A Treatise On Problems of Maxima and Min

This is a reproduction of a library book that was digitized
by Google as part of an ongoing effort to preserve the

information in books and make it universally accessible.
https://books.google.com
BIBLIOGRAPHIC RECORD TARGET
Graduate Library
University of Michigan
Preservation Office
Storage Number:
ABQ9766
ULFMTBRTaBLmT/C DT 09/12/88 R/DT 09/12/88 CC STATmmE/Ll
010: : | a 04012718//r85
035/1: : | a (RLIN)MIUG86-B48476
035/2: : | a (CaOTULAS)160121093
040: : |aDLC/ICU |cICU |dMiU
050/1:0: |aQA563 |b.R17
100:1 : | a Ramchundra, | d 1821-1880.
245:02: | a A treatise on problems of maxima and minima, | b solved by algebra.
| c By Ramchundra. Reprinted by order of the Honourable court of directors of
the East-India company for circulation in Europe and in India, in
acknowledgment of the merit of the author, and in testimony of the sense
entertained of the importance of independent speculation as an instrument of
national progress in India. Under the superintendence of Augustus De Morgan ...
260: : | a London, | b W.H. Allen & co., | c 1859.
300/1: : | a xxiii, v p., 1 L., 185 p., 1 L. | b diagrs. on 8 pi. | c 23 cm.
500/1: : | a Has also special t.-p. with imprint: Calcutta, Printed by P.S.
d'Rozario and co., 1850.
650/1: 0: | a Maxima and minima
700/1:1 : | a De Morgan, Augustus, | d 1806-1871. | e ed.
998: : | c DPJ | s 9120
Scanned by Imagenes Digitales

Nogales, AZ
On behalf of
Preservation Division
The University of Michigan Libraries
Date work Began: _

Camera Operator: .
TEEATI8E ON PROBLEMS
OF
MAXIMA AND MINIMA,

SOLVED BY ALGEBKA.
BY
RAMCHUNDRA,
LATE TEACHER OF SCIENCE, DELHI COLLEGE.
REPRINTED BY ORDER OE THE HONOURABLE COURT OP DIRECTORS

OE THE EAST-INDIA COMPANY EOR CIRCULATION IN EUROPE AND IN
INDIA, IN ACKNOWLEDGMENT OE THE MERIT OE THE AUTHOR,
AND IN TESTIMONY OE THE SENSE ENTERTAINED OE THE
IMPORTANCE OE INDEPENDENT SPECULATION AS AN
INSTRUMENT OE NATIONAL PROGRESS IN INDIA.
WLvtizx tfye <8upmtttm&mce o£

AUGUSTUS DE MORGAN, F.KA.S. F.C.P.S.
OF TRINITY COLLEGE, CAMBRIDGE}
PROFESSOR OF MATHEMATICS IM" UNIVERSITY COLLEGE, LONDON.
LONDON;
WM. H. ALLEN & CO. 7, LEADENHALL STREET.
1859.
LONDON :
COX AND WYMAN, PRINTERS, GREAT QUEEN STREET,
lincoln's-inn FIELDS.
EDITOK'S PBEFACE.
In the year 1850, iny friend the late* J. E. Drinkwater-

Bethune forwarded to England a number of copies of a work on
* John Elliot Drinkwater, the eldest son of Lieutenant-Colonel John

Drinkwater, author of the " History of the Siege of Gibraltar," was born
July 12, 1801, was educated at Westminster and at Trinity College, Cam
bridge, and took the degree of B,A., as fourth wrangler, in 1823. He was
called to the bar about 1827, and wThen Lord Grey came into office in 1831,
was employed by the Government on various commissions. He was for about
fourteen years counsel to the Home Office, and had much to do with the Par
liamentary Reform Bill, the Municipal Reform Bill, Medical Reform Bills,
the establishment of the Queen's Colleges in Ireland, the organization of the
County Courts, and other important measures. During this time he published,
in the Library of Useful Knowledge, part of a treatise on algebraical expres
sions, which was never finished, and lives of Galileo and Kepler, which exhibit
great research and acumen, and stand high among modern English efforts in
scientific biography. He also printed for private circulation a translation of
Schiller's " Maid of Orleans," and of some of Tegner's Swedish poems. In
1836 his mother inherited the estate of Bethune of Balfour, in Fife, and the
whole family added the name of Bethune to their surname. In 1848 he sailed
for India as fourth ordinary member of the Supreme Council, and president
of the Law Commission. Lord Dalhousie added the presidentship of the
Council of Education. He devoted more attention to the cause of education
than even to his legislative duties. In his private capacity he founded at
Calcutta a school for Hindoo girls of the higher classes. He bequeathed the
laud and building to the East-India Company, on condition that the school
should become a Government institution. The offer was accepted, and the
school is in successful operation. He also prevailed upon the representatives
of the family of Tippoo Saib to throw open to Mohamedans of good family
the school which had been endowed for the exclusive use of that family. He
procured an enactment by which natives converted to Christianity are not
deprived of their rights of inheritance. He had to encounter much virulence
of opposition, both from natives and Europeans ; but his character and manners
Maxima and Minima, by.Bamchundra, teacher of science, Delhi
College, with directions to present copies to various persons,
and among others to myself. On examining this work I saw in
it, not merely merit worthy of encouragement, but merit of a
peculiar kind, the encouragement of which, as it appeared to me,
was likely to promote native effort towards the restoration of the
native mind in India. Mr. Drinkwater-Bethune's lamented death,
which took place shortly after he had dispatched the books, pre
vented my knowing whether he also entertained any opinion simi
lar to mine as to the distinctive character of Kamchundra's work ;
but, from his own knowledge of the history of mathematics, I
think it highly probable. I addressed my thanks for the present
to his successor, Mr. Colvile, with some remarks on the subject.
Having taken further time to think of it, I determined to call the
attention of the Court of Directors to Eamchundra's work, 'in the
hope that it would lead to acknowledgment of his deserts. I
accordingly addressed a letter (July 24, 1856) to Colonel Sykes,
the Chairman, to whom I had previously mentioned the matter
at a casual meeting. This letter was at once forwarded to the
Lieutenant-G-overnor of the JNTorth-"W"est Provinces, with instruc
tions to procure a report on the case, and to suggest, on the
supposition of a public reward being approved of, the kind of
reward which should be given. Answers were received by the
Court, which were communicated to me on the 3rd of March,
1858. They contained various replies to the questions proposed,
by H. Stewart Eeid, Esq., Director of Public Instruction in the
North-West Provinces, and other gentlemen connected with the
procured him the respect he deserved before his death, which took place
August 12th, 1851, from inflammation of the liver. He lived two lives of
real utility, one in England and one in India ; and as many in either country
know nothing of his career in the other, and this work is intended for both, a
short abstract of his life is here given. This abstract is the more appropriate
as his encouragement of Kamehundra was the first of the train of circumstances.
which produced the reprint now before the reader.
same department. With some difference of opinion as to the
mode of acknowledgment, there was nnanimons appreciation of
Bamchundra's services to his country, and admission of the
desirableness of encouraging his efforts. The Court accompanied
the communication of these answers to me with a request that I
would point out how to bring Eamchundra under the notice of
scientific men in Europe. In my reply (March 18), assuming
distinctly that I conceived the question to be, not merely how
Kamchundra could be rewarded, but how his work might be made
most effective in the development of Hindoo talent, I recom
mended the circulation of the work in Europe, with a distinct
account of the grounds on which the step was taken. I entered at
some length into my own view of those grounds, and volunteered
to draw up the statement which should accompany the publication.
After some correspondence on details, the Court (July 1), ex
pressing entire satisfaction with my views, and characterizing
them as " deserving of the most attentive consideration by all
who are charged with the superintendence of education in India
in its higher grades," accepted my offer to superintend the present
reprint, for circulation in Europe and in India.
I shall at once proceed to a short account of these views ; after
which I shall give some account of Eamchundra, the author of the
work. Of course it will be remembered that the late Court of
Directors is in no way answerable for the details of my exposition,
though .their decided approbation was bestowed on the general
sketch which. I laid before them.
There are many persons, even among those who seriously turn
their thoughts to the improvement of India, who look upon the
native races as men to be dealt with in the same manner as Caffres
or New Zealanders. Judging by the lower races of the Peninsula,
and judging even these more by the grosser parts of their mytho
logy than by the state of domestic life and hereditary institutions,
they presume that the Indian question resolves itself into an
inquiry how to create a mind in the country^ and that mind fash
ioned on the English standard. They forget that at this very
moment there still exists among the higher castes of the country
—castes which exercise vast influence over the rest—a body of
literature and science which might well be the nucleus of a new
civilization, though every trace of Christian and Mohamedan civil
ization were blotted out of existence. They forget that there exists
in India, under circumstances which prove a very high antiquity,
a philosophical language which is one of the wonders of the world,
and which is a near collateral of the Greek, if not its parent form.
From those who wrote in this language we derive our system of
arithmetic, and the algebra which is the most powerful instru
ment of modern analysis. In this language we find a system of
logic and of metaphysics : an astronomy worthy of comparison with
that of Greece in its best days ; above comparison, if some books
of Ptolemy's Syntaxis be removed. "We find also a geometry, of
a kind which proves that the Hindoo was below the Greek as a
geometer, but not in that degree in which he was above the
Greek as an arithmetician. Of the literature, poetry, drama,
&c, which flourished in union with this science, I have not here
to speak.
Those who consult Colebrooke's translation of the Vija Ganita,
or the account given of it in the Fenny Cyclopedia, wrill see that I
have not exaggerated the point most connected with this preface.
For others I will quote the impression made, five-and-thirty
years ago, upon the mind of a mathematician wrhose subsequent
career and present position will give that weight to an extract
from his opinions which would have been given to any reader of the
whole article by the article itself, even had it been anonymous.
Sir John Herschel, in the historical article Mathematics, in JBreiv-
sterns Cyclopcedia, after some general account of the Hindoo
algebra, proceeds as follows :—

Vll
" The Brahma Sidd'hanta, the work of Brahmegupta, an Indian

astronomer at the beginning of the seventh century, contains a
general method for the resolution of indeterminate problems of
the second degree; an investigation which actually baffled the
skill of every modern analyst till the time of Lagrange' s solution,
not excepting the all-inventive Euler himself. This is matter of a
deeper dye. The Greeks cannot for a moment be thought of as
the authors of this capital discovery; and centuries of patient
thought, and many successive efforts of invention, must have
prepared the way to it in the country where it did originate. It
marks the maturity and vigour of mathematical knowledge, while
the very work of Brahmegupta, in which- it is delivered, contains
internal evidence that in his time geometry at least was on the
decline. For example, he mentions several properties of qua
drilaterals as general, which are only true of quadrilaterals
inscribed in a circle. The discoverer of these properties (which are
of considerable difficulty) could not have been ignorant of this
limitation, which enters as an essential element in their demon
stration. Brahmegupta then, in this instance, retailed, without
fully comprehending, the knowledge of his predecessors. "When
the stationary character of Hindu intelligence is taken into the
account, we shall see reason to conclude, that all we now possess
of Indian science is but part of a system, perhaps of much greater
extent, which existed at a very remote period, even antecedent to
the earliest dawn of science among the Greeks, and might autho
rize as well the visits of sages as the curiosity of conquerors."
Greece and India stand out, in ancient times, as the countries of

indigenous speculation. But the intellectual fate of the two nations
was very different. Among the Greeks, the power of speculation
remained active during their whole existence as a nation, even
down to the taking of Constantinople : it declined, indeed, but it
was never extinguished. Their latest knowledge was inquisitive,
as well as their earliest. They preserved their great writers
unabridged and unaltered; and Euclid did not degenerate into
what are called practical rules.
In India, speculation died a natural death. A taste for routine
—a thing to which inaccurate thinkers give the name of 'practical
—converted their system into a collection of rules and results.
Of this character are all the mathematical books which, have been
translated into English ; perhaps all which still exist. That they
must have had an extensive body of demonstrated truths is
obvious ; that they lost the power and the wish to demonstrate is
certain. The Hindoo became, to speak of the highest and best
class, the teacher of results which he could not explain, the
retailer of propositions on which he could not found thought. He
had the remains of ancestors who had investigated for him, and
he lived on such comprehension of his ancestors as his own small
grasp of mind would allow him to obtain. He fed himself and his
pupils upon the chaff of obsolete civilization, out of which Euro
peans had thrashed the grain for their own use.
But the mind thus degenerated is still a mind ; and the means
of restoring it to activity differ greatly from those by which a
barbarous race is to be gifted with its first steps of progress.
No man alive can, on sufficient data, reason out the restoration
of a decayed national intellect, possessed of a system of letters
and science which has left nothing but dry results, inveterate
habits of routine, great reverence for old teachers, and small
power of comprehending the very teaching which is held in tradi
tional respect. And this because the question is now tried for the
first time. Many friends of education have proposed that Hindoos
should be fully instructed in English ideas and methods, and made
the media through which the mass of their countrymen might
receive the results in their own languages. Some trial has been
given to this plan, but the results have not been very encouraging,
in any of the higher branches of knowledge. My conviction is,
that the Hindoo mind must work out its own problem ; and that
all we can do is to set it to worlc ; that is, to promote independent
speculation on all subjects by previous encouragement and sub
sequent reward. This is the true plan ; all others are neither fish
nor flesh.
That sound judgment which gives men well to know what is
best for them, as well as that faculty of invention which leads to
development of resources and to the increase of wealth and com
fort, are both materially advanced, perhaps cannot rapidly be
advanced wifchout, a great taste for pure speculation among the
general mass of the people, down to the lowest of those who can
read and write. England is a marked example. Many persons
will be surprised at this assertion. They imagine that our country
is the great instance of the refusal of all unpractical knowledge in
favour of what is useful. I affirm, on the contrary, that there is
no country in Europe in which there has been so wide a diffusion
of speculation, theory, or what other unpractical word the reader
pleases. In our country, the scientific society is always formed
and maintained by the people ; in every other, the scientific aca
demy—most aptly named—has been the creation of the govern
ment, of which it has never ceased to be the nursling. In all the
parts of England in which manufacturing pursuits have given the
artisan some command of time, the cultivation of mathematics and
other speculative studies has been, as is well known, a very fre
quent occupation. In no other country has the weaver at his loom
bent over the Principia of Newton ; in no other country has the
man of weekly wages maintained his own scientific periodical.
With us, since the beginning of the last century, scores upon
scores—perhaps hundreds, for I am far from knowing all—of
annuals have run, some their ten years, some their half-century,
some their century and a half, containing questions to be answered,
from which many of our examiners in the universities have culled
materials for the academical contests. And these questions have
b
always been answered, and in eases without number by the lower
order of purchasers, the mechanics, the weavers, and the printers'
workmen. I cannot here digress to point out the manner in which
the concentration of manufactures, and the general diffusion of
education, have affected the state of things ; I speak of the time
during which the present system took its rise, and of the circum
stances under which many of its most effective promoters were
trained. In all this there is nothing which stands out, like the
state-nourished academy, with its few great names and brilliant
single achievements. This country has differed from all others
in the wide diffusion of the disposition to speculate, which dis
position has found its place among the ordinary habits of life,
moderate in its action, healthy in its amount.
The history of England, as well as of other countries, having im
pressed me with a strong conviction that pure speculation is a
powerful instrument in the progress of a nation, and my own birth
and descent having always given me a lively interest in all that
relates to India, I took up the work of Eamchundra with a mingled
feeling of satisfaction and curiosity : a few minutes of perusal
added much to both. I found in this dawn of the revival of Hindoo
speculation two points of character belonging peculiarly to the
Greek mind, as distinguished from the Hindoo ; one of which may
have been fostered by the author's European teachers, but certainly
not the other.
The first point is a leaning towards geometry. Persons who
are not mathematicians imagine that all mathematicians are for all
mathematics. Nothing can be more erroneous. Not merely have
the two great branches, geometry and algebra, their schools of
disciples, each of which looks coldly upon the other ; but even
geometry itself, and algebra itself, have subdivisions of which the
same thing may be said. Eor example, Mr. Drinkwater-Bethune,
above mentioned, was by taste an algebraist ; as a practised eye
would at once detect from his unfinished work on equations.
XI
Business brought him to my house one morning, nearly thirty

