British Association, 1848; Philosophical Magazine, Dec. 1851.

534 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

the same field with a magnet, I shew the distribution of its equipotential
magnetic surfaces, cutting the lines of force at right angles.

In order to bring these results within the power of symbolical calculation,
I then express them in the form of the General Equations of the Electro-
magnetic Field. These equations express—

(A) The relation between electric displacement, true conduction, and the
total current, compounded of both.

(B) The relation between the lines of magnetic force and the inductive
coefficients of a circuit, as already deduced from the laws of induction.

(C) The relation between the strength of a current and its magnetic effects,
according to the electromagnetic system of measurement.

(D) The value of the electromotive force in a body, as arising from the
motion of the body in the field, the alteration of the field itself, and
the variation of electric potential from one part of the field to
another.

(E) The relation between electric displacement, and the electromotive force
which produces it.

(F) The relation between an electric current, and the electromotive force
which produces it.

(G) The relation between the amount of free electricity at any point, and
the electric displacements in the neighbourhood.

(H) The relation between the increase or diminution of free electricity and
the electric currents in the neighbourhood.

There are twenty of these equations in all, involving twenty variable
quantities.

(19) I then express in terms of these quantities the intrinsic energy of
the Electromagnetic Field as depending partly on its magnetic and partly on
its electric polarization at every point.

From this I determine the mechanical force acting, 1st, on a moveable con-
ductor carrying an electric current ; 2ndly, on a magnetic pole ; 3rdly, on an
electrified body.

The last result, namely, the mechanical force acting on an electrified body,
•/ives rise to an independent method of electrical measurement founded on its

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 535

electrostatic effects. The relation between the units employed in the two methods
is shewn to depend on what I have called the "electric elasticity" of the medium,
and to be a velocity, which has been experimentally determined by MM. Weber
and Kohlrausch.

I then shew how to calculate the electrostatic capacity of a condenser, and
the specific inductive capacity of a dielectric.

The case of a condenser composed of parallel layers of substances of different
electric resistances and inductive capacities is next examined, and it is shewn
that the phenomenon called electric absorption will generally occur, that is, the
condenser, when suddenly discharged, will after a short time shew signs of a

residual charge.

/

(20) The general equations are next applied to the case of a magnetic
disturbance propagated through a non-conducting field, and it is shewn that
the only disturbances which can be so propagated are those which are transverse
to the direction of propagation, and that the velocity of propagation is the
velocity v, found from experiments such as those of Weber, which expresses
the number of electrostatic units of electricity which are contained in one electro-
magnetic unit.

This velocity is so nearly that of light, that it seems we have strong
reason to conclude that light itself (including radiant heat, and other radiations
if any) is an electromagnetic disturbance in the form of waves propagated through
the electromagnetic field according to electromagnetic laws. If so, the agree-
ment between the elasticity of the medium as calculated from the rapid alterna-
tions of luminous vibrations, and as found by the slow processes of electrical
experiments, shews how perfect and regular the elastic properties of the medium
must be when not encumbered with any matter denser than air. If the same
character of the elasticity is retained in dense transparent bodies, it appears
that the square of the index of refraction is equal to the product of the specific
dielectric capacity and the specific magnetic capacity. Conducting media are
shewn to absorb such radiations rapidly, and therefore to be generally opaque.

The conception of the propagation of transverse magnetic disturbances to
the exclusion of normal ones is distinctly set forth by Professor Faraday* in
his "Thoughts on Kay Vibrations." The electromagnetic theory of light, as

* Philosophical JJagazitie, May 1846, or Experimental Researches, in. p. 447.

536 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

proposed by him, is the same in substance as that which I have begun to
develope in this paper, except that in 1846 there were no data to calculate
the velocity of propagation.

(21) The general equations are then applied to the calculation of the coef-
ficients of mutual induction of two circular currents and the coefficient of self-
induction in a coil. The want of uniformity of the current in the different
parts of the section of a wire at the commencement of the current is investi-
gated, I believe for the first time, and the consequent correction of the coefficient
of self-induction is found.

These results are applied to the calculation of the self-induction of the coil
used in the experiments of the Committee of the British Association on Standards
of Electric Resistance, and the value compared with that deduced from the
experiments.

PART II.

ON ELECTROMAGNETIC INDUCTION.

Electromagnetic Momentum of a Current.

(22) We may begin by considering the state of the field in the neigh-
bourhood of an electric current. We know that magnetic forces are excited in
the field, their direction and magnitude depending according to known laws
upon the form of the conductor carrying the current. When the strength of
the current is increased, all the magnetic effects are increased in the same pro-
portion. Now, if the magnetic state of the field depends on motions of the
medium, a certain force must be exerted in order to increase or diminish these
motions, and when the motions are excited they continue, so that the effect
of the connexion between the current and the electromagnetic field surrounding
it, is to endow the current with a kind of momentum, just as the connexion
between the driving-point of a machine and a fly-wheel endows the driving-point
with an additional momentum, which may be called the momentum of the fly-
wheel reduced to the driving-point. The unbalanced force acting on the driving-
point increases this momentum, and is measured by the rate of its increase.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 537

In the case of electric currents, the resistance to sudden increase or dimi-
nution of strength produces effects exactly like those of momentum, but the
amount of this momentum depends on the shape of the conductor and the
relative position of its different parts.

Mutual Action of two Currents.

(23) If there are two electric currents in the field, the magnetic force at
any point is that compounded of the forces due to each current separately,
and since the two currents are in connexion with every point of the field,
they will be in connexion with each other, so that any increase or diminution

of the one will produce a force acting with or contrary to the other.

/

Dynamical Illustration of Reduced Momentum.

(24) As a dynamical illustration, let us suppose a body C so connected
with two independent driving-points A and B that its velocity is p times that
of A together with q times that of B. Let u be the velocity of A, v that
of B, and w that of C, and let Sx, Sy, Sz be their simultaneous displacements,
then by the general equation of dynamics*,

where X and Y are the forces acting at A and B.

dw du dv

But dt=Pdi + ^Tt'

and Sz = p8x + q&y.

Substituting, and remembering that Sx and Sy are independent,

, ............................ (I)-

We may call Cp'u+Cpqv the momentum of C referred to A, and Cpqu + Cq*v
its momentum referred to B ; then we may say that the effect of the force
X is to increase the momentum of C referred to A, and that of Y to increase
its momentum referred to B,

* Lagrange, Mec. Anal. II. 2, § 5.

VOL. i. 68

538 A DYNAMICAL THEOnY OF THE ELECTROMAGNETIC FIELD.

If there are many bodies connected with A and H in & similar way but
with different values of p and q, we may treat the question in the same way
by assuming

L = 2 (Cjaf), M= 2 (Cpq), and N= X (Cf),

where the summation is extended to all the bodies with their proper values of
C, p, and q. Then the momentum of the system referred to A is

Lu + Mv,
and referred to B, Hu+Nv,

and we shall have •^~

(2),

where X and Y are the external forces acting on A and B.

(25) To make the illustration more complete we have only to suppose
that the motion of A is resisted by a force proportional to its velocity, which
we may call Ru, and that of B by a similar force, which we may call Sv, R and
S being coefficients of resistance. Then if £ and 77 are the forces on A and B,

If the velocity of A be increased at the rate -j- , then in order to prevent B
from moving a force, "n = -f. (Mil) must be applied to it.

This effect on B, due to an increase of the velocity of A, corresponds to
the electromotive force on one circuit arising from an increase in the strength
of a neighbouring circuit.

This dynamical illustration is to be considered merely as assisting the
reader to understand what is meant in mechanics by Reduced Momentum. The
facts of the induction of currents as depending on the variations of the quantity
called Electromagnetic Momentum, or Electrotonic State, rest on the experiments
of Faraday*, Felicif, &c.

• Experimental Kesearctits, Scries i., ix. t Annales de CMmie, s£r. 3, xxxiv. (1852), p. 64.

A DYNAMICAL THEOKY OF THE ELECTROMAGNETIC FIELD. 539

Coefficients of Induction for Two Circuits.

(26) In the electromagnetic field the values of L, M, N depend on the
distribution of the magnetic effects due to the two circuits, and this distri-
bution depends only on the form and relative position of the circuits. Hence
L, M, N are quantities depending on the form and relative position of the
circuits, and are subject to variation with the motion of the conductors. It will
be presently seen that L, M, N are geometrical quantities of the nature of lines,
that is, of one dimension in space ; L depends on the form of the first conductor,
which we shall call A, N on that of the second, which we shall call B, and
M on the relative position of A and B,

(27) Let £ be the electromotive force acting on A, 'x the strength of the
current, and R the resistance, then Rx will be the resisting force. In steady
currents the electromotive force just balances the resisting force, but in variable
currents the resultant force £—Rx is expended in increasing the "electro-
magnetic momentum," using the word momentum merely to express that which
is generated by a force acting during a time, that is, a velocity existing in a
body.

In the case of electric currents, the force in action is not ordinary
mechanical force, at least we are not as yet able to measure it as common force,
but we call it electromotive force, and the body moved is not merely the
electricity in the conductor, but something outside the conductor, and capable
of being affected by other conductors in the neighbourhood carrying currents.
In this it resembles rather the reduced momentum of a driving-point of a
machine as influenced by its mechanical connexions, than that of a simple
moving body like a cannon ball, or water in a tube.

Electromagnetic Relations of two Conducting Circuits.

(28) In the case of two conducting circuits, A and B, we shall assume
that the electromagnetic momentum belonging to A is

Lx + My,

and that belonging to B, Mx + Ny,

where L, M, N correspond to the same quantities in the dynamical illustration,
except that they are supposed to be capable of variation when the conductors
A or B are moved.

68—2

540 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

Then the equation of the current x in A will be

(4),

;md that of y in B TJ = Sy + -7- ( MX + Ny] ............................ (5),

where £ and 17 are the electromotive forces, x and y the currents, and R and <S
the resistances in A and B respectively.

Induction of one Current by another.

(29) Case 1st. Let there be no electromotive force on B, except that
which arises from the action of A, and let the current of A increase from 0
to the value x, then

n tf

whence Y = I ydt = — -~ x, .............................. (6)

Jo »

that is, a quantity of electricity Y, being the total induced current, will flow
through B when x rises from 0 to x. This is induction by variation of the
current in the primary conductor. When M is positive, the induced current
due to increase of the primary current is negative.

Induction by Motion of Conductor.

(30) Case 2nd. Let x remain constant, and let M change from M to M',
then

M'-M

so that if M is increased, which it will be by the primary and secondary
circuits approaching each other, there will be a negative induced current, the
total quantity of electricity passed through B being Y.

This is induction by the relative motion of the primary and secondary con-
ductors.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 541

Equation of Work and Energy.

(31) To form the equation between work done and energy produced,
multiply (1) by x and (2) by y, and add

-r (Mx + Ny) .......... (8).

-r

Here £x is the work done in unit of time by the electromotive force £ acting
on the current x and maintaining it, and r}y is the work done by the elec-
tromotive force r). Hence the left-hand side of the equation represents the work
done by the electromotive forces in unit of time.

Heat produced by the Current.

(32) On the other side of the equation we have, first,

Ry? + Sy* = H ................................. (9),

which represents the work done in overcoming the resistance of the circuits in
unit of time. This is converted into heat. The remaining terms represent
work not converted into heat. They may be written

Intrinsic Energy of the Currents.

(33) If L, M, N are constant, the whole work of the electromotive forces
which is not spent against resistance will be devoted to the development of
the currents. The whole intrinsic energy of the currents is therefore

$Lxi + Mxy + $Ny' = E ........................... (10).

This energy exists in a form imperceptible to our senses, probably as actual
motion, the seat of this motion being not merely the conducting circuits, but
the space surrounding them.

Mechanical Action between Conductors.

(34) The remaining terms,

dL . dM dN . ,

542 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

represent the work done in unit of time arising from the variations of L, M,
and N, or, what is the same thing, alterations in the form and position of the
conducting circuits A and JB.

Now if work is done when a body is moved, it must arise from ordinary
mechanical force acting on the body while it is moved. Hence this part of
the expression shews that there is a mechanical force urging every part of the
conductors themselves in that direction in which L, M, and N will be most
increased.

The existence of the electromagnetic force between conductors carrying
currents is therefore a direct consequence of the joint and independent action
of each current on the electromagnetic field. If A and B are allowed to approach
a distance ds, so as to increase M from M to M' while the currents are x
and y, then the work done will be

(M'-M)xy,

and the force in the direction of ds will be

<">•

and this will be an attraction if x and y are of the same sign, and if M is
increased as A and B approach.

It appears, therefore, that if we admit that the unresisted part of electro-
motive force goes on as long as it acts, generating a self-persistent state of
the current, which we may call (from mechanical analogy) its electromagnetic
momentum, and that this momentum depends on circumstances external to the
conductor, then both induction of currents and electromagnetic attractions may
be proved by mechanical reasoning.