years ago, at a time when I happened to be studying some of the
geometrical developments of the school of Monge. On my point
ing out to him some of the most remarkable of the conclusions,
he said, with a smile, " I see that sort of thing has charms for
you." Now the Hindoo was also an algebraist, as decidedly as
the Greek was a geometer : the first sought refuge from geometry
in algebra, the second sought refuge from arithmetic in geometry.
The greatness of Hindoo invention is in algebra ; the greatness of
Greek invention is in geometry. But B,amchundra has a much
stronger leaning towards geometry than could have been expected
by a person acquainted with the Vija Ganita ; but he has not the
power in geometry which he has in algebra. I have left one or
two failures—one very remarkable—unnoticed, for the reader to
find out. Should this preface—as I hope it will—fall into the
hands of some young Hindoos who are systematic students of
mathematics, I beg of them to consider well my assertion that
their weak point must be strengthened by the cultivation of pure
geometry. Euclid must be to them what Bhascara, or some
other algebraist, has been to Europe.
The second point is yet more remarkable. Greek geometry, as
all who have read Euclid may guess, gained its strength by
striving against self-imposed difficulties. It was not permitted to
take instruments from every conception which the human mind
could form ; definite limitation of means was imposed as a con
dition of thought, and it was sternly required that every feat of
progress should be achieved by those means, and no more. Just
as the Greek architecture studied the production of rich and
varied effect out of the simplest elements of form, so the Greek
geometry aimed at the demonstration of all the relations of
figure on the smallest amount of postulated basis. The great
problem of squaring the circle, now with good reason held in
low esteem, was the struggle of centuries to bring under the
dominion of the prescribed means what might with the utmost
ease have been conquered by a very small additional allowance.
The attempt was unsuccessful ; so was that of Columbus to dis
cover India from the west. But Columbus commenced the addition
of America to the known world ; and in like manner the squarers
of the circle, and their refuters, added field on field to the extent
of geometry, and aided largely in the preparation for the modern
form of mathematics. Yery few of these additions would have
been made, at or near the time when they were made, if it had
satisfied the Greek mind to meet each difficulty, as it occurred,
by permission to use additional assumptions in geometry.
The remains of the Hindoo algebra and geometry show to us no
vestige of any attempt to gain force of thought by struggling
against limitation of means : this, of course, because their mode of
demonstration does not appear in the works which are left, or at
least in those which have become known to Englishmen. But we
have here a native of India who turns aside, at no suggestion but
that of his own mind, and applies himself to a problem which has
hitherto been assigned to the differential calculus, under the con
dition that none but purely algebraical process shall be used. He
did not learn this course of proceeding from his European guides,
whose aim it has long been to push their readers into the diffe
rential calculus with injurious speed, that they may reach the full
application of mathematics to physics ; and who often allow their
pupils to read Euclid with eyes shut to his limitations. Eamchun-
dra proposed to himself a problem which a beginner in the diffe
rential calculus masters with a few strokes of the pen in a month's
study, but which might have been thought hardly within the possi
bilities of pure algebra. His victory over the theory of the difficulty
is complete. Many mathematicians of sufficient power to have
done as much would have told him, when he first began, that the
end proposed was perhaps unattainable by any amount of thought ;
next, that when attained, it would be of no use. But he found in
Xlll
the demands of his own spirit an impulse towards speculation of a

character more fitted to the state of his own community than the
imported science of his teachers. He applied to the branch of
mathematics which is indigenous in India, the mode of thought
under which science made its greatest advances in Greece. My
own strong suspicion that it was the want of this mode of thought
which allowed the decline of algebra in ancient India, coupled
with my thorough conviction that, whether or no, this mode of
thought yields the proper nutriment for mathematical science in
its early and feeble life, produced the recommendation to the
Court of Directors to which this reprint owes its existence.
Eamchundra's problem—and I think it ought to go by that
name, for I cannot find that it was ever current^ as an exercise of
ingenuity in Europe—is to find the value of a variable which will
make an algebraical function a maximum or a minimum, under
the following conditions. Kot only is the differential calculus to
be excluded, but even that germ of it which, as given by Permat
in his treatment of this very problem, made some think that he
was entitled to clain the invention. The values of §cc and of
<p (x+7i) are not to be compared ; and no process is to be allowed
which immediately points out the relation of §x to the derived
function (p'cc. A mathematician to whom I stated the conditioned
problem made it, very naturally, his first remark, that he could
not see how on earth I was to find out when it would be biggest,
if I would not let it grow. The mathematician will at last see
that the question resolves itself into the following:—Kequired
a constant, r, such that cpx—r shall have a pair of equal roots,
without assuming the development of §(oc-\-h), or any of its
consequences.
* It would not at all surprise me if it should be found that some one

inquirer has suggested the problem ; but, if so, I think the search which I
have made entitles me to say that the suggestion entirely failed to attract
attention, and to establish the difficulty as a recognized exercise.
It will readily be seen that a short paper, with a few examples,
would have sufficed to put the whole matter before a scientific
society. But it was Ramchundra's object fco found an elementary
wrork upon his theorem, for the use of beginners, with a large
store of examples. As to the method which he has adopted, Eu
ropeans must remember that his purpose is to teach Hindoos, and
that probably he knows better how to do this than they could tell
him. The excessive reiteration of details, and the extreme mi
nuteness of the algebraical manipulations, are excellent examples
of that patience of routine which is held to be a part of the
Hindoo character. I may make two remarks on matters which
would strike the most casual observer.
First, the constant occurrence of " the same solved without im
possible roots," and the transformation by which it is effected,
will remind the English mathematician who has his half-century
over his head, of the old " pure quadratic," and the victory which
was supposed to be gained when the " adfected quadratic " was
evaded by attention to the structure of the given equation. Eam-
chundra and Dr. Miles Bland, &c. &c, are here precisely on the
same scent, both making much of the same little.
Secondly, in the confusion of terms which sometimes appears,
in language implying that an equation is a factor of an equation,
instead of an expression a factor of an expression, we have the
same incorrectness which appears in more than one edition of
"Waring' s Meditationes Algelraicce, and which occasioned some
amount of objection to the whole theory from those who could not
see the inaccuracy and its correction. As in
"Scribatur x— a=0, x— /3=0, x— y = 0, a?— 3=0, &c. et per

sequationem ex horum factorum continua multiplicatione
(x — a) X (x— /3) X (x—y) X (#— S) X &C. = 0
generatam dividatur data sequatio."
I believe that selections from Kamchundra's work might advan

tageously be introduced into elementary instruction in this coun
try. The exercise in quadratic equations which it would afford,
applied as it is to real problems, would advantageously supersede
some of the conundrums which are manufactured under the name
of problems producing equations.
In the printing I have followed the original in every point,
altering nothing except obvious errata, including the restoration
of the numeral symbol 0, which in the original is always the
letter o. This again is a mistake into which "Waring allowed his
printer to fall in almost all his writings. I thought that the Euro
pean reader would be more curious to look at the way in which
the Calcutta printer treated mathematical manuscript when his
author was no nearer than Delhi, than to see the manner in which
I could mend it. My printers, Messrs. Cox & "Wyman, have
entered fully into the plan, and have produced as nearly a fac
simile as possible. I may add that the Calcutta printer has
acquitted himself in a manner entitled to especial notice and high
praise.
Hamchundra, the author of this work, has transmitted to me
some notes of his own life, from which I collect as follows. He was
born in 1821, at Paneeput, about fifty miles from Delhi. His
father, Soondur Lall, was a Hindoo Kaeth, and a native of Delhi,
and was there employed under the collector of the revenue. He
died at Delhi in 1831-32, leaving a widow (who still survives) and
six sons. After some education in private schools, Eamchundra
entered the English Government school at Delhi, to every pupil
of which two rupees a month were given, and a scholarship of five
rupees a month to all in the first and second classes. In this
school he remained six years. It does not appear that any par
ticular attention was paid to mathematics in this school ; but,
shortly before leaving it, a taste for that science developed itself
in Eamchundra, who studied at home with such books as he
could procure. After leaving school, he obtained employment as
a writer for two or three years. In 1841, changes took place in
the educational department of the Bengal presidency ; the school
was formed into a college ; and Ramehundra obtained, by compe
tition, a senior scholarship, with thirty rupees a month. In 1844,
he was appointed teacher of European science in the Oriental
department of the college, through the medium of the vernacular,
with fifty rupees a month additional. A vernacular translation
society was instituted, and Eamchundra, in aid of its object,
translated or compiled works in Oordoo, and also on algebra,
trigonometry, &c, up to the differential and integral calculus.
"These translations"—! now proceed to quote Kamchundra's
words—" were introduced into the Oriental department as class-
books ; so that in two or three years many students in the Arabic
and Persian departments were, to a certain extent, acquainted
with English science : and the doctrines of the ancient philosophy,
taught through the medium of Arabic, were cast into the shade
before the more reasonable and experimental theories of modern
science. The old dogmas, such as ' that nature abhors a vacuum,'
and c that the earth is the fixed centre of the universe,' were gene
rally laughed at by the higher students of the Oriental, as well as
by those of the English departments of the Delhi College. But
the learned moulwees, &c, who lived in the city and had no
connection with the college, did not like this innovation on their
much-beloved theories of the ancient Greek philosophy, which
from centuries past had been cultivated among them.
" I, with the assistance of the higher students of the English
and Oriental departments, formed a society for the diffusion of
knowledge among our countrymen. We were ambitious enough
to imitate the plan of the Spectator. "We first commenced a
monthly, and then a bi-monthly periodical, called the JFawdeddnnd-
zireen (i. e. useful to the reader), at the cheap price of four annas
a month, in which notices of English science were given, and in
which not only were the dogmas of the Mohamedan and Hindoo
philosophy exposed, but also many of the Hindoo superstitions
and idolatries were openly attacked. The result of this was that
many of our countrymen, the Hindoos, condemned us as infidels
and irreligious ; but as we did not advocate Christianity, but only
recommended a kind of deism, and as we never lost our caste pub
licly, by eating and drinking, all our free discussions did not much
alarm our Hindoo friends. "When in private meetings our friends,
seeing us so warmly advocating English science and knowledge,
taunted us by saying we will become Christians, as such and such
pundit had become, then we considered this as an insult, and
stated in reply, that the pundit referred to had not received any
English education, and that he was ignorant, and was therefore
deceived by the missionaries, whom we considered as ignorant and
superstitious as our own uneducated friends. We went so far as
to challenge our Hindoo friends to bring any Christian missionary
to us, and see whether he can persuade us. It was then my
conscientious belief that educated Englishmen were too much
enlightened to believe in any bookish religion except that of reason
and conscience, or deism. Sometimes, when the late Baptist
missionary, Mr. Thompson, stopped me in the bazaar, and required
me to think of my eternal concerns, and gave me some tracts, &c.
in Persian and Oordoo, I did not speak to him much,—received
parts of the New Testament, &c, and when I returned home I
put them in a corner, and never read them.
" Once a learned Mohamedan came to me with a copy of the
New Testament in Oordoo, and having read some portion of St.
Paul's epistles, spoke greatly against the apostle, and the mission
aries in general, because St. Paul teaches that circumcision is of
no use for salvation. His object in reading this to me was to get
an English scholar and a teacher of English science to agree with
him in saying how absurd Christianity and Christians were.
Though what he read was in my mother tongue, still it was
wholly Greek to me; I did not understand the question. In
order to put a stop to this talk, in which I had then no interest,
I briefly told him that, for my part, I considered not only Chris
tianity, but also Mohamedanism, and all bookish religions, as
absurd and false. Upon this all Hindoos and Mohamedans pre
sent paid me the compliment of being a philosopher, and departed
with marks of approbation and goodwill.
" A respectable and learned Mohamedan, secretly assisted by
some other celebrated moolwees of the city, published a treatise
in Oordoo in refutation of the motion of the earth, on the prin
ciples of Aristotelian philosophy ; the whole train of reasonings
being copied almost verbatim from a metaphysical work in Arabic,
called Myloodee. But no sooner was this publication made
over to us, than a moolwee, and some higher students of the Ara
bic department, got up a sharp reply, and published it ; to which
no answer was returned. Afterwards, in addition to the bi
monthly periodical, we commenced a monthly magazine, called
the MooMh-i-Hind, or the Friend of India. Eut it must be
confessed that we did not receive sufficient support from the native
public, and it was principally through the patronage of English
authorities, as Sir John Lawrence (the magistrate of Delhi), Mr.
A. A. Eoberts (ditto ditto), Dr. A. Eoss, Mr. J. F. Grubbins (then
judge at Delhi), who subscribed for several copies of our perio
dicals, that we got sufficient money to pay the expenses of our
publications. But afterwards, times and circumstances being
changed, we were compelled to discontinue them; so that, in 1852,
the bi-monthly periodical was also discontinued, after being kept
up more than five years.
" In 1850 I published the mathematical wrork to which this
account of my humble life is intended to be attached. As the
work was published in Calcutta, I requested a friend of mine
there to present copies of it to distinguished men in that city ; but
the reviews published in some Calcutta papers were generally
unfavourable to the publication." In another letter Eamchundra
says, " "When I composed my work on ' Problems of Maxima and
Minima,' I built many castles in the air ; but Calcutta reviewers,
&c. destroyed these empty phantasms of my brain." He also
describes himself as subjected to kind rebukes from some of the
best friends of native education in the North-West Provinces, for
his ambition in publishing his work in 'English.
" During the examination vacation in 1851, having obtained
three months' leave from the college, I went down to Calcutta, of
which I had heard much, and which I was very desirous of seeing.
When I arrived there, I happened to read a number of the Cal
cutta Review, in which a very unfavourable notice was given of
my work. My friends then advised me to write an answer to
it, which I did, and the editor of the Englishman very kindly
published it in his paper.
" Dr. Sprenger, who was formerly principal of the Delhi Col
lege, introduced me to the Honourable D. Bethune, of the
Supreme Council, who very kindly received from me thirty-six
copies of my work, and paid me 200 rupees as a donation." It
should be noted that Hamchundra had published the work entirely
at his own expense. " I afterwards learned that he sent a
number of these copies to England."
After mention of the correspondence, &c. described at the begin
ning of this Preface, Bamchundra proceeds as follows :—
" The honourable members of the Court of Directors were
pleased to confer honours upon me, and the Government in this
country sanctioned a khillut (dress of honour) of five pieces, which
I am told I will obtain at Delhi, and also a reward of 2,000
rupees, which I have already received at the hands of Captain
Eobert Maclagan. I am much thankful to the English Govern
ment that they are so bent upon encouraging science and know
ledge among the natives of this country, as to take notice of a
poor native of Delhi like myself.
" The most important event of my life, at least what I consider
to be as such, was, that by God's unsearchable and gracious Provi
dence I was brought to the knowledge of the Saviour. After I
had finished my mathematical work, and before I went down to
Calcutta on leave, I had become a believer in the Gospel. Before
this belief had taken possession of my heart, there were two erro
neous notions in my head (and which I believe must ever be in
the heads of nearly all native youths educated in Government
colleges and schools, as long as the system of instruction conti
nues to be pursued as it is till now)." The first of these notions
was that the English themselves did not believe in Christianity,
because they did not, as a Government, exert themselves to teach
it. The second was that a person who believes in one God stands
in need of no other religion. I omit the details of Bamchundra's
reasoning, because this publication is expressly intended for India
as well as England, and because I do not feel authorized to intro
duce into a work published by the late and present Government
of India, what might originate a discussion on a most difficult
question of Indian policy. Bamchundra proceeds thus :—
" Both of these erroneous notions were dispelled in the following
manner. Once a Brahmin student was sent by an English officer
from Kotah to the Delhi College, and was recommended to the
principal's notice. This stranger in Delhi waited to see the church
during divine service. The principal, Mr. Taylor, also requested
me to go with the Brahmin student to see the divine service in
the church, if I liked. And thus, out of mere curiosity, we went
there, and saw several English gentlemen whom I respected as
well-informed and enlightened persons. Many of them kneeled
down, and appeared to pray most devoutly. I was thus unde
ceived of my first erroneous notion, and felt a desire to read the
Bible. Mr. Taylor recommended me first to go through the ISTew
Testament. I commenced it, and read through it with attention ;
and thus I became aware that salvation is not merely in knowing
that there is one God, and that polytheism and idolatry are false,
but that it is in the name of our most blessed Saviour, the Lord
Jesus Christ ; and in this manner I was cured of my second error.
I afterwards read the English translation of the Koran, by Sale,
the Geeta in English, and had conversations and discussions with
those who knew these books in the original languages ; and at last
I was persuaded that what is required for man's salvation was
in Christianity, and nowhere else. I then read many Christian
books, together with some treatises of Hindooism and Mohame-
danism, and had frequent discussions with the professors of each
of them, but particularly with the latter. But the final step of
baptism was difficult for me to take ; for by this I was sure to
lose caste and dissolve all family connection, &c. ; and therefore
I wished to believe that baptism and a public profession were not
necessary for becoming a Christian. When I went down to Cal
cutta, Mrs. very kindly gave me a letter to the late Professor
Sturt, of the Bishop's College there ; and when by means of this
letter I was introduced to him, though he gave me reasons for the
necessity of baptism according to the Gospel, I very obstinately
did not agree with him. Near the end of March, 1851, 1 returned
to Delhi, and for more than a year I remained in great distress of
mind, until the 11th of May, 1852, when I and the late Sub-
Assistant Surgeon Chimmun Lall, who had formerly obtained
some Christian knowledge in Calcutta, were, by God's special
grace, brought to submit to baptism by the late Hev. — Jennings,
chaplain of Delhi."
Eamchundra continued as teacher at Delhi College, the princi
pal of which was Mr. E. Taylor, of whom he speaks in terms of
the highest gratitude and respect. Mr. Taylor was one of the
victims of the mutiny, as was also Chimmun Lall, just mentioned.
" The mutineers also inquired after me ; but my younger brothers,
who are as yet Hindoos, concealed me in the female apartments
of my family's house, in a lane, and my neighbours and acquaint
ances were kind enough not to betray me. On the evening of the
third day, that is, on 13th May, 1857, when it was dark, I
escaped out of the city, accompanied by two faithful servants, who
took me to the village of Matola, about ten miles distant from
Delhi, I remained in this village about a month, in great danger
of being betrayed by those who were opposed to the zemindar who
had very kindly lodged me in his house. Here I daily used to
persuade the zemindars that it was wrong that the English were
gone for ever, by telling them the vast resources, the power, and
the knowledge of the English nation. On 10th June, 1857, a
body of mutineers passed by this village, and some one told them
that a Christian was living in it ; but my old servant was warned
of this a few minutes before : he awakened me, and told me of my
danger. At first I hid myself in the zemindar's cottage, expecting
to be found out and killed ; but a very prudent Brahmin zemindar
advised me and my servant to fly to the jungles before the muti
neers could arrive. "We did so ; but before we could ran three
quarters of a mile, we heard a great noise in the village, bullets
were whistling about us, and horsemen appeared to be in our
pursuit, for the noise of galloping was distinctly heard. I then
rushed into a thorny little bush, not minding the thorns that
went into my flesh. By God's merciful providence the mutineers,
after plundering and giving a good beating to the zemindars, &c.
with whom I lived in the village, did not penetrate into the
jungle, but went their way towards Delhi. When there was
quiet towards the village, I and my old Jaut servant traversed
the whole juugle, and with great difficulty reached the English
camp on the 12th June, 1857. Here I was employed as an
English translator of daily news from Delhi, for the information
of the general and other commanders, and remained in the camp
till the capture of Delhi on the 20th September, 1857. In Janu
ary, 1858, 1 was appointed as native head master in the Thomason
Civil Engineering College at Eoorkee, on 250 rupees a month ;
which situation I held for eight months, and in the beginning of
the present month, September, 1858, I was appointed as head
master of the school (not a college) which is being organized at
Delhi."
Having thus given the reader the account which he will

naturally expect of the reasons for this publication, and of the
author of it, I leave those reasons to his attentive consideration,
and that author to his kindly criticism, and to the interest which
must be excited in the mind of any one who is capable of feeling
curiosity about the history of human progress, by the revival in
India, fostered by Europeans, of speculation on one of the sciences
for which Europe is indebted to India.
A. DE MORGAN.
Uniyeksity College, Lohpok,

January 17, 1859.
A TREATISE
ON
PBOBLEMS OF MAXIMA AND MINIMA,
SOLYED BY ALGEBRA,
BY BAMCHUNDKA,
TEACHER OF SCIENCE, DELHI COLLEGE,
" The problems which relate to the Maxima and Minima, or the greatest or least values
of variable quantities, are among the most interesting in the Mathematics; they are
connected with the highest attainments of wisdom and the greatest exertions of power ;
and seem like so many immoveable columns erected in the infinity of space, to mark
the eternal boundary which separates the regions of possibility and impossibility from one
another."
2nd Diss. Ency. Brit.
CALCUTTA:
PRINTED BY P. S. D'ROZAKIO AND CO., TANK SQUARE.
1850.
PREFACE,
For the last four or five years I was desirous of solving

almost all problems of Maxima and Minima by the principles
of Algebra, and not by those of the Differential Calculus.
All those problems which brought out equations of the
second degree were of course easily solved by the method of
imaginary roots given in some works on Algebra, particu
larly in Wood's Algebra by Lund, and the Encyclopaedia
Metropolitana. But even these problems in several cases
required particular artifices, without which it was impossible
for me to solve them. All these problems are solved in the
first chapter of this little work. Besides the method of
imaginary roots, I have given another, quite independent
of Imaginary quantities, quantities which to many beginners
of Mathematics, appear somewhat mysterious and unintel
ligible. This latter method I may venture to call a new
method, because in all mathematical works which I have had
access to, I have never seen a single problem of Maxima or
Minima solved by it, though it is used to reduce an adfected
quadratic to a pure one in a great many works on Algebra.
Thus far I have spoken of the first chapter.
All the problems solved in the second chapter bring out

cubic equations, the solution of which on the condition of
Maximum or Minimum, required a new method, which I
could not find, though I searched for it in several works
enumerated hereafter. I then resolved to find out a method,
and in intervals of leisure during three years I continually-
thought on the subject, and at last found it out. This is a
method which appears extremely simple and easy, though it
bafHed all my endeavours for the space of three years. I
may call it new, for I did not find it in any book I looked
into.
The third and fourth chapters, and the supplement con

tain problems and general solutions of particular equations
of the fourth, fifth, and the sixth degree, together with those
problems in which two or three variable quantities enter.
The methods used in these parts of the work, though more
difficult and intricate than that used in the second chapter,
were easily discovered.
This work contains about 130 problems taken chiefly

from the following works : Simpson's Fluxions, Hall's Dif
ferential Calculus, Gregory's Examples, Conner s Differential
Calculus, Walton's Differential Calculus, Ritchie's Differen
tial Calculus, Young's Differential Calculus, Encyclopaedia
Britarmica, Hirsch's Geometry, works on Mixed Mathema
tics, &c. Besides the problems solved here, many more may
be solved by the methods given in this treatise.
I have also given definitions, formulae, and propositions

necessary for the study of this work in the Introduction.
In conclusion, I flatter myself with the hope that my
labours will be of some use to those Mathematical students
who are not advanced in their study of the Differential
Calculus, and that the lovers of science, both in India and
Europe, will give support to my undertaking.
Owing to the necessity of having the work printed in

Calcutta, and my consequent inability to superintend the
sheets passing through the press, many errors, almost insepa
rable from a work of this nature, have unavoidably crept in ;
for these I must beg the indulgence of my readers.
RAMCHUNDRA.
Delhi,
16th February, 1850.
TABLE OE CONTENTS.
Page
Introduction , 1
Chapter I.—Problems of Maxima and Minima in the

solution of which simple and quadratic
equations are used 12
Chapter II.—Problems of Maxima and Minima in

the solution of which cubic equations
are used 80
Chapter III.—Problems of Maxima and Minima in

the solution of which equations of the
fourth, fifth, sixth, and seventh degrees
are used 127
Chapter IV.—Problems of Maxima and Minima in

which two or more variable quantities
are used 151
Supplement 1 78
INTRODUCTION.
(1.) REDUCTIONS OF EQUATIONS.
[Definitions.]
1. An equation is an algebraical expression of equality
between two quantities.
2. A root of an equation is that number, or quantity,
which, when substituted for the unknown quantity in the
equation, verifies that equation.
3. A function of a quantity is any expression involving
a —\~ x
that quantity ; thus, ax2 +. b, &c. These functions are
x
usually expressed by / (w) .
Prop. Any function of x, of the form xn -f- poo0--1 -f-
qxn~2 + &c, when divided by x — a or x + a, will leave a
remainder, which is the same function of a or — a that the
given polynomial is of x.
Let/ {x) —xll-{-pxn~~l-\-qxn~'2-{- &c. ; and, dividing/ (a?)
by# — a or x-\-a, let Q denote the quotient thus obtained,
and R the remainder, which does not involve x ; hence, by
the nature of division, we have / (x) = Q (x — a) + R or
/ (x) = Q (x + a) + R- Now these equations must be true
for every value of x ; hence, if x = a the first equation
becomes / (a) = R and if x = — a the second equation
becomes / (— a) =R, and hence it appears that / (a) or R is
the same function of a as the given polynomial is that
of x. If / (x) = 0 and a be a root of this equation, then by
definition (2) we must have / (a) = 0 or i? = 0, and hence
B
( 2 )
m.= -ô or i^-= -°-=0 or Q is in both