What I have called electromagnetic momentum is the same quantity which
is called by Faraday * the electrotonic state of the circuit, every change of which
involves the action of an electromotive force, just as change of momentum
involves the action of mechanical force.

If, therefore, the phenomena described by Faraday in the Ninth Series of
his Experimental Rescarclics were the only known facts about electric currents,
the laws of Ampere relating to the attraction of conductors carrying currents,

* Experimental Researches, Series i. 60, <fec.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 543

as well as those of Faraday about the mutual induction of currents, might be
deduced by mechanical reasoning.

In order to bring these results within the range of experimental verifica-
tion, I shall next investigate the case of a single current, of two currents, and
of the six currents in the electric balance, so as to enable the experimenter
to determine the values of L, M, N.

Case of a single Circuit.

(35) The equation of the current a; in a circuit whose resistance is R,
and whose coefficient of self-induction is L, acted on by an external electro-
motive force £, is /

.............................. (13).

When $ is constant, the solution is of the form

where a is the value of the current at the commencement, and b is its final
value.

The total quantity of electricity which passes in time t, where t is great, is

,L

(14)-

The value of the integral of of with respect to the time is

(15).

The actual current changes gradually from the initial value a to the final value
b, but the values of the integrals of x and a? are the same as if a steady

current of intensity \ (a + b) were to flow for a time 2 -^ , and were then suc-
ceeded by the steady current b. The time 2 -p is generally so minute a fraction

xc

of a second, that the effects on the galvanometer and dynamometer may be
calculated as if the impulse were instantaneous.

If the circuit consists of a battery and a coil, then, when the circuit is
first completed, the effects are the same as if the current had only half its

544 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

6nal strength during the time 2-^. This diminution of the current, due to

induction, is sometimes called the counter-current.

(36) If an additional resistance r is suddenly thrown into the circuit, as
by breaking contact, so as to force the current to pass through a thin wire

of resistance r, then the original current is « = •«, and the final current is

-R + r'

The current of induction is then %£ //7p^' \ » anc^ continues for a time

2 -JT . This current is greater than that which the battery can maintain in

fC T~ 7*

the two wires R and r, and may be sufficient to ignite the thin wire r.

When contact is broken by separating the wires in air, this additional
resistance is given by the interposed air, and since the electromotive force across
the new resistance is very great, a spark will be forced across.

If the electromotive force is of the form Esinpt, as in the case of a coil
revolving in the magnetic field, then

E

* • ' •-*),

where p* = R1 + Up*, and tan a = -• .

Case of two Circuits.

(37) Let R be the primary circuit and S the secondary circuit, then we
liuve a case similar to that of the induction coil.

The equations of currents are those marked A and Bt and we may here
assume L, M, N as constant because there is no motion of the conductors.
The equations then become

(13*).

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 545

To find the total quantity of electricity which passes, we have only to
integrate these equations with respect to t ; then if x0, ya be the strengths of
the currents at time 0, and xu yl at time t, and if X, Y be the quantities
of electricity passed through each circuit during time t,

(14*).

When the circuit R is completed, then the total currents up to time t,
when t is great, are found by making /

then X = Xlt-, Y=~x, ..................... (15*).

The value of the total counter-current in R is therefore independent of the
secondary circuit, and the induction current in the secondary circuit depends only
on M, the coefficient of induction between the coils, S the resistance of the
secondary coil, and xl the final strength of the current in R.

When the electromotive force £ ceases to act, there is an extra current
in the primary circuit, and a positive induced current in the secondary circuit,
whose values are equal and opposite to those produced on making contact.

(38) All questions relating to the total quantity of transient currents, as
measured by the impulse given to the magnet of the galvanometer, may be
solved in this way without the necessity of a complete solution of the equa-
tions. The heating effect of the current, and the impulse it gives to the
suspended coil of Weber's dynamometer, depend on the square of the current
at every instant during the short time it lasts. Hence we must obtain the
solution of the equations, and from the solution we may find the effects both
on the galvanometer and dynamometer ; and we may then make use of the
method of Weber for estimating the intensity and duration of a current uniform
while it lasts which would produce the same effects.

VOL. I. 69

546 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

(39) Let n,, «, be the roots of the equation

(LN-M')n'+(RN+lS)n + RS=0 ..................... (16),

and let the primary coil be acted on by a constant electromotive force Re, so
that c is the constant current it could maintain ; then the complete solution of
the equations for making contact is

x^n^_ (IS +N\^fS \ « + s*^} ............ (17)

Sn,-nt \\nl / \n, / n,na J

_ ..............................................

S Hi-n^ l

From these we obtain for calculating the impulse on the dynamometer,

The effects of the current in the secondary coil on the galvanometer and
dynamometer are the same as those of a uniform current

MR

~*°RN+LS

i
for a time

(£+£>

(R + S>-

(40) The equation between work and energy may be easily verified. The
work done by the electromotive force is

Work done in overcoming resistance and producing heat,

R]y?dt + Stfdt = c< (Rt - fL).
Energy remaining in the system,

(41) If the circuit R is suddenly and completely interrupted while carrying
a current c, then the equation of the current in the secondary coil would be

M -f«

y = c-^e

M

This current begins with a value c-, and gradually disappears.

A DYNAMICAL THEOBY OF THE ELECTROMAGNETIC FIELD.

547

M M3

The total quantity of electricity is c -~ , and the value of tydt is c2 •

The effects on the galvanometer and dynamometer are equal to those of a
M

unilonn current

f +' »

tor a time 2 - .

The heating effect is therefore greater than that of the current on making-
contact.

(42) If an electromotive force of the form g=Ecospt acts on the circuit
R, then if the circuit S is removed, the value of x will be

E

: = -. sm(pt-a),

where
and

tan a =

it=R*+Ly>
j£.

R '

The effect of the presence of the circuit S in the neighbourhood is to
alter the value of A and a, to that which they would be if R became

and L became

R+p*

L-p>

MS

MN
tf+pW

Hence the effect of the presence of the circuit S is to increase the apparent
resistance and diminish the apparent self-induction of the circuit R.

On the Determination of Coefficients of Induction by the Electric Balance.

(43) The electric balance consists of six con-
ductors joining four points, A, C, D, E, two
and two. One pair, AC, of these points is con-
nected through the battery B. The opposite pair, A,
DE, is connected through the galvanometer G,
Then if the resistances of the four remaining
conductors are represented by P, Q, R, S, and
the currents in them by x, x — z, y, and y + z,

69—2

548

A DYNAMICAL THEOKY OF THE ELECTROMAGNETIC FIELD.

the current through G will be z. Let the potentials at the four points be A, C,
D, E. Then the conditions of steady currents may be found from the equations

= A-D, Q(x-z) = D-C

Ry = A-E, S(y + z)=E-C
Gz=D-E, E(

Solving these equations for z, we find

(21).

EG

In this expression F is the electromotive force of the battery, z the current
through the galvanometer when it has become steady. P, Q, R, S the resistances
in the four arms. B that of the battery and electrodes, and G that of the
galvanometer.

(44) If PS=QR, then 2 = 0, and there will be no steady current, but a
transient current through the galvanometer may be produced on making or
breaking circuit on account of induction, and the indications of the galvano-
meter may be used to determine the coefficients of induction, provided we
understand the actions which take place.

We shall suppose PS=QR, so that the current z vanishes when sufficient
time is allowed, and

Let the induction coefficients between P, Q, R, S
l« given by the following Table, the coefficient of in-
duction of P on itself being p, between P and Q, h,
and so on.

Let g be the coefficient of induction of the gal-
vanometer on itself, and let it be out of the reach of
the inductive influence of P, Q, R, S (as it must be
in order to avoid direct action of P, Q, R, S on the
needle) Let Xt Y, Z be the integrals of x, y, z with respect to t. At

?)(* +

W

\&o

)•

P

Q

R

S

P

P

h

k

I

Q

h

(l

m

n

R

k

m

r

0

S

I

n

0

s

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

549

making contact x, y, z are zero. After a time z disappears, and x and y reach
constant values. The equations for each conductor will therefore be

PX + (p + h)x + (k+l}y = \Adt-

(24).

RY + (k+m)x + (r + o)y = \Adt-\Edt

S(Y+Z}+(1 + n)x + (o +s)y= \Edt-\Cdt
GZ= \Ddt-\Edt.

Solving these equations for Z, we find

^...(25).

(45) Now let the deflection of the galvanometer by the instantaneous
current whose intensity is Z be a.

*Let the permanent deflection produced by making the ratio of PS to QR,
p instead of unity, be 6.

Also let the time of vibration of the galvanometer needle from rest to rest
be T.

Then calling the quantity

'1
P~ Q~ R^ S~* "\P~Ql '

we find

tan 6 TT l—p

.(26),
(27).

* [In those circumstances the values of x and y found in Art. 44 require modification before
l>eing inserted in equation (24). This has been pointed out by Lord Rayleigh, who employed the
method described in the text in his second determination of the British unit of resistance in
absolute measure. See the Philosophical Transactions, 1882, Part II. pp. 677, 678.]

550 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

In determining T by experiment, it is best to make the alteration of resist-
ance in one of the arms by means of the arrangement described by Mr Jenkin
in the Report of the British Association for 1863, by which any value of p
from 1 to 1*01 can be accurately measured.

We observe (a) the greatest deflection due to the impulse of induction
when the galvanometer is in circuit, when the connexions are made, and when
the resistances are so adjusted as to give no permanent current.

We then observe (/J) the greatest deflection produced by the permanent
current when the resistance of one of the arms is increased in the ratio of
1 to p, the galvanometer not being in circuit till a little while after the con-
nexion is made with the battery.

In order to eliminate the effects of resistance of the air, it is best to vary
p till )8 = 2a nearly ; then

£ (28).

If all the arms of the balance except P consist of resistance coils of very
fine wire of no great length and doubled before being coiled, the induction
coefficients belonging to these coils will be insensible, and T will be reduced

to lp. The electric balance therefore affords the means of measuring the self-
induction of any circuit whose resistance is known.

(46) It may also be used to determine the coefficient of induction between
two circuits, as for instance, that between P and S which we have called ra;
but it would be more convenient to measure this by directly measuring the
current, as in (37), without using the balance. We may also ascertain the

fin rt

equality of *p and i< by there being no current of induction, and thus, when

we know the value of p, we may determine that of q by a more perfect method
than the comparison of deflections.

Exploration of the Electromagnetic Field.

(47) Let us now suppose the primary circuit A to be of invariable form,
and let us explore the electromagnetic field by means of the secondary circuit
ft, which we shall suppose to be variable in form and position.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 551

We may begin by supposing B to consist of a short straight conductor
with its extremities sliding on two parallel conducting rails, which are put in
connexion at some distance from the sliding-piece.

Then, if sliding the moveable conductor in a given direction increases the
value of Mt a negative electromotive force will act in the circuit B, tending
to produce a negative current in B during the motion of the sliding-piece.

If a current be kept up in the circuit B, then the sliding-piece will itself
tend to move in that direction, which causes M to increase. At every point
of the field there will always be a certain direction such that a conductor moved
in that direction does not experience any electromotive force in whatever direc-
tion its extremities are turned. A conductor carrying a current will experience
no mechanical force urging it in that direction or the opposite.

This direction is called the direction of the line of magnetic force through
that point.

Motion of a conductor across such a line produces electromotive force in
a direction perpendicular to the line and to the direction of motion, and a con-
ductor carrying a current is urged in a direction perpendicular to the line and
to the direction of the current.

(48) We may next suppose B to consist of a very small plane circuit
capable of being placed in any position and of having its plane turned in any
direction. The value of M will be greatest when the plane of the circuit is
perpendicular to the line of magnetic force. Hence if a current is maintained
in B it will tend to set itself in this position, and will of itself indicate, like
a magnet, the direction of the magnetic force.

On Lines of Magnetic Force.

(49) Let any surface be drawn, cutting the lines of magnetic force, and
on this surface let any system of lines be drawn at small intervals, so as to
lie side by side without cutting each other. Next, let any line be drawn on
the surface cutting all these lines, and let a second line be drawn near it, its
distance from the first being such that the value of M for each of the small
spaces enclosed between these two lines and the lines of the first system is
equal to unity.

In this way let more lines be drawn so as to form a second system, so

552 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

that the value of M for every reticulation formed by the intersection of the
two systems of lines is unity.

Finally, from every point of intersection of these reticulations let a line be
drawn through the field, always coinciding in direction with the direction of
magnetic force.

(50) In this way the whole field will be filled with lines of magnetic force
at regular intervals, and the properties of the electromagnetic field will be com-
pletely expressed by them.

For, 1st, If any closed curve be drawn in the field, the value of M for
that curve will be expressed by the number of lines of force which pass through
that closed curve.