off — a off — a off -\-a off -\-a
cases = 0.
Ex. Let / {off) = off3— off*-\- r = 0 and —■ a a root of this
equation :
off + ajofft — off*+r [ off*— {a + I) off +[a {a+l)
offZ+aoff* I = Q=0 Ans.
- (a + 1) #2
— (a + 1) #2— a (a -f 1) <#
a (a + 1) ^ + r
a (a + 1) a? + o* (« + 1)
r-a*{a + I)=R = 0.
This last equation expresses the condition of a, being a
negative root of the given equation.
(2.) TO FIND THE EQUATION TO THE PARABOLA. (Fig. 1.)
Let a point S be taken without the right line CB, and let
the indefinite line Sm revolve about the point S in the plane
SBC ; also, let Cm, which is perpendicular to CB, cut Sm
in m ; then, if Sm be always equal to Cm, the locus of the
point m is a parabola.
Through fif draw BSP at right angles to CB, and if SB
be bisected in A, the curve will pass through A, as appears
by the construction; draw mP perpendicular to BP, and let
APôc, Pm = y, AS = a; then SP*+Pm* = {Sm*= Cm*) =
BP*, or {off— a)*-\-y* — {off+a)*; that is, Off* — 2aoff + a* + y*=z
Off* + 2aoff+ a*, or y*=4ao?. This equation is called the equa
tion of the Parabola, because it expresses the relation between
the lines AP & Pm which determine the position of points
on the curve.
( 3 )
(3.) TO FIND THE EQUATION TO THE ELLIPSE. (Fig. 2.)
Let two indefinite lines Sm, Hm, revolve, in a given plane,

about the points S, H, and cut each other in m, in such a
manner that Sm -f mH may be an invariable quantity;
then the locus of the point m is an Ellipse. Bisect SH in
C, and from m draw mP perpendicular to SH, or SH pro
duced; let CP = x, Pm = y, CS = c, Sm + Hm = 2 a.
Then a/>SP2+ Pm8= Sm, and \/HP*+ Pm* = Hm ; there
fore \/£P2 + -P^2 + VHP2 + Pm2 = £m + flw, or
a/(c — #)8 + y2 + a/(c + ^)2 + ?/2 = 2a : hence \/(c— x)2-\-y2
= 2« — \/(c + ^)+^ and squaring both sides, tf — lcx + xf1
+ y« = 4 a* — 4 a \/(c + <^)2 + «/2 + c2 + 2 c# + #* + yg ;
that is, by transposition, 4a2 + 4 ex = 4a \/(£ + <^)2 + V2
or a2 + £# =# a/ (c + «^)2 + y2; and again squaring both
sides, a4 + 2a2c# + c2^2 = a2c2 + 2a2cx + a2^2 + a2?/2, or
a*y2 = a4 - a*d* - (a2 - c2) at*-, let a2 - c2 = b~, then
62
a2«/2 = a262 - 62^2, and y2 = — (a2 - <r2) ; this equation is
CI
called the equation of the Ellipse, because it expresses the
relation between the lines cP and Pm, which determine the
positions of points on the curve.
(4.) TO FIND THE EQUATIONS TO THE ELLIPSOID,

SPHEROID, AND SPHERE. (Fig. 3.)
An Ellipsoid is a solid figure, such that sections of it per

pendicular to its three axes are all Ellipses, and consequently
its three axes are unequal.
A Spheroid is a solid figure, generated by the revolution of
an Ellipse about its major or minor axes, and consequently
two of its axes are equal to each other, and sections of it
( 4 )
perpendicular to the axis, about which the* revolution is con
ceived to be performed, are all circles.
A Sphere is a solid figure, generated by the revolution of
a circle about one of its diameters. Figure 3 represents the
eighth part of an Ellipsoid.
AB is part of the Ellipse in the plane wy
AD xz
BD yz
And the section QPR parallel to ocy is also an Ellipse.
The surface may be conceived to be generated by a vari
able Ellipse CAB moving upwards parallel to itself, with its
centre in CZ. Let nQR be one position of this variable
Ellipse; and let
Cn = z CA = a nR = oox
nm = oo CB = b nQ = y1
mP = y CD = c
then from the Ellipse QPR we have
— + Hi = i
oo* y*
Also from the Ellipses DRA and DQB we have
oo* z* _ ,y* z* .
a* c* b* c*
X* V *
therefore -y = ^ ; and, multiplying the first equation by
oo* ., _ w« . a?2 , y* w* , ^2
— or its equal ff, we have — + f- = -~ = 1 5
a2 ^ 62 ' a2 b* a* c*
00* v* z*
. ' . -t; + t^ + -rr = 1 equation to the Ellipsoid :
a* b* c*
let a = # .•. 00*
— + II* z* = 1 equation to the Spheroid;
^ -f —
cl a c
$>% _j_ 2/2 4- £2
let 0 = J = c .\ ~ = 1 equation to the Sphere.
( 5 )
(5.) TO FIND THE AREA OE A TRIANGLE. (Fig. 4.)
Rule 1st.—Multiply the base by the perpendicular height,

and half the product will be the area. The truth of this rule
is evident, because any triangle is the half a parallelogram
of equal base and altitude, by Euclid, prop. 41, 1st Book.
Rule 2nd.—When the three sides are given : add all the
three sides together, and take half that sum. Next, subtract
each side severally from the said half sum, obtaining three
remainders.
Lastly, multiply the said half sum and those three remain
ders all together, and extract the square root of the last pro
duct, for the area of the triangle. For let a3b3c, denote the
sides opposite respectively to A,B,C, the angles of the tri
angle ABC; then by prop. 13, of Euclid, book 1st, we have
BC* = AB* + AC2 - 2AB. AP, or a2 = ¥ + c2 - 2c. AP
ffi JL- (f> q2
or AP = : hence we have
2c
CP2 = b2 _ (62 + c2-a2)2 = 4ffc*- (b2 + c2-a2)2
4c2 4c2
- (%fo + b2 + c2 - a2) (2bc - b2 - c2 + a2)
"" 4c2
.-. 4c2 CP2 = { {b + c)2 - a2 } { a2 - (c - b)2 }
= (a + b + c) (— a + b + c) {a — b + c) (a + b —. c)
a — b + c a + b — c } ,— r—p ^w r
g 2 J=V5(^«) (5-6) (s-c)
where s ~ \ (a + b -\- c) = half the sum of the three sides.
( 6 )
(6.) TO FIND THE DIAMETER AND CIRCUMFERENCE OF ANY

CIRCLE, THE ONE FROM THE OTHER. (Fig. 5.)
This may be done by the following proportion, viz. As

1 is to 3*1416, so is the diameter to the circumference. For,
let ABCD be any circle, whose centre is E, and let AB, BC,
be any two equal arcs. Draw the several chords as in the
figure, and join BE; also draw the diameter DA, which
produce to F, till BF be equal to the chord BD. Then the
two isosceles triangles DEB, DBF, are equiangular, be
cause they have the angle at D common; consequently
DE : DB:: DB : DF. But the two triangles AFB, DCB, are
identical, or equal in all respects, because they have the angle
F == the angle BDC, being each equal to the angle ADB,
these being subtended by the equal arcs AB, BC; also the
exterior angle FAB of the quadrangle ABCD, is equal to
the opposite interior angle at C ; and the two triangles have
also the side BF = side BD ; therefore the side AF is also
equal to the side DC. Hence the proportion above, viz.
DE : DB:: DB : DF=DA + AF becomes DE : DB ::DB :
2DE + DC. Then by taking the rectangles of the ex
tremes and means, it is DB2 = 2DE* + DE. DC. Now if
the radius DE—1, this expression becomes DB*=2 + DC,
and hence DB = \/2 + DC. That is, if the measure of
the supplemental chord of any arc be increased by the num
ber 2, the square root of the sum will be the supplemental
chord of half that arc. Let AC = a side of the inscribed
regular hexagon = 1 ,\ DC =:V AD* -AC2 = \Z~22 ~ 1
= V 3"= 1*7320508076, the supplemental chord of \ of the
periphery. Then, by the foregoing theorem, by always
bisecting the arcs, and adding 2 to the last square root, there
will be found the supplemental chords of the 12th, the 24th
the 48th, 96th, &c, to the 1536th part of the periphery; thus
( 7 )
it is found that 3*9999832669 is the square of the sup
plemental chord of the 1536th part of the periphery ; let
this number be taken from 4, the square of the diameter,
and the remainder = 0*0000167331 .\ ^ 0*0000167331 =
0*0040906112 = xirVe of the periphery; this number then
being multiplied by 1536, gives 6*2831788 for the perimeter
of a regular polygon of 1536 sides inscribed in the circle =
the circumference very nearly when the diameter of the
circle = 2.
(7.) THE AREA OF ANY CIRCLE = RECTANGLE OF \ CIRCUM

FERENCE AND \ ITS DIAMETER. (Fig. 6.)
Conceive a regular polygon to be inscribed in a circle;

and radii drawn to all the angular points, dividing it into as
many equal triangles as the polygon has sides, one of which
is ABC, of which the altitude is the perpendicular CD from
the centre to the base AB.
Then the triangle ABC is equal to a rectangle of half the
base AD and the altitude CD; consequently, the whole
polygon, or all the triangles added together which compose
it, is equal to the rectangle of the common altitude CD,
and the halves of all the sides, or the half perimeter of the
polygon.
Now, conceive the number of sides of the polygon to be
indefinitely increased ; then will its perimeter coincide with
the circumference of the circle, and consequently the altitude
CD will become equal to the radius, and the whole polygon
equal to the circle. Consequently, the space of the circle,
or of the polygon in that state, is equal to the rectangle of
the radius and half the circumference. Q.E.D.
( 8 )
(8.) EVERY SPHERE IS TWO-THIRDS OF ITS CIRCUMSCRIBING

CYLINDER. (Fig. 7.)
By prop. 12 of Euclid, Book 12th, the cones AIB and
QIM are in the triplicate ratio of IF and IK, that is to say
we have this proportion-
Cone AIB : cone QIM :: IF3 : IK3 :: FH3
: {FH-2FK)3
.-. Cone AIB : frustum ABMQ :: FH* I FH3
- (FH-2FK)3
:: FiJ3 : 6FII2FK- 12 FHFK2 + SFK3 but cone
-4IB = one-third of the cylinder ABGE, hence;
Cylinder AG : frustum yiJBMQ :: SFH3 I 6FH\FK-
12FH.FK2 + 8FK3.
Now cylinder AL : cylinder -4G :: FiT : ZZ".
.-. Cylinder -4L : ABMQ::6FH2 : 6FH2- 12FH.FK +
SFK2, (1)
Now it is evident that IK = KM.-. IK2 + KN2 =
KM2 + iCV2 = JiV2 = IG2 = JEZA Now circles are to each
other as the squares of their diameters, or of their radii;
therefore the circle described by KL is equal to both the
circles described by KM and KN ; or the section of the
cylinder is equal to both the corresponding sections of the
sphere and cone. And as this is always the case in every
parallel position of KL, it follows that the cylinder EB,
which is composed of all the former sections, is equal to the
hemisphere EFG and cone IAB, which are composed of all
the latter sections. By proportion (1) we find
Cylinder AL : segment PEN:: 6FII2 : 12FHFK-8FK2div.
lliFH* I FK{SFH-2FK)
But cylinder AL = circular base, whose diameter is AB
or FH multiplied by the height FK; hence cylinder AL=
circle EFGHx FK. "
( 9. )
EFGH
.-. Segment PFN = §. circle FZT (SFH - 2FK)
FK* . (2)
If FK = FH3 then the sphere = § cylinder. Q.^.D.
Note.—For the cylinder AL = frustrum ABMQ + segment PJPiV
and .'. cylinder AL — frustrum ABMQ = segment PFN.
(9.) TO FIND THE AREA OF AN ELLIPSE. (Kg. 8.)
The equation to the Ellipse is y = - n/ a2 — ^2 and to the
circle described on the major axis as diameter is y1 — %/ a2— a?21.

Comparing these two equations we find
y = -y1 ox 2 a y = 2byx, and .' . y \ yl : : '26 .' 2a.
In the diagram annexed 2a — A1A, 2b = BlB}% = any
of the lines or abscissas measured on the line CA or
CA1 from the point C, y = any of the perpendicular lines
denoted by pm or ^1m which are called the ordinates of
the Ellipse, and y1 = any of the perpendicular lines
denoted by Pm or P1m which are called the ordinates
of the Circle. Now if the area of the Ellipse and Circle
be supposed to be divided into bands perpendicular to the
axis major A A1, by ordinates Ppm, placed so closely to
gether that the arcs of the curves between them may be con
sidered to be straight lines, the areas of the spaces of the
Ellipse and Circle between every pair of contiguous ordinates
will be proportional to those ordinates, and as all the ordi
nates are in the same ratio, the sum of all the areas between
the elliptical ordinates, that is, the area of the Ellipse itself,
will be to the sum of all the areas included between the cir
cular ordinates, that is, to the area of the Circle itself, as any
elliptical ordinate is to the corresponding circular ordinate,
that is, as the axis minor of the Ellipse is to its axis major,
c
( io )
By article 6th we find that the circumference of the Circle
described upon the major axis is to its diameter as 2 is to
6*2831 &c. or 1 : 3*1415 &c. (which let=p) :: 2a : circum
ference = 2pa .' . -J circum. = pa and a = semi-diameter .\
the area of the Circle = pa x a = pa2, we therefore find area
of the Ellipse : pa2 : : 2b : 2a . • . area of the Ellipse = pab.
(10.) TO FIND THE SUM OF U TERMS OF THE SERIES

1 + 4 + 9 + 16 + 25 + n2.
Assume 1+4+9+16+25+ n2 = Pn? + Qn2 +
Rn + S, and since there are four co-efficients to be deter
mined^ we must have a corresponding number of indepen
dent equations; hence
when n=l we have P + Q+P+$=l
n=2 8P + 4Q + 2P + /S=l + 4=5
n=S 27P + 9Q + 3P + £=l + 4 + 9 = 14
rc=4 64P+16Q + 4P + £=l + 4 + 9 + 16 = 30.
And from these four simple equations we find, by continued
subtraction, P=^ Q=<b R=i an(l #=0; therefore the
sum of 1 + 4 + 9 + 16 + 25 + n2 = ±n3 + in2 +
i^ = n /rt o + 3rc
- (2n2 « + -.v
1) = n(n
-^—-+ 1) -—-—J
(2^+1) . TjC. .. sup-
If ^ be
posed to be indefinitely great, ^ and 2 ft may be put instead

of (ra+1) and (2^ + 1) and .*. in this case the sum of the
nZ (A \
series = — \A.)
o
( 11 )
(11.) TO FIND THE AUEA OF A PARABOLA. (Fig. 9.)
The equation to the parabola is y2 = 4&aos and consequently

we have the following equations :—
AH2
Kf = 4>aAK.\ AW = 4aHp or Hp =
4a
AG*
Ln2 = AaAL .-.AG* = 4>aGn or Gn = -— &c. = &c.
4a
^F2 = 4aFr or Fr = ^=- &c. = &c. = &c.
4a
AD
Let AH=HG = GF = FD = &c. and each = —
^4il2 ,„. ^#3 AD*
.\iect.HK=Hp x AH=~ x AH =
4a 4a 4a^3
4#G2 4^Z>3
rect. Gq = HG x Gn^HG x
4a 4a^3
rect. Fw= GF x Fr = GF x —A— = -;—r &c. = &c.

4a 4a#
. • . The sum of these rectangles = -—i + "1—£" "+" 9AD3
AD3 4<AD3 "7—T "^ ^c*
^■D2 AD (n+ 1) (2k + 1) __ r ^ (» + l) (2» + 1)

4a X n2 2. 3 "" X n* 2. 3
It is evident that if the number of parts into which the
line AD is divided be infinitely great, the sum of the rect
angles must be equal to the area ApnrCD and also by art.
10, equa. (A) ^——L \—_—L — ._ ... the area Apnr CD
&• O O
DC x AD n* DC x AD ^ A „_ „ , ,
== x — = ^ .*. the area ^/mr CD ot the
n3 3 3
parabola = rect. ^4D x AB — .4/wr CD = DC x AD —
DCxAD 2DCxAD
- = -—~ . Q.E.D.
A TREATISE ON PROBLEMS OF MAXIMA AND MINIMA
SOLVED BY ALGEBRA.
CHAPTER I.
Problems in the solution of which simple and quadratic
equations only are used.
PROB. (1.) TO DIVIDE A GIVEN NUMBER INTO TWO SUCH
PARTS THAT THEIR PRODUCT MAY BE THE GREATEST
POSSIBLE.
Put the given number = a, one of the parts required =

oos and consequently a — x = the other part, .'. x (a — x) =
ax — x2 = product = maximum, which let = r . * . x2 — ax
= — r. Solving this quadratic equation we find x =
Cb / (ft
- d= A/ —- — r. Now it is evident that r cannot be greater
than — for if it be so, the value of x becomes impossible ;

a2 =
therefore the product ax - /J or r is greatest when —
r> and .•.# = —-
The same solved without impossible roots.

In the expression ax—-x2 which is to become a maximum,
let x = y + h where the value of y determined by the
condition of ax — x2 being a maximum, will show whether

it is positive, zero, or negative. "We now find
a2 a2 a2
ax — x2 = ay -{- —— y2 ■— ay — — = -r — #2, which is evi
dently a maximum when y = 0, .•.#=- as before.

( 13 )
PROB. (2.) TO DETERMINE THE GREATEST RECTANGLE

INSCRIBED IN A GIVEN TRIANGLE. (Fig. 10.)
Let the base AC of the given triangle = b, and its alti

tude BD = a, and let the altitude B8 of the inscribed rect
angle mc (considered as variable) be denoted by x, Then,
because of the parallel lines AC, ac3 we find the proportion,
BD : AC : : DS : ac or a \ b : : a — x : ac or ac
• ; whence the area of the rectangle or ac x BS
a
= bax —* boo = b_ (aoo _ wv\ — max. It is evident that
a a K J
when a quantity is a maximum, any determinate part, multi-*
pie or power of it must also be a maximum, and consequently
the determinate ax — x2 of -a (ax
v — x2)' must be = max.
which let = r .*. ax — x2 = r or x2 — ax = — r. Solving
this quadratic equation we find,
/y /ft2
a? = - d= a/ -— r, and it is manifest now that ax •
or r cannot be greater than -j- (for the reason stated in the

last problem), and, therefore, when r = max. we must have r
a2
= —..•,# a Whence the greatest inscribed rectangle is
= -.
that whose altitude is just half the altitude of the triangle.

In the expression ax—x2, which is to become a maximum,
let x - y + £ .'. ax - x2 = a (y -f- |) - (y + |)2 = ay +
a? a2 a2
~2~~~y2"~'ay~~~T:=~A— V2> which is evidently = max.
when y = Q, or x = - as before.
( 14 )
PROB. (3.) OF ALL RIGHT-ANGLED PLANE TRIANGLES HAVING

THE SAME GIVEN HYPOTHENUSE, TO FIND THAT (ABC)
WHOSE AREA IS THE GREATEST POSSIBLE. (Fig. 11.)
Let AC = a, AB = x and BC = y. Then, x2 + y2 being

= a2 we shall have y = \/ a2 —- x2, and consequently
■—■ = - \/fl2 — #2 = the area of the triangle = max. and
CuOu -— X4*
consequently the square of the area, or —— = max.
and also four times this, or a2oc2 — x4* = max. which let = r.
.*. x* — a2<2?2 = — r. Solving this quadratic equation we
a2 faF
find x2 = ~ ± A/ — — r, and it is manifest that a2x2 — x*
or r cannot be greater than — ; therefore when r = max. we
must have r = —, . • . x2 =
?,.r — and a? =
= -anu, /-, and y = \/i
= v1)
a
= •— . Hence it appears that the right-angled plane tri
angle contains the greatest area whose two sides containing
the right angle are equal to each other.

In the expression a2w2 — x4, which is to become maximum
let x2 = y2 + ~ .\ a2x2 - x* = a2 (y2 + ~) - (y2 + ^)2
cfi — y* — a2y2 — —■
= a2y2 + --- eft* = -eft j/4, which is evidently
= maximum when y4* = 0, and .\ x2 = ~ or x — /- as

before.
( 15 )
PR0B. (4.) OF ALL RIGHT-ANGLED PLANE TRIANGLES CON

TAINING THE SAME GIVEN AREA, TO FIND THAT WHEREOF
THE SUM OF THE TWO SIDES, AB + BC, IS THE LEAST
possible. (See Fig. 11.)
Let one leg AB, be denoted by wy and the area of the
2d
triangle by a ; then the other side will be denoted by —, and
00
2a
the sum of the two legs will be oo + — = minimum, which
° oo
let = r ,\ oo1 — roc = — 2a (1.)
V
Solving this quadratic equation we find w = - dtz
v T
-— 2a, and it is evident now that r cannot be so
small as to make -r less than 2a; therefore, when r =

4 '
a»2 . y
min. we must have —• = 2a ,\ r = 2 v 2a and oo = -
4 2
= VYa = AB. Whence BC = — is also = \/2«. There
at
fore the two legs are equal to each other.

From equation (1) in the preceding solution we have #?2—
r
too = — 2a, and .\ roo — oo2 = 2a. Let a? = «/ + « .'. ^^
= 2a, or -— y2 = 2a, .'. r2 ~ 8a + 4y2. Now it is evident

4
r or r2 is the least possible when y = 0, . * . r = 2 v2a and
a? = \/%a as before.
( 16 )
PROB. (5.) DIVIDE A GIVEN LINE, AB, INTO TWO PARTS, SO

THAT THE SUM OF THE AREAS OE THE SQUARES DESCRIBED
ON THESE PARTS SHALL BE THE LEAST POSSIBLE.
Let a = the given line, x one of the parts, then a — x will

be the other part. Then, x2-\- {a— x)2 is a minimum, that is
2x2 + a2— 2ax is a minimum. Now a2 is a given determinate
quantity and therefore when 2x2 + a2 — 2ax = minimum
we must also have 2x2 — 2ax = minimum or its half, viz.
x2 — ax = minimum, which let = r . • . x2 — ax = r. Solving
Ct / d2
this quadratic equation we find # = -:+: A/ r + —- Now
r can be less than zero, that is it may become negative;
but, when negative, it cannot be so great as to make the
radical impossible. Therefore, when the least possible, r must
become a negative quantity == — — and hence x = -. This
problem may be solved by the following method which is
more elegant.
Let 2x2 + a2 —• 2ax = r, . • . x2 — ax = -V — —
a (1)
Solving this quadratic equation we find,

a . /r T
Now r or - cannot be so small as to
= 2~ Vl
t* a
make - less than —, because in this case the radical quantity
becomes impossible; therefore when r is the least possible,
we must have -t = --
a2 and .•.# = -■a Hence the riven line
2 4 2 &
must be bisected.
In the equation (1) let x == y ~f ~ and therefore we find

( 17 )
(y +|)2 ~ a ^ + 1) = \ ~~ j or v% + \ = ^ and therefore

a2
r = 2y2 + 17. Now it is evident that r is the least possible
<v
when y or 2?/2 = 0, . • . x — -- as before.

Z
It must here be remarked that when in the solution of
problems of minima we leave out some given negative quan
tity, we sometimes make r, or the minimum quantity, less
than zero or negative, as is done in the first method of solu
tion of the preceding problem.
PROB. (6.) OF ALL CONES UNDER THE SAME GIVEN SUPER

FICIES (s) TO FIND THAT ABB WHOSE SOLIDITY IS THE
GREATEST. (Fig. 12.)
Let the radius of the base AC = x} and the length of
the slant side AB = y, and let p denote the periphery
(3*14 &c.) of the circle whose diameter is unity. Then the
circumference of the base will be 2px, the area of the base
= px2} and the convex superficies of the cone = pxy (which
last is found by multiplying half the periphery of the base
by the length of the slant side) . Wherefore, since the whole
g
superficies is = px2 -f pxy = s, we have y = x;
2s
whence the altitude CB = \/AB2 — AC2 — A/-4-, —
V p"W
vx p '
'DX
which
ch multiplied
multiplied by ±—, or | of the area of the base, gives
o
TJ'JT / s 2s
-7z~ A / -o-s for the solid contents of the cone : which
3 V p2x2 p
2lQS
being a maximum, its square -J- -{s2x2 — 2psx4) = —r—
( 18 )
( — x2 — x*) must also be a maximum. Since —-- is a con-

\2p 9
s
stant given quantity, therefore — x2 — x4 must also be =
s x2 — x* — r, and x4 — —
maximum, which, let = r . • . — s
#2 = — r. Solving this quadratic equation we find x2 =

s
•7-=*= A//~lp .
TTT-o — r, .*. whenr = max. itmustbe s2 .*.
= —t-^
4p V 16j92 16/r
#2 = 4p
— and 5? = a/ —.
V 4p* Now vy = px
— — a? = /s
4p
Vs ___ v^s / s __ 4\A — \A __ 3\/s
f . __ # Hence
4» ~ pV
4p V Am ~~
V 4p x/lUi
v^ ~~ 2\/p
2a/^'
it appears that the greatest cone under a given surface (or a
given cone under the least surface) will be, when the length
of the slant side is to the semi-diameter of the base in the
ratio of 3 to 1, or (which comes to the same thing) when the
square of the altitude is to that of the whole diameter in the
ratio of 2 to 1,
The same solved without impossible roots,

s s
In the expression — x2 — x* = max. let x2 = — + y .'.
s 2 4 _ ^2 s Is \2 _ s* 1 s s*
""^^""^"Tp""^2" max' when y = 0y •'• ^ = ^

and # = A/ 7- as before.
( 19 )
PROB. (7.) TO DETERMINE THE POSITION OP THE RIGHT

LINE DE, WHICH, PASSING THROUGH A GIVEN POINT P
SHALL CUT TWO RIGHT LINES AR AND AS, GIVEN IN
POSITION, IN SUCH SORT THAT THE SUM OF THE SEGMENTS,
AD AND AE, MADE THEREBY, MAY BE THE LEAST POSSI
BLE. (Fig. 13.)
Make PB, parallel to AS, = a, and PC, parallel to AR,

= b ; and let BD = x. Then by reason of the parallel
7
lines, we will have the proportion x \ a : ; b I CE = — :

X
Therefore AD -f AE = b + a + x H = minimum. Now

x
b -\~ a, being a constant given quantity, x -\ is also a
x
minimum, which let = r, r.w-\ = r or x2 — rx = ~ ab.
x
T / r2
Solving this quadratic we find x = - zh A/ ~- — ab or r =
2\/ab and x = - = \/ab.

Since x2 — rx = — ab, we find rx — x2 = ab. Let
® = y + gr ••• ™? - %2„ = ry + —
r2 - (y
, + -J
r\2 = ry + ~
r2 -
y2 — n/ — — = —• — y2 = ab or r2 = 4#6 -+- 4y2 = min.
when y = 0 and therefore r = 2 v«§ and # = - = Va^

as before.
20 )
PROB. (8.) IF TWO BODIES MOVE AT THE SAME TIME, FROM

TWO GIVEN PLACES A AND B, AND PROCEED UNIFORMLY
FROM THENCE IN GIVEN DIRECTIONS, AP AND BQ, WITH
CELERITIES IN A GIVEN RATIO, IT IS PROPOSED TO FIND
THEIR POSITION, AND HOW FAR EACH HAS GONE, WHEN THEY
ARE THE NEAREST POSSIBLE TO EACH OTHER. (Fig. 14.)
Let M and N be two cotemporary positions of the bodies,

and upon AP let fall the perpendiculars NE and BD ; also
let QB be produced to meet AP in C, and let MN be drawn :
moreover, let the given celerity in BQ be to that in AP, as
n is to m, and let AC, BC, and CD (which are also given), be
denoted by a, b, and c, respectively, and make the variable
distance CN—x: Then, by reason of the parallel lines NE
and BD, we shall have CB : CN : : CD : CE or b : x : : c \
CX
CE .'. CE = —. Also, because the distances, BN and AM,
b
gone over in the same time, are as the celerities, we likewise
have, n \ m \\ BN : AM or n \ m : : x — b \ AM, or AM—
—, and consequently CM ■=, [AC — AM)— a +
— — d __ — (by writing d = a H ). Whence MN2 =
CM' + CN' -CMx2CE= (d - —)* + a* - (d - — )

2 ex yo 2dmx m2x2 9 2cdx 2cmx2
n2 b nb
,
# + m2 ^ 2 cm
2cm j -r+ H r-
1 + __
nb ■ / n* nb
( 21 )
Now let the quantity without the brackets = Q, the co-effi-
d2
cient of x = A and m2 2 cm = B, and we shall
n2 nb
therefore find Q (x2 —- Ax + B) — minimum or x2 — ^4<# 4-
B = min. which let = r, and . • . x2 — Ax -\- B = r or x2
- Aa? = r - 5 (1.)
-42
Before solving this equation we must show that — is less
than B. Since c = CD and b = .BC .- . b v c and Z>2 -7 c2

. • . n2b2 v n2c2 (2.)
2 dm 2cd
Now A. = —n n b — 2nd (bmv 4- en)J • A2 —
2cm m2b + n2b 4- 2mwc * " 4
-2 + l + —r
nb
n2d2 (bm + en)2 . ^ Arf'W
(m26 + /&2# 4- 2mnc)2 and i* = m2b 4- waA 4- 2mwc
ft2d2 (m262 -f ?z262 + 2mrcfo) m ^ » n -, ™ ^2
—t^ ^ 4- 2mncy—-.
(mz# + w26 We therefore find B : —
4
m2&2 4- ^2#2 4- 2mnbc \ m2b2 4- 2mnbc 4- A2. Now as
= m2#2, 2mnbc — 2mnbcy and %2#2 -7 %2c2 by inequation (2) .'.
the third term of this proportion is greater than the fourth
A2 A2
.-. B is greater than — and .-. —— B = a negative quan
tity, and may therefore be supposed = — P. The equation
(1) gives x2 — ^b = r — S .'. oc = — ± /y/ r 4- -7 ■#
^4 =iz \/V — P. Now r cannot be less than P . • . r = min.