2ndly. If this curve be a conducting circuit and be moved through the
field, an electromotive force will act in it, represented by the rate of decrease
of the number of lines passing through the curve.

3rdly. If a current be maintained in the circuit, the conductor will be
acted on by forces tending to move it so as to increase the number of lines
passing through it, and the amount of work done by these forces is equal to
the current in the circuit multiplied by the number of additional lines.

4thly. If a small plane circuit be placed in the field, and be free to turn,
it will place its plane perpendicular to the lines of force. A small magnet will
place itself with its axis in the direction of the lines of force.

Sthly. If a long uniformly magnetized bar is placed in the field, each pole
will be acted on by a force in the direction of the lines of force. The number
of lines of force passing through unit of area is equal to the force acting on
a unit pole multiplied by a coefficient depending on the magnetic nature of the
medium, and called the coefficient of magnetic induction.

In fluids and isotropoic solids the value of this coefficient p, is the same
in whatever direction the lines of force pass through the substance, but in
crystallized, strained, and organized solids the value of p. may depend on the
direction of the lines of force with respect to the axes of crystallization, strain,
or growth.

In all bodies p. is affected by temperature, and in iron it appears to diminish
as the intensity of the magnetization increases.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 553

On Magnetic Equipotential Surfaces.

(51) If we explore the field with a uniformly magnetized bar, so long that
one of its poles is in a very weak part of the magnetic field, then the mag-
netic forces will perform work on the other pole as it moves about the field.

If we start from a given point, and move this pole from it to any other
point, the work performed will be independent of the path of the pole between
the two points ; provided that no electric current passes between the different
paths pursued by the pole.

Hence, when there are no electric currents but only magnets in the field,
we may draw a series of surfaces such that the work done in passing from one
to another shall be constant whatever be the path pursued between them. Such
surfaces are called Equipotential Surfaces, and in ordinary cases are perpendicular
to the Lines of magnetic force.

If these surfaces are so drawn that, when a unit pole passes from any one
to the next in order, unity of work is done, then the work done in any motion
of a magnetic pole will be measured by the strength of the pole multiplied by
the number of surfaces which it has passed through in the positive direction.

(52) If there are circuits carrying electric currents in the field, then there
will still be equipotential surfaces in the parts of the field external to the con-
ductors carrying the currents, but the work done on a unit pole in passing
from one to another will depend on the number of times which the path of
the pole circulates round any of these currents. Hence the potential in each
surface will have a series of values in arithmetical progression, differing by the
work done in passing completely round one of the currents in the field.

The equipotential surfaces will not be continuous closed surfaces, but some
of them will be limited sheets, terminating in the electric circuit as their common
edge or boundary. The number of these will be equal to the amount of work
done on a unit pole in going round the current, and this by the ordinary
measurement =4iry, where y is the value of the current.

These surfaces, therefore, are connected with the electric current as soap-
bubbles are connected with a ring in M. Plateau's experiments. Every current
y has i-ny surfaces attached to it. These surfaces have the current for their
common edge, and meet it at equal angles. The form of the surfaces in other
parts depends on the presence of other currents and magnets, as well as on
the shape of the circuit to which they belong.

VOL. i. 70

554 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

PART III.
GENERAL EQUATIONS OF THE ELECTROMAGNETIC FIELD.

(53) Let us assume three rectangular directions in space as the axes of
x, y, and z, and let all quantities having direction be expressed by their com-
ponents in these three directions.

Electrical Cuivents (p, q, r).

(54) An electrical current consists in the transmission of electricity from
one part of a body to another. Let the quantity of electricity transmitted in
unit of time across unit of area perpendicular to the axis of x be called p, then
jj is the component of the current at that place in the direction of x.

We shall use the letters p, q, r to denote the components of the current
per unit of area in the directions of x, y, z.

Electrical Displacements (f, g, h).

(55) Electrical displacement consists in the opposite electrification of the
sides of a molecule or particle of a body which may or may not be accom-
panied with transmission through the body. Let the quantity of electricity which
would appear on the faces dy . dz of an element dx, dy, dz cut from the body
be f .dy .dz, then f is the component of electric displacement parallel to x. We
shall use f, g, h to denote the electric displacements parallel to x, y, z respectively.

The variations of the electrical displacement must be added to the currents
ft, q, r to get the total motion of electricity, which we may call p', q', r, so that

.(A).
dh

A DYNAMICAL THEORY OF THE ELECTBOMAGNETIC FIELD. 555

Electromotive Force (P, Q, R).

(56) Let P, Q, R represent the components of the electromotive force at
any point. Then P represents the difference of potential per unit of length in
a conductor placed in the direction of x at the given point. We may suppose
an indefinitely short wire placed parallel to x at a given point and touched,
during the action of the force P, by two small conductors, which are then
insulated and removed from the influence of the electromotive force. The value
of P might then be ascertained by measuring the charge of the conductors.

Thus if I be the length of the wire, the difference of potential at its ends
will be PI, and if (7 be the capacity of each of the small x conductors the charge
on each will be \CPl. Since the capacities of moderately large conductors,
measured on the electromagnetic system, are exceedingly small, ordinary electro-
motive forces arising from electromagnetic actions could hardly be measured in
this way. In practice such measurements are always made with long conductors,
forming closed or nearly closed circuits.

Electromagnetic Momentum (F, G, H).

(57) Let F, G, H represent the components of electromagnetic momentum
at any point of the field, due to any system of magnets or currents.

Then F is the total impulse of the electromotive force in the direction of
x that would be generated by the removal of these magnets or currents from
the field, that is, if P be the electromotive force at any instant during the
removal of the system

F-jPdt.

Hence the part of the electromotive force which depends on the motion of
magnets or currents in the field, or their alteration of intensity, is

dF dG dH

Electromagnetic Momentum of a Circuit.
(58) Let * be the length of the circuit, then if we integrate

s (30)

70—2

55G A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

round the circuit, we shall get the total electromagnetic momentum of the circuit,
or the number of lines of magnetic force which pass through it, the variations
of which measure the total electromotive force in the circuit. This electromag-
netic momentum is the same thing to which Professor Faraday has applied the
name of the Electrotonic State.

If the circuit be the boundary of the elementary area dtjdz, then its electro-
magnetic momentum is

fdll d

and this is the number of lines of magnetic force which pass through the
area dy dz.

Magnetic Force (a, ft, y).

(59) Let a, ft, y represent the force acting on a unit magnetic pole placed
at the given point resolved in the directions of x, y, and z.

Coefficient of Magnetic Induction (/x).

(GO) Let n be the ratio of the magnetic induction in a given medium to
that in air under an equal magnetizing force, then the number of lines of force
in unit of area perpendicular to x will be fia (p is a quantity depending on
the nature of the medium, its temperature, the amount of magnetization already
produced, and in crystalline bodies varying with the direction).

(61) Expressing the electric momentum of small circuits perpendicular to
the three axes in this notation, we obtain the following

Equations of Magnetic Force,

mdH_dQ

~ dy • !:

dF dH (B)

dG dF

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 557

Equations of Currents.

(62) It is known from experiment that the motion of a magnetic pole
in the electromagnetic field in a closed circuit cannot generate work unless the
circuit which the pole describes passes round an electric current. Hence, except
in the space occupied by the electric currents,

adx + fidy + ydz = d<f> (31)

a complete differential of <£, the magnetic potential.

The quantity <f> may be susceptible of an indefinite number of distinct values,
according to the number of times that the exploring poini passes round electric
currents in its course, the difference between successive values of <£ corre-
sponding to a passage completely round a current of strength c being 4-n-c.

Hence if there is no electric current,

dy dz
but if there is a current p',

dy dfi _ ,
dy dz *

dp _ da
dx dy

We may call these the Equations of Currents.

Electromotive Force in a Circuit.

(63) Let £ be the electromotive force acting round the circuit A, then

,„„.
s ........................... (32)'

where ds is the element of length, and the integration is performed round the
circuit.

N

558 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

Let the forces in the field be those due to the circuits A and B, then
the electromagnetic momentum of A is

where « and v are the currents in A and B, and

(34).

Hence, if there is no motion of the circuit A,

__
dt dx

-,,— S* (35),

dt dy

dt dz

j

where »/» is a function of x, y, z, and t, which is indeterminate as far as regards
the solution of the above equations, because the terms depending on it will
disappear on integrating round the circuit. The quantity $ can always, however,
be determined in any particular case when we know the actual conditions of
the question. The physical interpretation of \l> is, that it represents the electric
potential at each point of space.

Electromotive Force on a Moving Conductor.
(64) Let a short straight conductor of length a, parallel to the axis of

x, move with a velocity whose components are , and let its ex-

at dt at

ds
tremities slide along two parallel conductors with a velocity -r- . Let us find

the alteration of the electromagnetic momentum of the circuit of which this
arrangement forms a part.

(i,oc dt] dz
In unit of time the moving conductor has travelled distances -j- , -£ , -5-

along the directions of the three axes, and at the same time the lengths of

ds
the parallel conductors included in the circuit have each been increased by -7- .

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 559

Hence the quantity

f/j-.dx s^dv -rTdz\ ,
\(F-r + G~f-+H-r)ds
J \ ds ds dsj

will be increased by the following increments,

IdFdx , dFdy dFdz\ ,

a -y- -J- + -j- -77 + -j- -j. , due to motion of conductor,
\dx dt dy dt dz dt/

ds IdFdx , dGdy clH dz\ ,

' of circmt-

The total increment will therefore be

fdF_dG\dy_ (dH_d_F_\dz
*\dy dx) dt~ a( dx ~ dz ) dt '/

or. by the equations of Magnetic Force (8),

d dz

If P is the electromotive force in the moving conductor parallel to x referred
to unit of length, then the actual electromotive force is Pa ; and since this is
measured by the decrement of the electromagnetic momentum of the circuit, the
electromotive force due to motion will be

(65) The complete equations of electromotive force on a moving conductor
may now be written as follows : —

Equations of Electromotive Force.

fl,t, 1

. ___

" ~ dx

dz dx\ dG d\ii
dt-'Y-dt)--dt-dy

dx dy\ dll d\l>

-'--~

The first term on the right-hand side of each equation represents the electro-
motive force arising from the motion of the conductor itself. This electromotive

560 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

force is perpendicular to the direction of motion and to the lines of magnetic
force ; and if a parallelogram be drawn whose sides represent in direction
and magnitude the velocity of the conductor and the magnetic induction at that
point of the field, then the area of the parallelogram will represent the electro-
motive force due to the motion of the conductor, and the direction of the force
is perpendicular to the plane of the parallelogram.

The second term in each equation indicates the effect of changes in the
position or strength of magnets or currents in the field.

The third term shews the effect of the electric potential $. It has no effect
in causing a circulating current in a closed circuit. It indicates the existence
of a force urging the electricity to or from certain definite points in the field.

Electric Elasticity.

(66) When an electromotive force acts on a dielectric, it puts every part
of the dielectric into a polarized condition, in which its opposite sides are
oppositely electrified. The amount of this electrification depends on the electro-
motive force and on the nature of the substance, and, -in solids having a structure
defined by axes, on the direction of the electromotive force with respect to these
axes. In isotropic substances, if k is the ratio of the electromotive force to the
electric displacement, we may write the

Equations of Electric Elasticity,

(E).

Electi-ic Resistance.

(67) When an electromotive force acts on a conductor it produces a current
of electricity through it. This effect is additional to the electric displacement
already considered. In solids of complex structure, the relation between the
electromotive force and the current depends on their direction through the solid.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 561

In isotropic substances, which alone we shall here consider, if p is the specific
resistance referred to unit of volume, we may write the

Equations of Electric Resistance,

P=-pp]

Q=-pq\ (F).

R=-pr\

Electric Quantity.

(68) Let e represent the quantity of free positive electricity contained in
unit of volume at any part of the field, then, since this arises from the electri-
fication of the different parts of the field not neutralizing each other, we may
write the

Equation of Free Electricity,

e+df+dff + dh = 0 /Q\

dx eZy dz

(69) If the medium conducts electricity, then we shall have another con-
dition, which may be called, as in hydrodynamics, the

Equation of Continuity,

(70) In these equations of the electromagnetic field we have assumed twenty
variable quantities, namely,

For Electromagnetic Momentum F G H

,, Magnetic Intensity a /3 y

„ Electromotive Force P Q R

„ Current due to true Conduction p q r

„ Electric Displacement f g h

„ Total Current (including variation of displacement) p' q' r'

„ Quantity of Free Electricity e

„ Electric Potential ¥

VOL. I. 71

< -' A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

Between these twenty quantities we have found twenty equations, viz.

Three equations of Magnetic Force ................................. (B)

„ Electric Currents .............................. (C)

„ Electromotive Force ........................... (D)

„ Electric Elasticity .............................. (E)

„ Electric Kesistance .............................. (F)

Total Currents .................................. (A)

One equation of Free Electricity .................................... (G)

„ Continuity ......................................... (H)

These equations are therefore sufficient to determine all the quantities which
occur in them, provided we know the conditions of the problem. In many
questions, however, only a few of the equations are required.

Intrinsic Energy of tlie Electromagnetic Field.

(71) We have seen (33) that the intrinsic energy of any system of currents
is found by multiplying half the current in each circuit into its electromagnetic
momentum. This is equivalent to finding the integral

E = &(Fp'+G<I' + Hr')dV. .......................... (37)

over all the space occupied by currents, where p, q, r are the components of
currents, and F, G, II the components of electromagnetic momentum.

Substituting the values of p', q', r' from the equations of Currents (C),
this becomes

tdy d$\ (da. dy\
- -

Integrating by parts, and remembering that a, /8, y vanish at an infinite
distance, the expression becomes

IdH dG\ ldF dll\ fdG

where the integration is to be extended over all space. Referring to the equa-
tions of Magnetic Force (B), p. 556, tliis becomes

(38),

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 563

where a, /8, y are the components of magnetic intensity or the force on a unit
magnetic pole, and p,a, p./3, p.y are the components of the quantity of magnetic
induction, or the number of lines of force in unit of area.

In isotropic media the value of /a is the same in all directions, and we
may express the result more simply by saying that the intrinsic energy of any
part of the magnetic field arising from its magnetization is

f-r

Sir
per unit of volume, where / is the magnetic intensity.

(72) Energy may be stored up in the field in a different way, namely,
by the action of electromotive force in producing electric displacement. The
work done by a variable electromotive force, P, in producing a variable dis-
placement, f, is got by integrating

from P = 0 to the given value of P.

Since P = kf, equation (E), this quantity becomes

Hence the intrinsic energy of any part of the field, as existing in the
form of electric displacement, is

The total energy existing in the field is therefore

W-

The first term of this expression depends on the magnetization of the field,
and is explained on our theory by actual motion of some kind. The second
term depends on the electric polarization of the field, and is explained on our
theory by strain of some kind in an elastic medium.

(73) I have on a former occasion * attempted to describe a particular kind
of motion and a particular kind of strain, so arranged as to account for the
phenomena. In the present paper I avoid any hypothesis of this kind ; and in

* "On Physical Lines of Force," Philosophical Magazine, 1861—62. (In this vol. p. 451.)

71—2

564 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

using such words as electric momentum and electric elasticity in reference to
the known phenomena of the induction of currents and the polarization of
dielectrics, I wish merely to direct the mind of the reader to mechanical pheno-
mena which will assist him in understanding the electrical ones. All such phrases
in the present paper are to be considered as illustrative, not as explanatory.

(74) In speaking of the Energy of the field, however, I wish to be under-
stood literally. All energy is the same as mechanical energy, whether it exists
in the form of motion or in that of elasticity, or in any other form. The
energy in electromagnetic phenomena is mechanical energy. The only question
is, Where does it reside ? On the old theories it resides in the electrified bodies,
conducting circuits, and magnets, in the form of an unknown quality called
potential energy, or the power of producing certain effects at a distance. On
our theory it resides in the electromagnetic field, in the space surrounding the
electrified and magnetic bodies, as well as in those bodies themselves, and is
in two different forms, which may be described without hypothesis as magnetic
polarization and electric polarization, or, according to a very probable hypothesis,
as the motion and the strain of one and the same medium.

(75) The conclusions arrived at in the present paper are independent of
this hypothesis, being deduced from experimental facts of three kinds :

1. The induction of electric currents by the increase or diminution of
neighbouring currents according to the changes in the lines of force passing
through the circuit.

2. The distribution of magnetic intensity according to the variations of a
magnetic potential.

3. The induction (or influence) of statical electricity through dielectrics.

We may now proceed to demonstrate from these principles the existence
and laws of the mechanical forces which act upon electric currents, magnets, and
electrified bodies placed in the electromagnetic field.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 565

PART IV.

MECHANICAL ACTIONS IN THE FIELD.

Mechanical Force on a Moveable Conductor.

(76) We have shewn (§§ 34 & 35) that the work done by the electro-

magnetic forces in aiding the motion of a conductor is equal to the product

of the current in the conductor multiplied by the increment of the electro-
magnetic momentum due to the motion.

Let a short straight conductor of length a move parallel to itself in the
direction of x, with its extremities on two parallel conductors. Then the incre-
ment of the electromagnetic momentum due to the motion of a will be

fdFdx dGdy dHdz\ g
\dx ds dx ds dx ds)

That due to the lengthening of the circuit by increasing the length of the
parallel conductors will be

fdFdx dFdy dF_dz\8
\dx ds dy ds dz ds)

The total increment is

8 Wf/1^ «M% */d^ 'dH\\

\ds\dx dy) ds\dz dx ) J '

which is by the equations of Magnetic Force (B), p. 556,

dz

Let X be the force acting along the direction of x per unit of length of
the conductor, then the work done is XaSx.

Let C be the current in the conductor, and let p', q, r' be its com-
ponents, then

XaSx = CaSx (-^ py - ~ i

566 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

or

Similarly, Y= par' — /iyp'

(J).

These are the equations which determine the mechanical force acting on a
conductor carrying a current. The force is perpendicular to the current and
to the lines of force, and is measured by the area of the parallelogram formed
by lines parallel to the current and lines of force, and proportional to their
intensities.

Mechanical Force on a Magnet.

(77) In any part of the field not traversed by electric currents the dis-
tribution of magnetic intensity may be represented by the differential coefficients
of a function which may be called the magnetic potential. When there are no
currents in the field, this quantity has a single value for each point. When
there are currents, the potential has a series of values at each point, but its
differential coefficients have only one value, namely,

dd> d4> dd>

-r-=a, -j- = p, -f = y-
dx dy dz r

Substituting these values of a, $, y in the expression (equation 38) for the
intrinsic energy of the field, and integrating by parts, it becomes

+ +

dy dz

The expression 2 + - + - dV^mdV .. .. (39)

\ dx dy dz /

indicates the number of lines of magnetic force which have their origin within
the space V. Now a magnetic pole is known to us only as the origin or
termination of lines of magnetic force, and a unit pole is one which has 4n-
lines belonging to it, since it produces unit of magnetic intensity at unit of
distance over a sphere whose surface is 47r.

Hence if m is the amount of free positive magnetism in unit of volume,
the above expression may be written i-rrm, and the expression for the energy
of the field becomes

(40).

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 567

If there are two magnetic poles m^ and m.2 producing potentials ^ and <£2
in the field, then if m3 is moved a distance dx, and is urged in that direction
by a force X, then the work done is Xdx, and the decrease of energy in the
field is

and these must be equal by the principle of Conservation of Energy.

Since the distribution ^ is determined by ml} and <£, by ma, the quantities
and ^am2 will remain constant.

It can be shewn also, as Green has proved (Essay, p. 10), that

so that we get
or

-

where c^ represents the magnetic intensity due to mr

(K).

Similarly, Y= mJ3lt

Z = m3y1.

So that a magnetic pole is urged in the direction of the lines of magnetic
force with a force equal to the product of the strength of the pole and the
magnetic intensity.

(78) If a single magnetic pole, that is, one pole of a veiy long magnet,
be placed in the field, the only solution of <f> is

*=-~f (41)>

where ml is the strength of the pole, and r the distance from it.
The repulsion between two poles of strength in\ and mt is

(42).

In air or any medium in which \L = 1 this is simply -^ , but in other

media the force acting between two given magnetic poles is inversely propor-
tional to the coefficient of magnetic induction for the medium. This may be
explained by the magnetization of the medium induced by the action of the
poles.

568 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

Mechanical Force on an Electrified Body.

(79) If there is no motion or change of strength of currents or magnets
in the field, the electromotive force is entirely due to variation of electric
potential, and we shall have (§65)

~dx' V~

Integrating by parts the expression (I) for the energy due to electric
displacement, and remembering that P, Q, R vanish at an infinite distance, it becomes

or by the equation of Free Electricity (G), p. 561,

By the same demonstration as was used in the case of the mechanical
action on a magnet, it may be shewn that the mechanical force on a small
body containing a quantity et of free electricity placed in a field whose
potential arising from other electrified bodies is ¥,, has for components

(D).

' dz~

So that an electrified body is urged in the direction of the electromotive
force with a force equal to the product of the quantity of free electricity and
the electromotive force.

If the electrification of the field arises from the presence of a small
electrified body containing el of free electricity, the only solution of ¥, is

*.-££ <43>'

where r is the distance from the electrified body.

The repulsion between two electrified bodies elt et is therefore

(. ? = _•_' (44)

* dr 47r r1 '

A DYNAMICAL THEOEY OF THE ELECTROMAGNETIC FIELD. 569

Measurement of Electrostatic Effects.

(80) The quantities with which we have had to do have been hitherto
expressed in terms of the Electromagnetic System of measurement, which is
founded on the mechanical action between currents. The electrostatic system of
measurement is founded on the mechanical action between electrified bodies,
and is independent of, and incompatible with, the electromagnetic system ; so
that the units of the different kinds of quantity have different values according
to the system we adopt, and to pass from the one system to the other, a
reduction of all the quantities is required.

According to the electrostatic system, the repulsion between two small
bodies charged with quantities fy, 77, of electricity is

where r is the distance between them.

Let the relation of the two systems be such that one electromagnetic unit
of electricity contains v electrostatic units; then ijl = vel and ty, = vea, and this
repulsion becomes

eP^.A^! by equation (44) .................. (45),

whence k, the coefficient of "electric elasticity" in the medium in which the
experiments are made, i. e. common air, is related to v, the number of electro-
static units in one electromagnetic unit, by the equation

k = 4vv* .................................... (46).

The quantity v may be determined by experiment in several ways. Ac-
cording to the experiments of MM. Weber and Kohlrausch,

v = 310,740,000 metres per second.

(81) It appears from this investigation, that if we assume that the medium
which constitutes the electromagnetic field is, when dielectric, capable of receiving
in every part of it an electric polarization, in which the opposite sides of every
element into which we may conceive the medium divided are oppositely elec-
trified, and if we also assume that this polarization or electric displacement is
proportional to the electromotive force which produces or maintains it, then we

VOL. I. 72

570 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

can shew that electrified bodies in a dielectric medium will act on one another
with forces obeying the same laws as are established by experiment.

The energy, by the expenditure of which electrical attractions and repul-
sions are produced, we suppose to be stored up in the dielectric medium which
surrounds the electrified bodies, and not on the surface of those bodies them-
selves, which on our theory are merely the bounding surfaces of the air or other
dielectric in which the true springs of action are to be sought.

Note on the Attraction of Gravitation.

(82) After tracing to the action of the surrounding medium both the
magnetic and the electric attractions and repulsions, and finding them to depend
on the inverse square of the distance, we are naturally led to inquire whether
the attraction of gravitation, which follows the same law of the distance, is
not also traceable to the action of a surrounding medium.

Gravitation differs from magnetism and electricity in this ; that the bodies
concerned are all of the same kind, instead of being of opposite signs, like
magnetic poles and electrified bodies, and that the force between these bodies
is an attraction and not a repulsion, as is the case between like electric and
magnetic bodies.

The lines of gravitating force near two dense bodies are exactly of the
same form as the lines of magnetic force near two poles of the same name ;
but whereas the poles are repelled, the bodies are attracted. Let E be the
intrinsic energy of the field surrounding two gravitating bodies Mlt Mtl and
let E* be the intrinsic energy of the field surrounding two magnetic poles,
TO,, m^ equal in numerical value to M0 Hv and let X be the gravitating
force acting during the displacement Sx, and X' the magnetic force,

now X and X' are equal in numerical value, but of opposite signs ; so that

8E= -8R,
or E=C-E'

where a, ft, y are the components of magnetic intensity. If R be the resultant

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 571

gravitating force, and R the resultant magnetic force at a corresponding part
of the field,

R= -R, and 0:
Hence

07T

(47).

The intrinsic energy of the field of gravitation must therefore be less where-
ever there is a resultant gravitating force.

As energy is essentially positive, it is impossible for any part of space to
have negative intrinsic energy. Hence those parts of space in which there is
no resultant force, such as the points of equilibrium in >the space between the
different bodies of a system, and within the substance of each body, must have
an intrinsic energy per unit of volume greater than

where R is the greatest possible value of the intensity of gravitating force in
any part of the universe.

The assumption, therefore, that gravitation arises from the action of the
surrounding medium in the way pointed out, leads to the conclusion that every
part of this medium possesses, when undisturbed, an enormous intrinsic energy,
and that the presence of dense bodies influences the medium so as to diminish
this energy wherever there is a resultant attraction.

As I am unable to understand in what way a medium can possess such
properties, I cannot go any further in this direction in searching for the cause
of gravitation.