= —
2
^ ^4 wm£e£ 4- n2cd „ ,
when r — P .'. x = ~- = -57—;—2A , 0 ; from whence
2 m26 4- w26 4- 2mnc
BN} AM, and M2V are also given.
( 22 )
A
In the expression x2 — Ax -f B = min. let x = z/ + ~
.-. *• - Ace + 5 = (j/ + ^f - A (y + |) + B = f +

J2 J2 M M
Ay + - - Ay - - + B = f + B ~ -; hut- - B =
A2
— P .\ B - := P . \ w2 + P = min. which is the case when
4 *
^4 m^M + w2cc? , -
y = 0 .*. x = -r- = —^ 5! as before.
2 wra + wo + 2m%c
PR0B. (9.) LET THE BODY M MOVE UNIFORMLY, FROM A

TOWARDS Q, WITH THE CELERITY M, AND LET ANOTHER
BODY N PROCEED FROM B, AT THE SAME TIME, WITH THE
CELERITY 71. NOW IT IS PROPOSED TO FIND THE DIREC
TION BD OF THE LATTER, SO THAT THE DISTANCE MN
OF THE TWO BODIES WHEN THE LATTER ARRIVES IN THE
WAY OF DIRECTION AQ OF THE FORMER, MAY BE THE
GREATEST POSSIBLE. (Fig. 14.)
Let BC be perpendicular to AQ, and make AC = a, BC

= b, and BN = x. Therefore, if the position M, be supposed
cotemporary with N, we shall have
n I m :: x I AM .• . AM = — : whence CM = a, and
n n
r 7}tX
consequently MN = (CN — CM) = Vx2 — b2 h a —
71
max. which let =r, .*. \/x2 — b2 — r -\ a, and .\ x2 —

79 , mx so „ 2rmx mV rt 2am<2?
62 = ( r-j- n a)2 = r2 + n + —^
nL 2ar n
+ a, .• . 2ar — r-1 — oz — £r = 5—- x- — ™ x
nr n
( 23 )
• •. x2 - r—r1 oc = - g 5 — (1) and
2 2mn{a — r) m2n2(a — r)2 _ m2n2(a — r)2

m2 — ^2 (m2 — %2)2 ~~ (m2 — ^2)2
(2ar
^ — r2 — b2 — a2) n2(m2 — n2)
l and therefore # = mn(a
., . „ —\ — r)~
(m2 — ft2)2
/^4 (a — r)2 — n2b2(m2 — n2)
4-
(m2 — n2)2
Now it is evident that in order that this problem may be
possible, r must be less"* than a, and consequently r = max.
when rfi{a — r)2 = n2b2(m2 — n2)} for r cannot be taken so
great as to render n4(a — r)2 y n2b2(m2 —- n2)} and therefore
bv m2 — n2 , mn(a — r) mb ..
a — r == and a? = —5 ~* — /— s, and
n mr — w \/ mr — nr
CN=\/xJ~^J2 = /-^—9: Whence m : a :: 5iV : CN :
Radius : cosine iV.

It is also evident that this problem is impossible when
m z. n.

In the equation (1) let the coefficient of x = A and the
second member = B .*. x2 — Ax = B. Now let x — y +
^ .-. x2 - ^ = (2/ + jf - -4 (y + j) = y8 + -4y + -^
^2 ^2 jp
— Ay — = y2 — — = B, and therefore y2 = B + —
= rft
—(a — r)2
7—s— nrb2ot^(m2 — n2) .by substitution,
, ,., ,. , — r)2
.•. (a v0
(m2, — n?)1 J v
= y2
—K(m2
- — n2)2
- -f %2#2 (m2
i — %2)'-. and_ therefore
,, „ a — r =
* This is evident, because if r — a the root becomes impossible, and if

r 7 a, there can be no limit to its increase, that is, it cannot admit of being
a maximum.
( 24 )
or r = a
TV*
(B.'-^ + rfy(»'-rf)i Now it is evident that

w.4
r — max. when the quantity subtracted from a = min. which

can only happen when y = 0, . • . when r = max. we must
have r — a b \//—z
mL — r5 or a - r = hx/m? — n2 ana . • . #
n n
— A — mn(a
v — r)
■ — /—x mb , p
0 as before.
PROB. (10.) TO FIND THAT POINT (F) IN A GIVEN ELLIPSE

ABHD WHICH, OF ALL OTHERS, IS THE MOST REMOTE
FROM THE EXTREMITY B OF THE CONJUGATE AXIS.
(Pig. 15.)
Drawing FE parallel to the transverse axis AH, and
making AH = a, BD = b, and J32£ = <#, we have, by the
property of the curve BF2 = BE2 + EF2 = ^2 + ~ {bx - <r>)
= a?2 H—t A2 ^2- Now « is greater than b, -'• jo must

a2\ 0 b2 — a2
be greater than unity, and therefore (1 — -^\ x% =
J*/ 62
(/jf2 ___ 7)2v //7^ —■ Z>^\ / Cub

—aH *2 = (-p—) (rfTirji ^ - ^ = max- and
a2b
therefore -5 T6 x — x2 = max. which let = r, and we there-
a2 — b2
a2b
fore find <2?2 — -^ — 5? = — r. Solving this quadratic we
find x = 2 __ dfc A/ ( ~y—jz) — r. Now it is evident

( 25 )
that r = max. when (-|—^j

W — bLi = r. But from the nature
of the figure, the greatest value that x ( = BE) can possibly
admit of is b = jBD, therefore if the relation of a and 5 be
la2b
such that -~—75 is greater than b. this solution is mani-
cr — o2 °
festly impossible. To determine this limit, therefore, make
ia26
= b; then it will be found that 2 b2 = a2. Whence
the foregoing problem can only obtain when 2BD2 is equal
to, or less than AH2.

' ,, . a2b 9 . A a2b 4
In the expression —2 ~ <# — ar = max. let -^ „ = ^4
.*. ^ — x2 = max. Let # = ?/ + —-, therefore Ax — <^2

d>
A2 A2 A2
= Ay + — -y2 ~ Ay -—^ —~-y2=: max. when y = 0
.4 ^
.■.a?=:y + -2=^= i«2Z> as .before-
^TZT2 .
PROB. (11.) GIVEN THE BASE AND PERPENDICULAR OF A

TRIANGLE, TO DESCRIBE IT SO THAT THE VERTICAL ANGLE
MAY BE A MAXIMUM. (Fig. 16.)
Let AB = c, DC = p, and AD — x, .• . D.B = c — x, AD

-=r^
x DB c — x
= tan a — —, ^-^ — tan b = and .*. tan C = tan
p DC p
X C — X
—
tan a + tan b _ p + p cp
^ ' ~~ 1 — tan « tan 6 1 — x (c — x) ~~ p2 —- ex -f #2
( 26 )
2) -— CSC "4~ OC
= maximum .*. = min. which let == r .'. oo2
cp
Q
— coo — rpc — p2. Solving this quadratic we find oo = -
c2 4»2\
pc(r H J— )• This problem has three cases:
1st. Let c z. 2p and .*. —-—— must be a negative quantity,

and therefore in this case r cannot be taken so small as to be
less than this negative quantity, .\ when r = min. it must =
4^2 __ c2 c
—44>pc and .'.# = -.
2 2nd. Let c = 2p
r .\ c2 = 4p2 .'.
(& 4??2 c /
—-—~ = 0. and oo = - -f V per. Now when r, or the
4pc 2
co-tangent or the tangent of the complement of the vertical
c
angle C = min. it must = 03 or oo — -. In this case, since
2
the complement of C= 0, the angle itself must = 90 degrees.
3rd. Let c y 2p or c2 v 4<p2. In this case when r or the
co-tangent of the angle C = min. it must be a negative
quantity, equal to the positive quantity —-— > an^ •'• w
= -. In this third case it is evident that the vertical angle

C must be obtuse, because its co-tangent is negative. It is
also evident that in every case the triangle is isosceles.

c
In the expression oo2 — coo + p2 = min. let oo = y + - . * .
w2— ex + p2= y2+ cy + -c2 c2 + p2 = ?/2 — —•

cy — — c2 + p2
4* 2 4
= min. when y = 0, .'. a? = - as before.
( M )
PROS. (12.) TO FIND THE POINT D IN THE STRAIGHT LINE

CE, FROM WHICH AB SUBTENDS THE GREATEST ANGLE.
(Fig. 17.)
Let AC = a, CB = b> and CD — x.
It is evident that
AM _ BM
tan. ABB = tan. (iM - BDM) = MP MD
MA MB ~~
+ MD2
(^M-JBJIf) MD _ . , ._ L . A ,__ . fl
M& + AMBM' U 1S als° 6Vldent that = * Sm '
^filf = a — a? cos. 0 and BM =z b — x cos. 0, we, therefore,
„ , , (a — b) x sin 0
find tan. 6r — rsm^t/
_ . 9/1 + («
—^— x cos.
-—^-— ^ a maxi-
0) [b — x cos. 0)
mum . • . y = ,2?2sin20 + («
^-—;— oo cos.
=-7 0)Jr-~(b — x cos. 0) a mmi-
. .
(a — 0) x sm 0
mum ; and since (a — b) sin 0 is a constant given quantity
x2 sin2 0 + (a — x cos 0)— {b —■ x cos 0) must,11 also be a mmi-
x
mum which let = r, . • . x2 sin2 0 + ab — (a + b) cos 0 <# +
cos2 0- a?8 = <2?2 (sin2 0 + cos2 0) + ab — (a + *) cos 0- a? =
<2?2 -f ab — (a + b) cos 0*# = r#, therefore x2 —
j (a + 6) cos 0 + r > x = — ##. Solving this quadratic we find
(a + i) cos0 + r ^ A / f (a + b) cos 0 + ri 2
•= g ±v{ 1 j - ab-7
Now r cannot be taken so small (or, if necessary, negatively
so great) as to make \ - — [ less than ab,
because this supposition makes the value of x impossible . • .
•u • 4- -u J> + *) cos0 + r}2
when r = mm. we must have < - — > = a#,
O /T / , ja fl J (« + 6) COS 0 + r
.-. r = 2vab — (a + b) cos 0 and <# = '—
2\/«# /~r
= —~— = v #0.
( ^8 )

In the expression x2 •— \ {a -f b) cos 9 + r\ x = — ab
A
let the co-efficient of x = ^4 and let <s? = y -f -—, we there-
^2 ^2
fore find a?2 — Ax + ab = y2 + Ay + — — Ay — +
A2 A2 A2
ab = y2 — — + ab =: y2 + ab -- ~ == 0, or — = y2 -f
^2
«6. Now it is evident that r and — = min. when y == 0, . • .
A /— _
— = \/ ab ,\x =vab as before.
PROB. (13.) TO BISECT A TRIANGLE BY THE SHORTEST

LINE. (Fig. 18.)
Let ABC be the given triangle, and PQ the shortest line
required. Also let CP = x} CQ = y} PQ = u, and a, b, c
the three sides of the triangle, and Cthe angle BCA. Pm
and Bn are perpendiculars drawn from the points P and B
on the line CA. Now by similar triangles we find -j^p z=z
Bn = sin C, . • . Pm = x sin C and Bn = a sin C and . * .
77^
CQ x Pm xy sin C , CA x Bn ab sin C , , ,
—3 = ^— and —2— = -T- ; but by
CQ x ^m C4 x Bn xy sin C
supposition 2 x g = ^ . ' . 2 x —~ =
«6 sin C 7 rt «#
By Prop. 13, Book 2d of Euclid we find—

a2b2
u2 = ct?2 + if - 2#y cos ^=^+4-2- ab cos C = mm'
( 29 )
ich let := r, r.x* -f — «6 COS\C.x2 = rx*, and there-

4
a262
'e w* — {ab cos C -f r) a? = — Completing the
4 '
square and extracting the square root we find,
ab cos C + r t A /(ab cos C + r)2 — a2b2 ^T
w2 = = =±= A / ~ . Now a2b2
2 V 4
is greater than a2b2 cos2 C, .-.in order that the value of x2
may not become impossible, we must have ab cos C -f r =
ai, ,\ r = ab — ab cos (7, and . • . when r = min. we must
have^0 = ab cos C + r
_ ab
=_.-.^= A /ab
Y_andy_ a&
=- =
\f -K •'• ^ = "o" + "9 a* cos C = «6 (1 — COS C) =
<**. {(2ab + c2 - (a2 + 62))

2ab }= c2 - (a - 6)2
2 ^ and, •'• U =
(c — a + 6) (c + a — b)
a/
a2b2
Let ab cos C + r = ^4 and —— = JB .*. the equation <3?4 —
4
a262
(#6 cos C + r) #2 = — —r- becomes a?4 — Ax2 = — B.
Also let a?2 = y + —> .\ oc* — -<4#2 = ?/2 + Ay + -: Ay

<l 4
u42 ^42 A2, A2
— — 2^4= y2 = — B. .- . y2
y + B = — 4 .' . when —4 and
r = min. y = 0, .• . B = —- or -j- = — and ab — A — ab

4 4 4
cos C + r and r ~ ab — ab cos C ~ ab (1 — cos C) and
9 ^4
2" = #6
' cos2C + t = "a"'
ab ••• ^ = V
A /ab , -
"2 as before'
( 30
PROB. (14.)
X2
Let y = x tan 0 — - ^ : find <# that w may be a maxi-
XT 4p cos2 0 tan 0# a? — x2 , .
mum. JNow ^ = ~ 5-7T == max. and since
* 4p cos2 0
4p cos2 0 is a constant given quantity, we must have 4p cos2 0
tan 0 a? — a?2 = max. which let = r. Also let the coeffi
cient of x in this equation = 2A, and we therefore find
2Ax — x2 = max. = r or 2-<4# — a?2 = r and hence a?2
— 2Ax = — r. Solving this quadratic we find a? = ^4 +
V A2 — r, .\ when r = max. we must have A2 = r .' . x =
.4 = 2A
— = 4p— cos2 0 tan 0 = 2p cosz 0 tan 0 = 2p
o sin
• a0 cos /)0
^ a j ox/i-/i /i^P2 sin2 0 cos2 0

and we find *y = 2p
r tan 0 sin 0 cos 0 —4p cos2^-^0 =
2p sin2 0 — p sin2 0 ~ p sin2 0. The equation is that of the
path of a projectile, and the maximum value of y is the
greatest altitude above the horizontal plane.