72—2

572 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

PART V.

THEORY OF CONDENSERS.

Capacity of a Condenser.

(83) The simplest form of condenser consists of a uniform layer of insulating
matter bounded by two conducting surfaces, and its capacity is measured by the
quantity of electricity on either surface when the difference of potentials is unity.

Let S be the area of either surface, a the thickness of the dielectric, and
k its coefficient of electric elasticity; then on one side of the condenser the
potential is ¥„ and on the other side ¥t+l, and within its substance

<«>

Since -j— and therefore f is zero outside the condenser, the quantity of electricity
on its first surface = — Sf, and on the second +Sf. The capacity of the con-
denser is therefore Sf=-r in electromagnetic measure.

Specific Capacity of Electric Induction (D).
(84) If the dielectric of the condenser be air, then its capacity in electro-

o

static measure is — (neglecting corrections arising from the conditions to be

fulfilled at the edges). If the dielectric have a capacity whose ratio to that of

DS

air is D, then the capacity of the condenser will be - - .

4 Trot

Hence D-i (49),

where kt is the value of k in air, which is taken for unity.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 573

Electric Absorption.

(85) When the dielectric of which the condenser is formed is not a perfect
insulator, the phenomena of conduction are combined with those of electric dis-
placement. The condenser, when left charged, gradually loses its charge, and in
some cases, after being discharged completely, it gradually acquires a new charge
of the same sign as the original charge, and this finally disappears. These
phenomena have been described by Professor Faraday (Experimental Researches,
Series XL) and by Mr F. Jenkin (Report of Committee of Board of Trade on
Submarine Cables), and may be classed under the name of " Electric Absorption."

/

(86) We shall take the case of a condenser composed of any number of

parallel layers of different materials. If a constant difference of potentials between
its extreme surfaces is kept up for a sufficient tune till a condition of perma-
nent steady flow of electricity is established, then each bounding surface will
have a charge of electricity depending on the nature of the substances on each
side of it. If the extreme surfaces be now discharged, these internal charges
will gradually be dissipated, and a certain charge may reappear on the extreme
surfaces if they are insulated, or, if they are connected by a conductor, a certain
quantity of electricity may be urged through the conductor during the re-
establishment of equilibrium.

Let the thickness of the several layers of the condenser be aa a., &c.
Let the values of k for these layers be respectively ku ka, kt, and let

ajct + ajct + &c. =ak (50),

where k is the "electric elasticity" of air, and a is the thickness of an equiva-
lent condenser of air.

Let the resistances of the layers be respectively ru ri} &c., and let
r1 + rt + &c. = r be the resistance of the whole condenser, to a steady current
through it per unit of surface.

Let the electric displacement in each layer be fu fv &c.
Let the electric current in each layer be pu pv &c.

Let the potential on the first surface be ¥„ and the electricity per unit
of surface e,.

-.74

A DYNAMICAL THEORY OP THE ELECTROMAGNETIC FIELD.

Let the corresponding quantities at the boundary of the first and second
surface be V, and eu and so on. Then by equations (G) and (H),

^1 = — Jit ~jTf — ~Pv

•(51),

&c.
But by equations (E) and (F),

¥,-¥, = «,£,/,= -r.fr

(52)'
&c. &c. &c. J

After the electromotive force has been kept up for a sufficient time the
current becomes the same in each layer, and

T

where ¥ is the total difference of potentials between the extreme layers. We
have then

yj = ^ , ft= ^- 1 &c.

J * C*i»Vj i Utt/v* / f n\

and I (53).

ei ~ r • et= — ( ~~r r ) > &c-

These are the quantities of electricity on the different surfaces.

(87) Now let the condenser be discharged by connecting the extreme surfaces
through a perfect conductor so that their potentials are instantly rendered equal,
then the electricity on the extreme surfaces will be altered, but that on the
internal surfaces will not have time to escape. The total difference of potentials
is now

whence if e', is what et becomes at the instant of discharge,

e', = - -^- - -j~ei--T (55).

r a,k, ok ak

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 575

The instantaneous discharge is therefore — p or the quantity which would

be discharged by a condenser of air of the equivalent thickness a, and it is
unaffected by the want of perfect insulation.

(88) Now let us suppose the connexion between the extreme surfaces

broken, and the condenser left to itself, and let us consider the gradual dissi-

pation of the internal charges. Let ¥' be the difference of potential of the
extreme surfaces at any time t ; then

(56);
but

Hence fl = Ate r- , ft = Af, r* ', &c. ; and by referring to the values of e\, e,,
&c., we find

1

_ I

1 ~~ r a fa ak

(57),

>Tf ™ \ff

A, = \ r

r ajct ak

&c.
so that we find for the difference of extreme potentials at any time,

— r? }e r* + &c (58).

r ak )

(89) It appears from this result that if all the layers are made of the
same substance, ¥* will be zero always. If they are of different substances,
the order in which they are placed is indifferent, and the effect will be the
same whether each substance consists of one layer, or is divided into any number
of thin layers and arranged in any order among thin layers of the other sub-
stances. Any substance, therefore, the parts of which are not mathematically
homogeneous, though they may be apparently so, may exhibit phenomena of
absorption. Also, since the order of magnitude of the coefficients is the same
as that of the indices, the value of ¥' can never change sign, but must start
from zero, become positive, and finally disappear.

576 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

(90) Let us next consider the total amount of electricity which would
pass from the first surface to the second, if the condenser, after being thoroughly
saturated by the current and then discharged, has its extreme surfaces connected
by a conductor of resistance R. Let p be the current in this conductor ; then,
during the discharge,

(59).

Integrating with respect to the time, and calling qlt qn q the quantities of
electricity which traverse the different conductors,

ff,r1 + 5trt + &c.-g£ ............................... (60).

The quantities of electricity on the several surfaces will be

&c. ;
and since at last all these quantities vanish, we find

q1 = e\ - q,

•\|r / if» j> \ -qrr

whence qR = - — ,' H -- r + &c. -- T,

r\a,&, ajc, ] ok

or q

a quantity essentially positive; so that, when the primary electrification is in
one direction, the secondary discharge is always in the same direction as the
primary discharge *.

* Since this paper was communicated to the Royal Society, I have seen a paper by M. Gaugain
in the Annales de Chimie for 1864, in which he has deduced the phenomena of electric absorption and
secondary discharge from the theory of compound condensers.

A DYNAMICAL THEORY OP THE ELECTROMAGNETIC FIELD. 577

PART VI.

ELECTROMAUNETIC THEORY OF LIGHT.

(91) At the commencement of this paper we made use of the optical
hypothesis of an elastic medium through which the vibrations of light are
propagated, in order to shew that we have warrantable grounds for seeking,
in the same medium, the cause of other phenomena as well as those of light.
We then examined electromagnetic phenomena, seeking for their explanation in
the properties of the field which surrounds the electrified or magnetic bodies.
In this way we arrived at certain equations expressing certain properties of
the electromagnetic field. We now proceed to investigate whether these pro-
perties of that which constitutes the electromagnetic field, deduced from electro-
magnetic phenomena alone, are sufficient to explain the propagation of light
through the same substance.

(92) Let us suppose that a plane wave whose direction cosines are I, m, n
is propagated through the field with a velocity V. Then all the electro-
magnetic functions will be functions of

z— Vt.

The equations of Magnetic Force (B), p. 556, will become

dH dG
U.CL = m — j -- n -; — ,
aw aw

B=n — -l-—

^ dw dw '

__
' " dw dw '

If we multiply these equations respectively by I, m, n, and add, we find

lfia + m^/3 + niJ.y = Q ............................... (62),

which shews that the direction of the magnetization must be in the plane of
the wave.

VOL. I. 73

578

A DYNAMICAL THEORY OP THE ELECTROMAGNETIC FIELD.

(93) If we combine the equations of Magnetic Force (B) with those of
Electric Currents (C), and put for brevity

dF dG dH^j d ^_..dL,d^= .
dx dy dz ~ dx* dy* dz*

(63),

(64).

If the medium in the field is a perfect dielectric there is no true conduction,
and the currents p, q', r are only variations in the electric displacement, or,
by the equations of Total Currents (A),

, df , dq . dh

*-%• *~% r=Tt <65>-

But these electric displacements are caused by electromotive forces, and by the
equations of Electric Elasticity (E),

P = kf, Q = kg, R = Tch (66).

These electromotive forces are due to the variations either of the electro-
magnetic or the electrostatic functions, as there is no motion of conductors in
the field ; so that the equations of electromotive force (D) are

_

'It dx

dG
~dt

~dy

dH

s

dt

1 —

dz

(67).

(94) Combining these equations, we obtain the following : —

.(68).

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

579

If we differentiate the third of these equations with respect to y, and
the second with respect to z, and subtract, J and ¥ disappear, and by remem-
bering the equations (B) of magnetic force, the results may be written

*

d*

(69).

(95) If we assume that a, /3, y are functions of Ix-j-my + nz- Vt = w, the
first equation becomes

d'a

or

, a a ,,,MO.

"Ptt^^V-V^

1

(70),

F=±

(71).

The other equations give the same value for V, so that the wave is propa-
gated in either direction with a velocity V.

This wave consists entirely of magnetic disturbances, the direction of mag-
netization being in the plane of the wave. No magnetic disturbance whose
direction of magnetization is not in the plane of the wave can be propagated
as a plane wave at all.

Hence magnetic disturbances propagated through the electromagnetic field
agree with light in this, that the disturbance at any point is transverse to
the direction of propagation, and such waves may have all the properties of
polarized light.

(96) The only medium in which experiments have been made to determine
the value of k is air, in which /*=!, and therefore, by equation (46),

V=v (72).

By the electromagnetic experiments of MM. Weber and Kohlrausch *,
v = 310,740,000 metres per second

* Leipzig Transactions, Vol. v. (1857), p. 260, or PoggendoriFs Annalen, Aug. 1856, p. 10.

73—2

580 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

is the number of electrostatic units in one electromagnetic unit of electricity,
and this, according to our result, should be equal to the velocity of light in
air or vacuum.

The velocity of light in air, by M. Fizeau's * experiments, is

7=314,858,000;

according to the more accurate experiments of M. Foucaultt,

F= 298,000,000.

The velocity of light in the space surrounding the earth, deduced from
the coefficient of aberration and the received value of the radius of the earth's
orbit, is

V= 308,000,000.

(97) Hence the velocity of light deduced from experiment agrees sufficiently
well with the value of v deduced from the only set of experiments we as yet
possess. The value of v was determined by measuring the electromotive force
with which a condenser of known capacity was charged, and then discharging
the condenser through a galvanometer, so as to measure the quantity of electricity
in it in electromagnetic measure. The only use made of light in the experiment
was to see the instruments. The value of V found by M. Foucault was
obtained by determining the angle through which a revolving mirror turned,
while the light reflected from it went and returned along a measured course.
No use whatever was made of electricity or magnetism.

The agreement of the results seems to shew that light and magnetism
are affections of the same substance, and that light is an electromagnetic dis-
turbance propagated through the field according to electromagnetic laws.

(98) Let us now go back upon the equations in (94), in which the
quantities J and "V occur, to see whether any other kind of disturbance can
be propagated through the medium depending on these quantities which disappeared
from the final equations.

* Comfits Rendiu, Vol. MIX. (1849), p. 90.
t Ibid. Vol. LV. (1862), pp. 501, 792.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

If we determine x from the equation

581

dz'
and F', G', H' from the equations

F' = F_dx Q' = Q^^K H' = H-^ (74)

dx ' dy' dz"

then

dx dy dz
and the equations in (94) become of the form

Differentiating the three equations with respect to x, y, and z, and adding, we
find that

dx

dt ' '•••\''/>

and that kV*F' = 4irp. —^-

(78).

Hence the disturbances indicated by F', G', H' are propagated with the velocity

F= /- - through the field; and since

dF' dG' dH'

-T- + -5-+-5— = 0,
dx dy dz

the resultant of these disturbances is in the plane of the wave.

(99) The remaining part of the total disturbances F, G, II being the part
depending on ^, is subject to no condition except that expressed in the equation

_
dt * dt>~

582 A DYNAMICAL THEORY OF TUB ELECTROMAGNETIC FIELD.

If we perform the operation V* on this equation, it becomes

Since the medium is a perfect insulator, e, the free electricity, is immove-
able, and therefore -j- is a function of x, y, z, and the value of J is either

constant or zero, or uniformly increasing or diminishing with the time ; so that
no disturbance depending on J can be propagated as a wave.

(100) The equations of the electromagnetic field, deduced from purely
experimental evidence, shew that transversal vibrations only can be propagated.
If .we were to go beyond our experimental knowledge and to assign a definite
density to a substance which we should call the electric fluid, and select either
vitreous or resinous electricity as the representative of that fluid, then we might
have normal vibrations propagated with a velocity depending on this density.
We have, however, no evidence as to the density of electricity, as we do not
even know whether to consider vitreous electricity as a substance or as the
absence of a substance.

Hence electromagnetic science leads to exactly the same conclusions as
optical science with respect to the direction of the disturbances which can be
propagated through the field ; both affirm the propagation of transverse vibra-
tions, and both give the same velocity of propagation. On the other hand, both
sciences are at a loss when called on to affirm or deny the existence of normal
vibrations.

Relation between the Index of Refraction and the Electromagnetic Character

of the substance.

(101) The velocity of light in a medium, according to the Undulatory
Theory, is

Iv

i y"

where i is the index of refraction and Vt is the velocity in vacuum. The
velocity, according to the Electromagnetic Theory, is

T

47T/t'

where, by equations (49) and (71), £ = -g£0, and &0 =

A DYNAMICAL THEORY OP THE ELECTROMAGNETIC FIELD. 583

Hence D = - (80),

P-

or the Specific Inductive Capacity is equal to the square of the index of refrac-
tion divided by the coefficient of magnetic induction.

Propagation of Electromagnetic Disturbances in a Crystallized Medium.

(102) Let us now calculate the conditions of propagation of a plane wave
in a medium for which the values of k and /u, are different in different direc-
tions. As we do not propose to give a complete investigation of the question
in the present imperfect state of the theory as extended to disturbances of
short period, we shall assume that the axes of magnetic induction coincide in
direction with those of electric elasticity.

(103) Let the values of the magnetic coefficient for the three axes be
X, fi, v, then the equations of magnetic force (B) become

. dH dG

\OL=-, -j-

dy dz

dF dH
~fa~^

dG dF

(81).

• dx dy

The equations of electric currents (C) remain as before.
The equations of electric elasticity (E) will be

................................. (82),

R = 4ircth\

where 4nu5, 4^6', and 4?rcJ are the values of k for the axes of x, y, z.

Combining these equations with (A) and (D), we get equations of the form

1 /. <W? d'F^ d*F\ 1 d
~'V~~~

^

* '

__^
dz ) ~ a\ dt* + dxdt)"'(^

(104) If /, m, n are the direction-cosines of the wave, and V its velocity,

and if

Ix + my + nz— Vt = w .............................. (84),

584

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

then F, G, H, and ¥ will be functions of w; and if we put Ff, G', H', V
for the second differentials of these quantities with respect to w, the equations
will be

' + — G' + a^ H'-IVV = 0

(85).

X- • | .•. "- ' — » \ /

v

?' , c'nm{

M * + •

If we now put

If , V

oW/P m' n'\ (86)'

we shall find

f W \J "^ V3L V \J ^~ V ........«..«».. ..P. .......... \O/ /•

\ /*

with two similar equations for G' and //'. Hence either

F = 0 (88),

U=0 (89),

or

VF' = lfr, VG'^m^ and VH' = n^ (90).

The third supposition indicates that the resultant of F', G', H' is in the
direction normal to the plane of the wave ; but the equations do not indicate
that such a disturbance, if possible, could be propagated, as we have no other
relation between ^ and F', G', H'.

The solution F=0 refers to a case in which there is no propagation.

* The solution U= 0 gives two values for F' corresponding to values of F',
G', H', which are given by the equations

a' *"b* <? "' ''

f>*7ji*

•-a'X) + ^,(a'X-&» = 0 (92).

* [Although it is not expressly stated in the text it should be noticed that in finding equations
(91) and (92) the quantity *' is put equal to zero. See § 98 and also the corresponding treat-
ment of this subject in the Electricity and Magnetism il § 796. It may be observed that the

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

(105) The velocities along the axes are as follows : —
Direction of propagation

585

Direction of the electric displacements

X

y

%

a2

a2

X

—

—

V

P-

b1

V

y

V

X

c2

c2

z

*/

X

Now we know that in each principal plane of a crystal the ray polarized
in that plane obeys the ordinary law of refraction, and therefore its velocity
is the same in whatever direction in that plane it is propagated.

If polarized light consists of electromagnetic disturbances in which the
electric displacement is in the plane of polarization, then

a* = b* = c' (93).

If, on the contrary, the electric displacements are perpendicular to the plane

of polarization,

\ = fi = v (94).

We know, from the magnetic experiments of Faraday, Plucker, &c., that in
many crystals X, ft, v are unequal.

equations referred to and the table given in § 105 may perhaps be more readily understood from
a different mode of elimination. If we write

' + vn' =

it is readily seen that

and \lf" + ft.mG' + vnJI ' =
7V*'-a'\Q

" V - a'\P '

with similar expressions for G', II'. From these we readily obtain by reasoning similar to that in
§ 104, the equation corresponding to (86),' viz. :

tn.fi nv

i • -wra i~o »-* ~r

P-a*APT V'-b'fiP ' V'-c?vP~
This form of the equation agrees with that given in the Electricity and Magnetism ii. § 797.

By means of this equation the equations (91) and (92) readily follow when *' = 0. The
ratios of F' : G' : H' for any direction of propagation may also be determined.]

VOL. I. 74

586 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

The experiments of Knoblauch * on electric induction through crystals seem
to shew that a, b and c may be different.

The inequality, however, of X, p., v is so small that great magnetic forces
are required to indicate their difference, and the differences do not seem of
sufficient magnitude to account for the double refraction of the crystals.

On the other hand, experiments on electric induction are liable to error
on account of minute flaws, or portions of conducting matter in the crystal.

Further experiments on the magnetic and dielectric properties of crystals
are required before we can decide whether the relation of these bodies to mag-
netic and electric forces is the same, when these forces are permanent as when
they are alternating with the rapidity of the vibrations of light.

Relation between Electric Resistance and Transparency.

(106) If the medium, instead of being a perfect insulator, is a conductor
whose resistance per unit of volume is p, then there will be not only electric
displacements, but true currents of conduction in which electrical energy is
transformed into heat, and the undulation is thereby weakened. To determine
the coefficient of absorption, let us investigate the propagation along the axis
of x of the transverse disturbance G.

By the former equations

by (A),

dtf~

If G is of the form

G = €.->** COB (qx + nt) (96),

we find that

27^n_2^F

• ......--••-•-.................... I »' i I.

p q pi

where V is the velocity of light in air, and i is the index of refraction. The
proportion of incident light transmitted through the thickness x is

«'*• (98)-

* Philosophical Magazine, 1852.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 587

Let R be the resistance in electromagnetic measure of a plate of the
substance whose thickness is x, breadth b, and length /, then

R-l£
~bx'

(107) Most transparent solid bodies are good insulators, whereas all good
conductors are very opaque.

Electrolytes allow a current to pass easily and yet are often very trans-
parent. We may suppose, however, that in the rapidly alternating vibrations
of light, the electromotive forces act for so short a time that they are unable to
effect a complete separation between the particles in combination, so that when
the force is reversed the particles oscillate into their former position without
loss of energy.

Gold, silver, and platinum are good conductors, and yet when reduced to
sufficiently thin plates they allow light to pass through them. If the resistance
of gold is the same for electromotive forces of short period as for those with
which we make experiments, the amount of light which passes through a piece
of gold-leaf, of which the resistance was determined by Mr C. Hockin, would
be only 10~w of the incident light, a totally imperceptible quantity. I find that
between -g J7 and 10100 of green light gets through such gold-leaf. Much of this
is transmitted through holes and cracks ; there is enough, however, transmitted
through the gold itself to give a strong green hue to the transmitted light.
This result cannot be reconciled with the electromagnetic theory of light, unless
we suppose that there is less loss of energy when the electromotive forces are
reversed with the rapidity of the vibrations of light than when they act for
sensible times, as in our experiments.

Absolute Valves of the Electromotive and Magnetic Forces called into play in the

Propagation of Light.

(108) If the equation of propagation of light is

9_

F=Aco9~(z-Vt),

A

74—2

588 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

the electromotive force will be

and the energy per unit of volume will be

where P represents the greatest value of the electromotive force. Half of this
consists of magnetic and half of electric energy.

The energy passing through a unit of area is

P1

W—
~

so that

where V is the velocity of light, and W is the energy communicated to unit
of area by the light in a second.

According to Pouillet's data, as calculated by Professor W. Thomson*, the
mechanical value of direct sunlight at the Earth is

8 3 '4 foot-pounds per second per square foot.

This gives the maximum value of P in direct sunlight at the Earth's distance

from the Sun,

P = 60,000,000,

or about 600 Daniell's cells per metre.

At the Sun's surface the value of P would be about

13,000 Daniell's cells per metre.

At the Earth the maximum magnetic force would be '193t.
At the Sun it would be 4'13.

These electromotive and magnetic forces must be conceived to be reversed
twice in every vibration of light ; that is, more than a thousand million million
times in a second.

* Transactions of MM Royal Society of Edinburgh, 1854 ("Mechanical Energies of the Solar
System ">

t The horizontal magnetic force at Kew is about 1*76 in metrical units.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 589

PART VII.
CALCULATION OF THE COEFFICIENTS OF ELECTROMAGNETIC INDUCTION.

General Methods.

(109) The electromagnetic relations between two conducting circuits, A and
B, depend upon a function M of their form and relative position, as has been
already shewn.

M may be calculated in several different ways, which must of course all
lead to the same result.

First Method. M is the electromagnetic momentum of the circuit B when
A carries a unit current, or

where F, G, H are the components of electromagnetic momentum due to a unit
current in A, and ds' is an element of length of B, and the integration is
performed round the circuit of B.

To find F, G, H, we observe that by (B) and (C)

+ i + ~ = ~ *P'

with corresponding equations for G and H, p, q', and r' being the components
of the current in A.

Now if we consider only a single element ds of A, we shall have

, dx , , dy , , dz ,

P=Tsd8' 1=fsds> r=dsds'

and the solution of the equation gives

„ u.dx , _, a dy , TT u. dz ,

F = '--rds, G=--r-ds, H = "-j-ds,

p ds p ds pas

590 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

where p is the distance of any point from ds. Hence

dz

where 0 is the angle between the directions of the two elements ds, (/*', and
p is the distance between them, and the integration is performed round both
circuits.

In this method we confine our attention during integration to the two linear
circuits alone.

(110) Second Method. M is the number of lines of magnetic force which
pass through the circuit B when A carries a unit current, or

M = 2 (pal + pfim + pyn) d&,

where pa, p/3, py are the components of magnetic induction due to unit current
in A, S' is a surface bounded by the current B, and I, m, n are the direction-
cosines of the normal to the surface, the integration being extended over the
surface.

We may express this in the form

M= /i2 -t sin e sin ff sin ^dS'ds,

where dS' is an element of the surface bounded by B, ds is an element
of the circuit A, p is the distance between them, 6 and ff are the angles
between p and ds and between p and the normal to dS? respectively, and <f> is
the angle between the planes in which 6 and ff are measured. The integration

O 1 O

is performed round the circuit A and over the surface bounded by B.

This method is most convenient in the case of circuits lying in one plane,
in which case sin 0=1, and sin^=l.

(111) Third Method. M is that part of the intrinsic magnetic energy of
the whole field which depends on the product of the currents in the two
circuits, each current being unity.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 591

Let a, /8, y be the components of magnetic intensity at any point due to
the first circuit, a, (?, y the same for the second circuit ; then the intrinsic
energy of the element of volume dV of the field is

The part which depends on the product of the currents is

Hence if we know the magnetic intensities / and /' due to the unit current
in each circuit, we may obtain M by integrating

47T

over all space, where 6 is the angle between the directions of / and /'.

Application to a Coil.

(112) To find the coefficient (M) of mutual induction between two circular
linear conductors in parallel planes, the distance between the curves being every-
where the same, and small compared with the radius of either.

If r be the distance between the curves, and a the radius of either, then
when r is very small compared with a, we find by the second method, as a
first approximation,

To approximate more closely to the value of M, let a and a, be the radii of
the circles, and 6 the distance between their planes; then

We obtain M by considering the following conditions: — •
1st. M must fulfil the differential equation

=Q
da- db* a da

This equation being true for any magnetic field symmetrical with respect to the
common axis of the circles, cannot of itself lead to the determination of M as
a function of a, a,, and 6. We therefore make use of other conditions.

592 A DYNAMICAL T11EOKY OF THE ELECTROMAGNETIC FIELD.

•2ndly. The value of M must remain the same when a and a, are exchanged.
3rdly. The first two terms of M must be the same as those given above.
M may thus be expanded in the following series : —

(113) We may apply this result to find the coefficient of self-induction
(L) of a circular coil of wire whose section is small compared with the radius
of the circle.

Let the section of the coil be a rectangle, the breadth in the plane of
the circle being c, and the depth perpendicular to the plane of the circle being 6.