In the expression 2Ax — x2 = max. let x = y + A> . • .
2â? - x2 = 2^% + 2-42 - y2 - 2.4*/ - ^42 = -<42 - */2
2^4
which is evidently = max. when y = 0.-.x = A=: ~7T =
2p sin 0 cos 0 as before.
PROB. (15.) DIVIDE A NUMBER a INTO TWO SUCH FACTORS

THAT THE SUM OF THEIR SQUARES SHALL BE A MINIMUM.
Let x ~ one of the factors. . • . - = the other factor, and

x
( 31 )
a2
tlieir squares = x2 \—g = min. = r .\ x% + «2 = ro?2, and
.-. a?4 — n*?2 = — a2. Solving this quadratic we find, x2 =
± A/- a2. It is now evident that when r = min. we

2 Vi
must have 7=a2or-: = «.,,,^=-=fl,,^=Vfl.
4 2 2
7%e same solved without impossible roots.

In the expression x* — r#2 = — a2 suppose #2 = y +
m iy£ y& Y"
- .\ x* ~ rx2 = y2 + ry + — - ry - j = y2 -- ~^ =
— a2 . • . y2 -f a2 = —, which is evidently a minimum when
y =0, .*.- = « and <2?2 = - = a> . • . <# = V a as before.

2 2
PROB. (16.) FIND THAT FRACTION WHICH EXCEEDS ITS SECOND

POWER BY THE GREATEST POSSIBLE NUMBER.
Let x be the fraction, and it is required to find such a

value for x which may make x1 — x2 a maximum. Let
x — x2 = r . • . x2 — x — — r, and solving this quadratic we
find x = \ -f \/\ — r. Now it is evident that r cannot be
greater than J, and therefore when r == max. it must be =
J and .*. x = i = the fraction required.

In the expression x —■ x2 = maximum, let # = y -f the
Or
coefficient of — = y + ^, and . • . we find x — x2 = y -\- \ —
y2 — y —- J = J — y2y which is evidently a maximum when

y = 0, .*. a? = | as before.
( 32 )
PROB. (17.) OF ALL TRIANGLES UPON THE SAME BASE, AND

HAVING THE SAME PERIMETER, EIND THAT WHICH HAS THE
GREATEST AREA.
Let 2P be the perimeter, a the given base, x and y the

remaining sides. It is demonstrated in the Introduction that
in any plane triangle whose sides are a, x and y and semi-
perimeter = P, the area = \/P (P — a) (P — x) (P —• y) ;
and because the square of a maximum is a maximum, we
must have P (P — a) (P — x) (P — y) = max. and
P (P — a) is a given constant quantity, we must also have
(P — x) (P — y) = max. Now y = 2P — a — x.\P — y
= P — 2P + « + #=# + <# — P, therefore by substitu
tion we find (P — x) (a -\- x — P) — max. and .* . aP — P2
+ (2P — «) a? — x2 = max. and as aP —- P2 is a constant
given quantity, we must also have (2P — a) x — x2 =
max. which let = r. We now have a?2 — (2P — a) x =
2P a
— r, and solving this quadratic we find x = h
r. It is evident that when r is a maximum,

a/
it must be = —- .*. x = -—-—- = P — - and y
= 2P — « — # = 2P — a — P + ~ ~ P — ^ and there-
fore y = x, or the triangle is isosceles.

In the expression (2P — a) x — x1 — max. let x =
E^ + y, ... (2p _ «) * _ J = (2P - «) y + i^^
- y» - (2P - a)^ y - (2P
v - o)4
^- = ^(2P r- a)2
'- — jr =
( 33 )
2P a
max. which happens when y = 0 .\ x = —- = P
- as before.
PROB. (18.) TO INSCRIBE THE GREATEST PARALLELOGRAM

WITHIN A GIVEN TRIANGLE ABC, THE ANGLE A BEING
ONE OF THE ANGLES OF THE PARALLELOGRAM. (Fig. 19.)
Let AEGFhe the greatest inscribed parallelogram required,
and ED the perpendicular let fall from one of its angles E,
upon one of its sides AF. Also let AB = c, AC = b and
AE = x.
The area of the parallelogram = AF x ED. The lines
EG and AC being parallel, the triangles ABC and EBG
must be similar, and consequently AB : EB \\ AC \ EG, or
c : AB — AE :: b : AF; or c : c — # : : b : ^F, .• . ^F =
/> (/» /yA
— and the perpendicular ED is evidently = EA x
sin A = a? sin ^4. Now substituting these values of AF and
i^rh a j *i.i. ni #sin.4x6(c — a?)
j&x>, we find area ot the parallelogram =
bsmA . „. _ . bsmA . ,
= c K(ex — ar)' = max. : and since c is a constant
given quantity, we must also have coo — x* = max. Let
ex — x% = max. = r .'. x* —• c# = — r, and therefore x =
.â/?-*
— . %/ , , , and hence it is evident that when r = max.
c2 c AB
it must be = —, and .•.#=- or ^1£ = ——.
4' 2 2

c
In the expression ex — x2 = max. let x = y + - and
( 34 )
<? — y» — c«/ — —
.• . ex — x2 = cy + — c2 = —
c2 — ?/2 which
*w 4 4
is evidently = max. when «/ = 0, .'.#=- as before.
PROB. (19.) OF ALL E QUI-ANGULAR, AND ISOPERIMETRICAL

PARALLELOGRAMS JFIND THAT WHICH HAS THE GREATEST
AREA. (Fig. 20.)

Let ACDE be the required parallelogram, AE = oo, AC
= y} and semi-perimeter = a. It is evident that the area of
this parallelogram = AC x EB (1.)
Now by supposition a? + y = «, ,\ y = a — x = ./4C
and I£B = .<42£ sin ^4 = x sin ^4 ; substituting these values
of AC and EB in equation (1) we find, area of the parallelo
gram = sin A (ax — x2) = max. Now as sin A is a con
stant given quantity, we must have also ax — x2 = max.
which let = r .*. x2 —- ax = — r. Solving this quadratic
a fa2
we find # = -=*= A/ — — r, and it is evident from this value
a2 a
of xt that when r = max. we must have r = — .\ a? = ~
' 4 2
and 2/ = a — ^ = «--- = ~ .\ x = y. Hence it appears
that of all equi-angular and isoperimetrical parallelograms,
the equi-lateral has the greatest area.
In the expression ax — x2 = max. let x = y -f « an^
therefore ax — x2 = ay + — — y2 — ay —r = i— 2/2 =
<s 4 4
max. when y = 0, ,\ # == - as before.

( 35 )
PROB. (20.) OF ALL TRIANGLES ON THE SAME BASE, AND

HAVING EQUAL VERTICAL ANGLES, TO FIND THAT WHICH
HAS THE GREATEST PERIMETER. (Fig. 21.)
Let ABC be the required triangle, of which the base AC

is given = b, and the vertical angle ABC = B : it is required
to find the mutual relation and magnitudes of the remaining
sides AB = x and BC = y when the perimeter or the sum
of all the sides is a' maximum. Let a perpendicular AD be
drawn to the line BC. It is evident, by the first principles
of trigonometry, that BD = AB cos B = x cos B} AD =
AB sin B = x sin B and .-. DC = V'AC* - AD2 =
\/b2 - a?2sin2.B, and .\ y = BD + DC = x cos 5 +
\/ Z>2 — #2 sin2 .B . • . perimeter = 6 + # + # cos B +
a/ Z>2 - x2 sin2 5 = 6 + (1 + cos B) x + V b2 - w2 sin2 2*
= max. and as S is a constant given quantity, we must also
have (1 + cos B) x + \/b2 — a?2 sin2 i? = max. which let = r
.'. \/b2 — <2?2sin2i? = r — (1 -f cosl?) x, and, squaring both
sides, we find b2 — x2 sin2 JB = r2 — 2r (1 + cos B) x +
(1 + cos 5)2 a?2, therefore {sin2 B + (1 + cos JB)2} a?2 -
^ /, ™ 70 o o 2 (1 + cos B) r
2 v 1 + cos B)' rx = b2 - r2 r.x2 - sm2
• 2 .B
p +, (1
n +, cos jB)2
PN2 x
b2 — r2
Solving this quadratic we find x =
sin2J3 + (1 + cos£) 2*
(1 + cos B) r /y/{sin2^ + (1 + cos£)2}62---sin2£r2
sin2J5+ (1 + COS.B)2 {sin^JB + (1 + cos£)2}2
Now it is evident that r or sin2.B r2 when a maximum, must be
/■•*», n , m*\ v> bV$ui2B + (1 + cosff)2
= <jsin2I? + (1 + cos B)2\ b2 or r = r-^ -
and therefore we find x = , ■

sinJBvWs 4- (1 + cos£)2
( 36 )
and y = ce cos B + \/b2—x2 sin2J3 = — , —

sm.£\/shi*B + (1 + cosB)2
bsinB __ b cos 5 + b cos2 J5 4- b sin2 5
r \/sin2J3 4- (1 4- cos£)2 ~~ sinJBvW-B + (1 + cos£)2
= 6(1 + cos B) . • . % = y. tt
Hence or* all
ii
sin B A/sin2 B 4- (1 4- cos£)2
triangles on the same base, having equal vertical angles, the
isosceles has the greatest perimeter.
In the equation oo% — . \—Tr—■ —^ a? =

^ sin2 B 4- (1 4- cos By
b2 r2
let
sin2 B 4- (1 + cos J5)2 sin2 B 4 (1 + cos £)2'
(1 4- cos-B) V _ ,
sin2 £ + (1 + cos Bf ~ m> sin2 B 4- (1 4- cosiS)2 " n
■ 2p . n , ™s = Q •'• ^2 - 3mr a? = » — §r2. Also

snr,B + (1 4- cosi?)2 *
let # = 2/ 4- mr and we therefore find y2 4- 2mry + m2r2 —
2mry — 2m2r2 =z y2 — m2r2 = n — gr2 .\ r2 = g
which is evidently = max. when y = 0, .'. r =

vg — m',2
and, .'. ^ = m\/- n = , b (1
v 4- cosJB)7 ,
=• by
Vq - m2 sin £ Vsin2 B + (1 4- cosJS)2
substitution as before.
PROB. (21.) TO INSCRIBE THE GREATEST RECTANGLE IN A

GIVEN SEMICIRCLE. (Fig. 22.)
Let CN — Xj and CA = a .-. JVP = vV — #2 and there

fore the rectangle required = 2PM x CM = 2x \/a2 — x2 =
max. .*. 4a2#2 — 4&* = max. and .\ o?x2 —• x* = max. which
( 37 )
a2 d= A//a*
let = r, .\ x4* — A2 = — r^ .• . <#2 = — —- — r2. It
is evident that when r = max. it is = a4 a2 . • . x =

— . # . x2 = —•
4 2
7%£ same solved without impossible roots.

a2
In the expression a2x2 — x4* = max. let x2 = y + — .\«V2
— x* = a2y + - y2 — a2y — -j- = - y2 = max. when
w2 = 0, . • . #2 = —- and x = —7= as before.