Let the mean radius of the coil be a, and the number of windings n ;
then we find, by integrating,

X = r— I M (xy x'y) dx dy dx' dy",

where M(xy x'y'} means the value of M for the two windings whose coordinates
are xy and x'y' respectively; and the integration is performed first with respect
to x and y over the rectangular section, and then with respect to x' and y
over the same space.

L = 4mi'a {log.— + -5j - \ ( 0 - 7) cot 26 - ^cos 20 - Jcot'fllogcos 6 - Jtan'fllogsin

T I — u \ *±J o O o

Here a= mean radius of the coiL
„ r= diagonal of the rectangular section = \/6* + c'.
,, 0= angle between r and the plane of the circle.
,, n= number of windings.

The logarithms are Napierian, and the angles are in circular measure.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 593

In the experiments made by the Committee of the British Association for
determining a standard of Electrical Resistance, a double coil was used, con-
sisting of two nearly equal coils of rectangular section, placed parallel to each
other, with a small interval between them.

The value of L for this coil was "found in the following way.

The value of L was calculated by the preceding formula for six different
cases, in which the rectangular section considered has always the same breadth,
while the depth was

A, B, C, A + B, B+C, A+B+C,
and n = I in each case.

Calling the results L(A), L (B), L(C), &c.,

we calculate the coefficient of mutual induction M(AO) of the two coils thus,
2ACM (AC] = (A +B+ C?L(A +B + C)-(A + B}*L

Then if n, is the number of windings in the coil A and «, in the coil C, the
coefficient of self-induction of the two coils together is

L = n?L (A ) + Zr^njtf (AC) + < L ( C) .

(114) These values of L are calculated on the supposition that the windings
of the wire are evenly distributed so as to fill up exactly the whole section.
This, however, is not the case, as the wire is generally circular and covered with
insulating material. Hence the current in the wire is more concentrated than it
would have been if it had been distributed uniformly over the section, and the
currents in the neighbouring wires do not act on it exactly as such a uniform
current would do.

The corrections arising from these considerations may be expressed as nu-
merical quantities, by which we must multiply the length of the wire, and they
are the same whatever be the form of the coil.

Let the distance between each wire and the next, on the supposition that
they are arranged in square order, be D, and let the diameter of the wire
be d, then the correction for diameter of wire is

VOL. I. 75

594 A DYNAMICAL THEORY OP THE ELECTROMAGNETIC FIELD.

The correction for the eight nearest wires is

+ 0-0236.
For the sixteen in the next row + 0'00083.

These corrections being multiplied by the length of wire and added to the
former result, give the true value of L, considered as the measure of the
potential of the coil on itself for unit current in the wire when that current
has been established for some time, and is uniformly distributed through the
section of the wire.

(115) But at the commencement of a current and during its variation the
current is not uniform throughout the section of the wire, because the induc-
tive action between different portions of the current tends to make the current
stronger at one part of the section than at another. When a uniform electro-
motive force P arising from any cause acts on a cylindrical wire of specific

resistance p, we have

dF

where F is got from the equation

d*F 1 dF

r being the distance from the axis of the cylinder.

Let one term of the value of F be of the form Trn, where T is a
function of the time, then the term of p which produced it is of the form

4 TT -p.
Hence if we write

p I 1* . 2' ofa1
' 1 d'T

The total counter current of self-induction at any point is
IP \ 1 , . air cW . . uV 1

P

f

from ( = 0 to t = oo .

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 595

When t = 0, p = 0, .: = P, = o, &c.

t /„

rrom £ = 0 to = oo .

Wlien t = 0, p = 0 throughout the section, .-. (*—) = P, K^ =0, &c.

\ dt /o \ dt /o

When<=oo,_p = 0 „ „ „ .-. K =0, t =0,&c.

Also if Z be the length of the wire, and R its resistance,

and if C be the current when established in the wire, C=-fr.

M

The total counter current may be written
7 7

Now if the current instead of being variable from the centre to the cir-
cumference of the section of the wire had been the same throughout, the value
of F would have been

where y is the current in the wire at any instant, and the total counter
current would have been

/7

J 0 J 0

Hence L = U —

or the value of L which must be used in calculating the self-induction of a
wire for variable currents is less than that which is deduced from the suppo-
sition of the current being constant throughout the section of the wire by

75—2

596 A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD.

where / is the length of the wire, and p. is the coefficient of magnetic induc-
tion for the substance of the wire.

(116) The dimensions of the coil used by the Committee of the British
Association in their experiments at King's College in 1864 were as follows: —

metro.

Mean radius =a= '158194

Depth of each coil =& = -oi608

Breadth of each coil = c = '01841

Distance between the coils ='02010

Number of windings »= 313

Diameter of wire ='00126

The value of L derived from the first term of the expression is 437440
metres.

The correction depending on the radius not being infinitely great compared
with the section of the coil as found from the second term is —7345 metres.

The correction depending on the diameter of the wire is 1

e , - + '44997

per unit ot length J

Correction of eight neighbouring wires + '0236

For sixteen wires next to these + -0008

Correction for variation of current in different parts of section — '2500

Total correction per unit of length '22437

Length 31T236 metres.

Sum of corrections of this kind 70 „

Final value of L by calculation 430165 ,,

This value of L was employed in reducing the observations, according to
the method explained in the Report of the Committee*. The correction de-
pending on L varies as the square of the velocity. The results of sixteen
experiments to which this correction had been applied, and in which the velocity
varied from 100 revolutions in seventeen seconds to 100 in seventy-seven seconds,

* British Association Reports, 18G3, p. 169.

A DYNAMICAL THEORY OF THE ELECTROMAGNETIC FIELD. 597

were compared by the method of least squares to determine what further cor-
rection depending on the square of the velocity should be applied to make the
outstanding errors a minimum.

The result of this examination shewed that the calculated value of L should
be multiplied by 1'0618 to obtain the value of L, which would give the most
consistent results.

We have therefore L by calculation 430165 metres.

Probable value of L by method of least squares 456748 ,,

Result of rough experiment with the Electric Balance (see § 46) 410000 ,,

The value of L calculated from the dimensions of the 'coil is probably much
more accurate than either of the other determinations.

[From the Philosophical Magazine, Vol. XXVII.]

* XXVI. On the Calculation of the Equilibrium and Stiffness of Frames.

THE theory of the equilibrium and deflections of frameworks subjected to
the action of forces is sometimes considered as more complicated than it really
is, especially in cases in which the framework is not simply stiff, but is
strengthened (or weakened as it may be) by additional connecting pieces.

I have therefore stated a general method of solving all such questions in
the least complicated manner. The method is derived from the principle of
Conservation of Energy, and is referred to in Lamp's Lemons sur VElasticite,
Lecon 7™*, as Clapeyron's Theorem ; but I have not yet seen any detailed
application of it.

If such questions were attempted, especially in cases of three dimensions,
by the regular method of equations of forces, every point would have three
equations to determine its equilibrium, so as to give 3s equations between
<• unknown quantities, if s be the number of points and e the number of
connexions. There are, however, six equations of equilibrium of the system
which must be fulfilled necessarily by the forces, on account of the equality
of action and reaction in each piece. Hence if

e = 3s-6,

the effect of any external force will be definite in producing tensions or pressures
in the different pieces ; but if e > 3s — 6, these forces will be indeterminate.
This indeterminateness is got rid of by the introduction of a system of e equa-
tions of elasticity connecting the force in each piece with the change in its
length. In order, however, to know the changes of length, we require to assume
3* displacements of the s points ; 6 of these displacements, however, are equiva-
lent to the motion of a rigid body so that we have 3s — 6 displacements of
points, e extensions and e forces to determine from 3s — 6 equations of forces, e

* [Owing to an oversight this paper is out of its proper place ; it should have been immediately
before the memoir on "The Electro-magnetic Field." (No. XXV.)]

ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES. 599

equations of extensions, and e equations of elasticity ; so that the solution is
always determinate.

The following method enables us to avoid unnecessary complexity by treating
separately all pieces which are additional to those required for making the frame
stiff, and by proving the identity in form between the equations of forces and
those of extensions by means of the principle of work.

On the Stiffness of Frames.

Geometrical definition of a Frame. A frame is a system of lines connecting
a number of points.

A stiff frame is one in which the distance between any two points cannot
be altered without altering the length of one or more of the connecting lines
of the frame.

A frame of s points in space requires in general 3s — 6 connecting lines to
render it stiff. In those cases in which stiffness can be produced with a smaller
number of lines, certain conditions must be fulfilled, rendering the case one of
a maximum or minimum value of one or more of its lines. The stiffness of
such frames is of an inferior order, as a small disturbing force may produce
a displacement infinite in comparison with itself.

A frame of s points in a plane requires in general 2s — 3 connecting lines to
render it stiff.

A frame of s points in a line requires s — 1 connecting lines.

A frame may be either simply stiff, or it may be self-strained by the intro-
duction of additional connecting lines having tensions or pressures along them.

In a frame which is simply stiff, the forces in each connecting line arising
from the application of a force of pressure or tension between any two points
of the frame may be calculated either by equations of forces, or by drawing
diagrams of forces according to known methods.

In general, the lines of connexion in one part of the frame may be affected
by the action of this force, while those in other parts of the frame may not
be so affected.

Elasticity and Extensibility of a connecting piece.

Let e be the extension produced in a piece by tension-unity acting in it,
then e may be called its extensibility. Its elasticity, that is, the force required

600 ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES.

to produce extension -unity, will be - . We shall suppose that the effect of

c

pressure in producing compression of the piece is equal to that of tension in
producing extension, and we shall use e indifferently for extensibility and com-
pressibility.

Work done against Elasticity.

Since the extension is proportional to the force, the whole work done will
l>e the product of the extension and the mean value of the force ; or if x is
the extension and F the force,

x = eF,

work = %Fx = \eF* = l-x*.

c

When the piece is inextensible, or e = 0, then all the work applied at one end
is transmitted to the other, and the frame may be regarded as a machine whose
efficiency is perfect. Hence the following

THEOREM. If p be the tension of the piece A due to a tension-unity
l>etween the points B and C, then an extension-unity taking place in A will
bring B and C nearer by a distance p.

For let X be the tension and x the extension of A, Y the tension and
// the extension of the line BC '; then supposing all the other pieces inextensible,
no work will be done except in stretching A, or

But X=pY, therefore y = —px, which was to be proved.

PROBLEM I. A tension F is applied between the points B and C of a
frame which is simply stiff; to find the extension of the line joining D and E,
all the pieces except A being inextensible, the extensibility of A being e.

Determine the tension in each piece due to unit tension between B and C,
and let p be the tension in A due to this cause.

Determine also the tension in each piece due to unit tension between D
and E, and let y be the tension in the piece A due to this cause.

Then the actual tension of A is Fp, and its extension is eFp, and the
extension of the line DE due to this cause is —Fepq by the last theorem.

ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES. 601

COR. If the other pieces of the frame are extensible, the complete value
of the extension in DE due to a tension F in BC is

where 2(epq) means the sum of the products of epq, which are to be found
for each piece in the same way as they were found for A.

The extension of the line BC due to a tension F in BC itself will be

2) may therefore be called the resultant extensibility along BC.

PROBLEM II. A tension F is applied between B and C ; to find the
extension between D and E when the frame is not /simply stiff, but has
additional pieces R, S, T, &c. whose elasticities are known.

Let p and q, as before, be the tensions in the piece A due to unit
tensions in BC and DE, and let r, s, t, &c. be the tensions in A due to
unit tension in R, S, T, &c. ; also let R, S, T be the tensions of R, S, T,
and p, a; T their extensibilities. Then the tension A

= Fp+Rr+Ss + Tt + &c.;
the extension of A

the extension of R

= - Ft(epr) - RZer1 - SZers - TZert + &c. = Rp ;
extension of S

= - F1(eps) - R2(ers) - S^es* - T1(est) = So- ;
extension of T

= - F2(ept) - R2(ert) - S2(est) - TZ(e?} = Tr ;
also extension of DE

= - F2(epq) - Rl(eqr) - S2(eqs) - T2(egt) = x,

the extension required. Here we have as many equations to determine R, S, T,
&c. as there are of these unknown quantities, and by the last equation we
determine x the extension of DE from F the tension in BC.
Thus, if there is only one additional connexion R, we find

and

VOL. I. 76

602 ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES.

If there are two additional connexions R and S, with elasticities p and or,
x=-F

_
) - (2S(er«))«

/ 2 (cpr) 2 (ers) S («?s) + 2 (eps) S (egr) 2 (ers) + 2 (ep?) 2e (r3 + p) 2e (s1 + o-^l
1-2 (epr) 2 (e(jr) 2e(s> + <r) - 2 (<>p*) 2 (eg*) 2e (r1 + /») - 2 (ep?) (2 (ers))' J '

The expressions for the extensibility, when there are many additional pieces,
are of course very complicated.

It will be observed, however, that p and q always enter into the equations
in the same way, so that we may establish the following general

THEOREM. The extension in BC, due to unity of tension along DE, is
always equal to the tension in DE due to unity of tension in BC. Hence we
have the following method of determining the displacement produced at any
joint of a frame due to forces applied at other joints.

1st. Select as many pieces of the frame as are sufficient to render all its
points stiff. Call the remaining pieces R, S, T, &c.

2nd. Find the tension on each piece due to unit of tension in the
direction of the force proposed to be applied. Call this the value of p for each
piece.

3rd. Find the tension on each piece due to unit of tension in the
direction of the displacement to be determined. Call this the value of q for
each piece.

4th. Find the tension on each piece due to unit of tension along R, S, T,
Ac., the additional pieces of the frame. Call these the values of r, s, t, &c.
for each piece.

5th. Find the extensibility of each piece and call it e, those of the
additional pieces being p, a, T, <fec.

6th. R, S, T, Ac. are to be determined from the equations
Rp + R2 (er1) + S(ers) + Tt (ert) + F2 (epr) = 0,
So- + RZ(ers) + S(e,f) + T2(est) +F2(eps) = 0,

as many equations as there are quantities to be found.

7th. x, the extension required, is then found from the equation
x= -FS,(epq)-RZ(erq)-S*(eqs)- Tt(eqt).

ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES. 603

In structures acted on by weights in which we wish to determine the
deflection at any point, we may regard the points of support as the extremities
of pieces connecting the structure with the centre of the earth ; and if the
supports are capable of resisting a horizontal thrust, we must suppose them
connected by a piece of equivalent elasticity. The deflection is then the
shortening of a piece extending from the given point to the centre of the
earth.

EXAMPLE. Thus in a triangular or Warren girder of length I, depth d,
with a load W placed at a distance a from one end, 0 ; to find the deflection
at a point distant b from the same end, due to the yielding of a piece of
the boom whose extensibility is e, distant x from the sam6 end.

The pressure of the support at 0 = W — j — ; and if a; is less than a, the

W

force at x will be ~jr x(l — a), or
al

If x is greater than a,
Similarly, if x is less than b,
but if x is greater than b,

x(l — a)

— \ _ '

dl

«('-&).

~ dl '

The deflection due to x is therefore Wepq, where the proper values of p
and q must be taken according to the relative position of a, b, and x.

If a, b, I, x represent the number of the respective pieces, reckoning from
the beginning and calling the first joint 0, the second joint and the piece
opposite 1, &c., and if L be the length of each piece, and the extensibility of
each piece =e, then the deflection of b due to W at a will be, by summation
of series,

76—2

604 ON THE CALCULATION OF THE EQUILIBRIUM AND STIFFNESS OF FRAMES.

This is the deflection due to the yielding of all the horizontal pieces.
The greater the number of pieces, the less is the importance of the last term.

Let the inclination of the pieces of the web be a, then the force on a
piece between 0 and o is W , . , or

I — a ,

p = f- — when x < a.
I Bin a

and

p' =.—=—- when x>a.
I sin a

Also

q'= , . when x < b,
I sin a

q'=j— ~ when x > b.
I sin a

If e be the extensibility of a piece of the web, we have to sum WZe'p'tf
to get the deflection due to the yielding of the web,

-
sin" a

INDEX TO VOL. I.

Absorption, Electric, 573
Ampere, 193

Beam : bent into a circular ring, 55 — 57, 65
Bending, Lines of, 87, 97, 105, 106, 107, 108
Brewster, Sir David, 43, 63, 68, 263, 413

Cauchy, 32, 40, 71

Challis, 453, 503, 505

Clapeyron, 30, 32, 70, 72

Clausius, 377, 386, 387, 405

Coil, Coefficient of induction of, 592—597

Colour-blindness, 119, 137, 441

Colour box, 420 ; Method of observation, 426

Coloured beams of light, Mixtures of, 143 — 147

Coloured discs, 128, 264, 265

Coloured powders, 142

Colour equations, 128, 129, 138—141, 148, 268—270,
427, 428, 442

Colour: Experiments on, 126, 263; Law of Per-
ception of, 130; Theory of the Perception of,
135

Colours : History of, 411 ; Mathematical theory of
Newton's diagram of, 416 ; Method of repre-
senting them by straight lines, 418 ; Theory of
Compound, 149, 243, 410; Three primary,
Theory of, 445

Colour sensations, Relations of to the pure rays of
the spectrum, 150

Colour top, 127, 147

Colour triangle, 121, 131—135, 268, 416

Compression, Equations of, 36

Condenser, Capacity of, 572 ; Theory of, 572—576

Contact, Conic of, 93, 94, 102

Cotes, 271

Coulomb, 33, 71

Current Electricity, Conduction of, 180

Currents: Action enclosed, 183; Intrinsic energy
of, 541; Mutual action between, 537; Produced
by induction, 185 ; Quantity and Intensity as
properties of, 189

Curvature, entire and specific, 89, 92

Cylinder : acted on by centrifugal force, 60 ; hollow,
exposed to pressure, 45 — 50 ; hollow, dilated
by heat, 62; of parallel wires twisted, 59;
twisted, 42, 43, 44, 66

D'Alembert, 248

Diagrams, Conditions of indeterminateness in draw-
ing reciprocal diagrams, 516; of Force, 514
Dielectrics, Theory of, 177
Dynamometer, Weber's, 546

Earth's motion, 259

Elastic bodies, Collisions of, 405

Elasticity : Axioms, 31 ; Equations of, 38 ; Co-
efficients of, 41

Elastic solids : Equilibrium of, 30 ; Pressures in the
interior of, determined by the action on polar-
ized light, 68

Electric absorption, 573

Electrical images, 209

Electromagnet, Effect of the core, 222

Electromagnetic disturbances, Propagation of, 578,
583

Electromagnetic Field : Dynamical theory of, 526 ;
General equations of, 534, 551—552, 554—564;
Mechanical actions in, 565 — 570

Electromagnetic Induction, 536 ; Calculation of
coefficients, 589

606

IXDEX TO VOL. I.

Electromagnetic momentum, 538 : Dynamical illus-
tration of.

Electromagnetism : Ampere's laws, 193
Elect rotonic state, 188, 205, 538
Equipotential surfaces, Magnetic, 553
Euler, 29, :iJ. i'lS 271

Faraday, 1.55, 188, 205, 241, 504, 629, 531, 535,

:>li', 573,585
Felici. 538
Figures : General relation between the numbers of

points lines and polygons in, 515; Reciprocal,

514

Fizeau, 500, 580
Fluid : Application to Lines of Force, 175; Theory

of motion, 160; through a resisting medium, 163
Foramen Centrale, Sensibility to light, 242
Forbes, Professor, 1, 124, 142, 145, 146,243
Force : Diagrams of, 514 ; Lines of, 155, 158, 241 ;

Magnetic Lines of, 551 ; Physical lines of, 451
Forces: Absolute values of, concerned in propagation

of light, 588; Electromotive, 181
Foucault, 229, 248, 580
Fourier, 361
Frames, Equilibrium and stiffness of, 598

Gases: Conductivity of, 403—405; Division of,
392, 403; Dynamical Theory of, 377; Fric-
tion, 390 ; Ratio of specific heats of, 409

Gauss, 81, 88—90, 271

Graham, 403

Grassman, 125, 414, 419

Gravitation, arising from the action of surrounding
medium, 571

Green, 196

Gregory, Dr, 126

Gutta percha, optical properties of, 43

Heavy body, descent of, in a resisting medium, 115
Helmholtz, 125, 141, 144, 145, 146, 152, 204, 243,

271, 414,415, 488, 533
Herapath, 377
Hersshel, Sir J. F. W., 142

Image, perfect defined, 273
Images, Electrical, Theory of, 209

Induction : by motion of conductor, 540 ; Coefficients
of, for two circuits, 539 ; Determination of
coefficients of, 547 ; Electromagnetic, ~>:,r, .
Magnecrystallic, Theory of, 180 ; of one cur-
rent by another, 540 ; Paramagnetic and dia-
magnetic, Theory of, 178

Instrument, perfect optical, 274

Isinglass, optical properties of, 43, 67

Jenkin, F., 532, 573
Joule, 377 .

Kohlrausch, 492, 499, 535, 569, 579

Lam«, 30, 32, 70, 72

Laplace, 292, 293, 294, 369

Leslie, Sir John, 16

Light : distinction between optical and chromatic
properties of, 411 ; Electromagnetic theory of,
502, 577 ; propagation of, forces called into
play, 587 ; propagation of, in a crystallised
medium, 583

Liquids, Compressibility of, 50

Listing, 271

Magnetic field of variable intensity, 214

Magnetic lines of force, 552

Magnetism, Quantity and Intensity as indicated by

lines of force, 192

Magnets, Permanent, Theory of, 178
Momentum, Electromagnetic, 538

Nasmyth, J., 57
Navier, 30, 31, 32, 72
Neumann, 208, 512, 527

Newton, 3, 124, 135, 142, 143, 144, 145, 146, 149,
151, 249, 267, 410, 411, 412

(Ersted, 30, 33, 50

Optical Instruments, General laws of, 27 1 ; Mathe-
matical treatment of, 281 — 285
Oval Curves, 1

Plate, bent by pressures, 57
Plateau, 243, 295
Platometer, New form of, 230

INDEX TO VOL. I.

607

Plucker, 585
Poinsot, 248, 250

Poinsot's Theory of Rotation, Instrument to illus-
trate, 246
Poisson, 30, 32, 72

Polyhedron, inscribed in a surface, 94, 98, 99
Pressures, Equations of, 37
Problems, Solutions of, 74

Ray, Reduced path of, 280

Reciprocal Figures, 514; Application^ to Statics,
522 ; Possibility of drawing, 516 ; Relation
between the number of points lines and polygons
in, 515

Refraction, index of, how related to specific in-
ductive capacity, 583

Regnault, 33, 71

Resistance, Electric, how related to transparency,
586

Resisting medium, descent of a heavy body in, 115

Rigid body, Stability of the steady motion about a
fixed centre of force, 374 ; motion of about a
sphere, 296

Ring : motion of, when rigid, about a sphere, 296 —
310 ; motion of, when the parts are not rigidly
connected, 310; of equal satellites, 360

Rings, Effect of long continued disturbances on a
system of, 352; mutual perturbations of two,
345 ; Fluid, Loss of energy due to friction, 354

Rolling Curves, 4 ; Examples of, 22—29

Rolling of Curves on themselves, 19

Rotation, Theory of, 249

Saturn's Rings, Stability of, 286, 288—376

Sources and sinks denned, 163

Spectrum, Relation of the Colours of, to Compound

Colours, 410
Sphere, hollow, dilated by heat, 64 ; exposed to

pressures, 51 — 55
Sphere: magnecrystallic, 217 ; magnetic in uniform

field of magnetic force, 212

Spherical shell: Electromagnetic, 220, 222; Per-
manent magnetism in, 220 ; Revolving in
magnetic field, 226

Spherical electromagnetic coil machine, 224

Spheres perfectly elastic : Motions and Collisions of,
378 ; Boyle's law, 389; Mean distance between
collisions, 386 ; Mean Velocity, 381 ; Mean Velo-
city-square, 381 ; Two systems in same vessel,
383

Spheres: Two, between poles of a magnet, 216;
Two, in uniform magnetic field, 215

Stokes, 32, 33, 71, 72, 143, 209, 391, 410

Struve, 292

Surfaces: Applicability of, 95; Conjugate systems
of curves on, 95, 96 ; Transformation of, by
bending, 80

Telescopes, perfect, 275 — 279

Thomson, Sir W., 157, 196, 199, 209, 212, 374,
453, 503, 505, 528, 529, 533, 588

Top, Dynamical, 248 ; Instantaneous axis, 255 ;
Invariable axis, 252 — 255 ; Method of observ-
ing the motion, 257

Tractory of a curve, 13

Tractory of circle, 15, 17

Tubes, unit, 161

Unit, Electrostatic and Electromagnetic, 569

Verdet, 504, 506, 507, 513, 529

Vortices, molecular : Applied to Electric Currents,

467 ; Applied to Magnetic phenomena, 451 ;

Applied to Statical Electricity, 489

Wave length, Method of determining, 423

Weber, 208, 492, 499, 507, 527, 535, 545, 569,

579

Wheatstone, 434
Wilson, Dr George, 137, 415

Young, 32, 60, 124, 136, 137, 412, 419, 447

CAMBRIDGE : PRINTED BY C. J. CLAY, M.A. AND SONS, AT THE UNIVERSITY PRESS.

N.

\

60

PLEASE DO NOT REMOVE
CARDS OR SLIPS FROM THIS POCKET

UNIVERSITY OF TORONTO LIBRARY

S Maxwell, James Clerk

M4653s Scientific Papers, ed. by

v.l Niven

BioMed