2 a/ 2
PROB. (22.) OF ALL SQUARES INSCRIBED IN A GIVEN SQUARE

TO FIND THAT WHICH IS THE LEAST. (Pig. 23.)
Let ABCD be the given square, and abed the required one.
Also let AB = BC = a, aB — x, .*. Aa = a — a?. Now it
is evident that «# = ac, the Z. ^4 = z. J5 and the angles â#
and ^4#a are together equal to 90 degrees = angles Aab
and Bac .\ Z. ^46# = Z. Z?#c .'. the third angle Aab = Z.
i?c# .*. Aa = 5c; but^4a = a — <# .*. 5c = a — a?. Now
it is evident that «52 -f Be2 = ac2 or x2 + (a — #)2 = ac2
= the area of the square required = a maximum, which let
= ry .*. 2x2 — 2ax + a2 = r, and by proceeding exactly as
in problem (5) we find x = - when r = max.
The same may be solved without impossible roots as in
problem (5.)
( 38 )

GIVEN ELLIPSE. (Pig. 24.)
Let AFGBED be the given Ellipse, and FDEG the in

scribed rectangle required. Also let mC (where C is the
centre) = Cn=: oo} AC = a &n&pC = b .*. mn = 2x. Now
by the property of the Ellipse demonstrated in the Introduc
tion we find mF = - \/a2 — x2 .\ 2mF = — \/a2 — x2
a a
and therefore the rectangle FE = FD x BE = FD x mn
= — a/«2 — a?2 x 2# = — \/a2x2 —• a?4 = max.. and as —
a a a
is a constant given quantity, we find \/a2x2 — oc* = max.
and also a2#2 — #4 = max. which let = r, .• . oc* — a2a?2 = —
«2 /a4
r. Solving this quadratic we find x2 = — d= A/ — — r, and
a*
hence it is manifest that r cannot be greater than — and
a4 a2
therefore when r = max. it must be = — and . ' . x2 = — and
4 2
a
x = -7=.
\/2
This problem may be solved without impossible roots,
exactly in the same way as problem (21.)
PROB. (24.) GIVEN THE BASE AND THE VERTICAL ANGLE OF

A TRIANGLE, SHOW THAT WHEN IT IS ISOSCELES ITS AREA
is a maximum. (See Fig. 21.)
Let ABC be the required triangle of which the base AC
is given = b, and the vertical angle AB C = B : it is required
to find the mutual relation of the remaining sides AB = x
( 39 )
and BC = y when the area of the triangle is a maximum.
Let a perpendicular AD be drawn to the line BC. It is
evident, by the first principles of Trigonometry, that BD =
AB cos B = x cos B, AD = AB sin B = a? sin i? and .\
DC = \/^C2 - AD2 = \/Z>2 - a? sin2 .# and r.y = BD +
DC— x cosB + VV — x2 sin2i?, therefore the area of the tri
angle AB C = AD x 5 C = 0? sin^ {a? cos^ + ^-^sm2^}
= sini? (#2 cosjB + a? a/62 — x2 sin2 5) = max. Now as
sin B is a constant given quantity, we must have x2 cos B +
0? \/ft2 — oc2 sin2i? = max. which let = r, . • . x Vb2 — x2 sin2I?
= r — x2 cos B or fi2a?2 — x4* sin2 5 = r2 —- 2r cos B x2 +
x* cos2^ or a?4 (cos2 B + sin2 5) - (62 + 2r cos.#) a?2 = —
r2 or x41 — (62 + 2r cos B) x2 = — r2. Solving this qua-
j x- we nnd
dratic £J xL« = b2+2rcosB_1_
-i-. aA // (b2
v<
v + 2r cosB)2
4
b2 + 2rcosB
2 *v
~ V
/b* - 4r (r sin2 B - b2 cos J9)
4
Now it is evident that r cannot be taken so great as to
make 4r (r sin2i? — b2 co$B) greater than 64, and therefore
when r = max. we must have 64 = 4r (r sin2 B ■— b2 cos i?)
and from this equation

u we find r2 . «B
sm2 ^ f = -;—.
4 snr9 i?
^
and, solving this quadratic, we find 2r = —-—r-y^ •

^2 i 2^ co$B
Substituting this value of 2r in the equation #2= ^
- , 2 62 (1 + cos^) , A /ft8 (1 + cosff)

we find #2 = — 0 . a-^—- and x = A/ — 0 . 2 p—'
2 sm2 i? V 2 sir i?
and y = BD + DC = a? cos B + V b2 - x2 sin2 J? =
» a /PQ- + MBB) , A AI 62(1 + cosJ9)sin2I?
wJV 2sin2i? + V*" 2sin2i? =
( 40 )
r> A A2 (1 + cos£) , . /ft2(l-cos#) . , .,

C0Sj9 V 2sin2.ff + V ~2 ' squaring both
sides of this equation we find yl = . a .

^ sin4 J? 2
_ A /ft* sin2 i? , ft2 (1 - cos 5) M p
+ 2 cos 5 A/ , ■ 2 p + — o ^ * cos 5
V 4 sm2 B 2
(cos2 5 (1 + cos^) + sin2^ (1 - cos 5)}
+ ft2 i - ^^ U = ft2 cosi?
2 sm2 i?
ft2 (1 + cos3^
+ —i 0 . -2 psin2^ cos.#)^ = ft2M cos 5
_ +,
2 sm2 B
{ 1 + cos ^ (1 - sin2 B) - cos .# sin2 B }
2 sin2 B
„ f _ 1 + cos^ — 2sin2J9cos.#)
= ft2 -J cos B + 0 . 2 p >
( 2 sm2 i? J
= — ^2 sin2
. o5^—- = oc2.' • " • V*
y = #2j' and .*. *y = x. Hence it
appears that the triangle must be isosceles, in order that its
area may be a maximum.
PROB. (25.) TO FIND THE LEAST TRIANGLE TCt, WHICH CAN

BE DESCRIBED ABOUT A GIVEN QUADRANT. (Fig. 25.)
Let CA = a, tC = x, and CT — y. It is evident that

the line or hypothenuse Tt is a tangent to the quadrant at
the point P, and therefore the angles tPC and CPâre right
angles. By similar triangles, according to Prop. 8, Book 6
of Euclid, we have tC : CP :: CP : CN or x : a ;: a :
CN .-. CN = - and NP =: CM = V CB2 - CiV2 =
#
aA2
V
-^
XL
= X
- V> - a2. Also CT : (7P :: CP : Clf
or y \ a \\ a \ - vx2 — a2, . \ y = . ——. and therefore

00 V x1 — a2
( 41 )
the area of the triangle TCt = \xy = \a x

v x2 — ft2
minimum. Now \a is a constant given quantity, .-. . -zz~.
V a?-— a2
/y>4
or its square
u —5
<3?2 — ft2 z = min. which let = r, . * . x* = 7\a?2 —
rft2 or x* — to2 = — ra2 (1)
iy
r AT*2 4/*ft2
Solving this quadratic we find x = - =fc A/ —
2
y* / v (t 4ft2)
— ± A / —-— - and here it is evident that r cannot be
2 V 4
less than 4ft2, . • . it must be = 4ft2 when it is a minimum, . • .
y» 4ft4
4ft2 ^ 0 -, /
/-7T . ft<2?
x' = - = —- = 2ft2 and x — a v 2 and y — . .—
& & vx2 — ft2
= ft \/2, .-. x = y. Hence it appears that the angle
ft
PTC must be = 45 degrees, or that the triangle described
must be isosceles when it is the least possible.

In the equation (1) viz. in x* — rx2 = — rft2 let x2 = y
+ - and therefore x* — rx2 = ^/2 + ry + ry ■— — =
n*2
2/2 — — = — rft2, . • . r2 — 4rft2 = 4?/2, and therefore we find
4
r = 2ft2 4- \Z^y2 + 4ft4, and here it is evident that when r =
min. we must have 4?/2 or y = 0, .\ r = 2ft2 + 2ft2 = 4ft2
and x2 = - = — = 2ft2, .•.# = # V^3 as before.
( 42 )
PROB. (26.) SUPPOSING A SHIP TO SAIL FROM A GIVEN PLACE

A, IN A GIVEN DIRECTION AQ, AT THE SAME TIME THAT
A BOAT FROM ANOTHER GIVEN PLACE B, SETS OUT IN
ORDER (IF POSSIBLE) TO COME UP WITH HER, AND SUP
POSING THE RATE AT WHICH EACH VESSEL PROGRESSES TO
BE GIVEN, IT IS REQUIRED TO FIND IN WHAT DIRECTION
THE LATTER MUST PROCEED, SO THAT IF IT CANNOT COME
UP WITH THE FORMER, IT MAY HOWEVER APPROACH IT AS
NEAR AS POSSIBLE. (Fig. 26.)
Let the celerity of the ship be to that of the boat in the

given ratio of m to n ; also let D and F be the places of the
two vessels when nearest possible to each other, and, from
the centre B, through F, suppose the circumference of a
circle to be described. Then the distance DF, being the
least possible, the point F must be in the right line DB,
joining the point D and the centre B ; because no other point
in the whole periphery, at which the boat from B might arrive
in the same time, is so near to D as that wherein the line DB
intersects the said periphery. Eut now, to get an expression
for DF, in algebraic terms, let jBCbe perpendicular to AQ
and make AC = a, BC = b, CD = w, and then BD will be
= \/ BC2 + CD2 = w b2 + x2 ; moreover, because m \ n
t: AD or a + x \ BF. we will have BF= , and conse-
m
quently DF = V b2 + x2 = V b2 + x2
= min. which let = q. .\ v b2 + x2 — — = q +

m m
— which let = r. Now it is evident that since — is a con-
m m
stant given quantity, and q = min. we must also have q +
— or r = min. .\ Vb2 + x2 = min. = r or V b2 + x2
m m
( 43 )
— r + — and therefore 6Z + # = rL -\ x H r . .
m mm*
m2—n2 0 2w\a? „ 70 „ 2rmn (r2—b2)m2 /nN
^ =r2--62or^2~ ^>= '— (1)
Solving this quadratic we find,
mnr / (r2 —• £2) m2 (m2 — ra2) + m2n2r2
x = m2 — n2 V {m2 — %2)
mnr /'{m2(m2 - rc2) + wV } r2 - 62m2 (m2 - rc2)
m2 ■— n2 ^ (m2 — n2)2
mnr , A /m¥ — #2m2 (m2 — ^2) TT ., ,,
= — g =fc A/ —^—4- ^. Here it must be
remarked, that this problem becomes impossible when m is
less than n, for in this case the quantity — b2(m2 — n2)m2
must become a positive quantity, and therefore there remains
no condition of r becoming a minimum. Now it is evident
that m¥ or r cannot be taken so small as to make the root
impossible, therefore when r = min. we must have m4r2 =
b\/m2 — n2 n mnr
b2m2(m2
K — n2) and ,\ r = ™ and x = —mL«—— w9
m
nb , D.F
; also _. _. = Vb2
/75 -\- x2r â +■ nx = r na
^/m* _ ^' mm
b \/m2 — n2 na b\/m2 — n2 — na , „ .J . ■
m m = m : whence the position
x
of F is known. From the above it is observable that, as DF
must be a real positive quantity (by the question), this
method of solution can only be of use when m is greater
than n, and b\/m2—n2, also greater than na : for in all other
cases the boat will be able to come up with the ship.

2mnr (r —~ b < vyl
In the equation
u (1)
w or x2 — m*
—s —■ nl-9 x = -—5
mL — n*v— let
half the co-efficient of x = A, and the second member of
( 44 )
the equation = B, . • . x2 — 2Ax = B. Now let x = A ■+■ y
.♦. x2 - 2Ax = «/2 + 2^/ + ^2 - 2^ - 2^£2 = y2 - ^42
= 5, .-. y2 = 5 + ^ and by substitution, 2/2 = B + ^42 =
(r2 — b2) m2 m¥r2 m2n2r2 -f (r2 — #2) m2 (m2 — %2)
m2 •— n2 (m2 — ^2)2 "~ (m2 — n2)2
m4r2 __ £2m2 /m2 __ ^2\
= —- ~—- £ .'. mr — oV vmr — ?r) =
(m2, — n2)2
[vyp1 — 7) \ If -\- u 777/ \7Yv —~ Yl J
(m2 — n2) 2y2, and thereforewefind r2 = — -A
which is evidently a minimum when y2 or y = 0, .'. r =
b\/m2 — n2 ^ nb , P
and a? = ■ , —= as before.
m vm2 — %2
PROB. (27.) TO FIND SUCH A VALUE TOR X AS WILL MAKE

b — (a? — a)2 a maximum.
Let #2 — (x — «)2 = y, .' . b — x2 + 2ax — a2 = r, and .• .
a?2 — 2ax = b — a2 — r. Solving this quadratic we find
x = a =fc \/& — r, and here it is evident that r cannot be
greater than b; therefore when r = max. we must have
r = b, .*. x ~ a.

In the equation x2 — 2ax = b — a2 — r, let a? = «/ + a
.'. x2 — 2ax = y2 + 2a?/ + a2 — 2^ — 2a2 = y2 — a2 =
b — a2 — r, .\ r = 6 — y2 which is evidently a maximum
when y2 or y = 0, . \ r = b and a; = a as before.
( 45 )
PROB. (28.) TO FIND SUCH A VALUE FOR X AS WILL

x
MAKE - 5 A MAXIMUM.
1 + X2
Since - 5 = max. .* . = min. which let = r. there-

\-\-x2 x
fore x2 — rx = — 1. Solving this quadratic we find x =
f+VS-*.
2rti'%/.
V 4 -1J and here it is manifest
^ that r or —
4 cannot
be taken so small as to be less than 1, therefore when r =
r2 = 1, . • . r = 2 and a? = -r = 2
mm. we must have -j- - = 1.
4 <s <4

T
In the equation x2 — rx = — 1, let # = 2/ + - and there
in
m& /y*£ (y*a
fore a;2 — rx = ?/2 + ry + — — n/ — -~ = 2/2 — -r = ■— 1*
o*2
. • . -— = ^2 + 1, which is evidently a minimum when y = 0,
.\ — = 1, ,\ r = 2 and # = — = — = las before.

4 ' 2 2
PROB. (29.) TO DETERMINE POR WHAT VALUE OP X THE

EXPRESSION a4 + b3X — C2#2 BECOMES A MAXIMUM.
/ a4 b3 \
Here' a4 + b3x — c2x2 = c2 1—$ H ^ x — x2) = max.
or —o- H—r x — ar = max. Now since —z is a constant

& & c2,
b3
given quantity, we must have — x — x2 also = max. which
b3 x — a;2 = r, or #2 — —g
let = r, . • . — b3 a; = — r. Solving
( 46 )
b3 /Ifi
this quadratic we find x = —2 + \f t~i — r> an<i here it is
b6
evident that r cannot be greater than -r—. and therefore when
r is a maximum we must have r = —: and or = 77-^.

4c4 2c2

bs b3
In the expression x ^ x = — r let # = y + ^-2 and
b3 _ 2 b3 b5 b3 b6
therefore *?_ —
—1 xar=y* + _y+___j,__=:
- y +^y + 4^--^y-2^
b6 = — r .•. 7* = —
y2 ■— 7-4 b6 — ?/2 which is evidently a max.
b3
when y = 0 ,\ x = -^—^ as before.
^ 2c2
PROB. (30.) TO DETERMINE SUCH A VALUE FOR X AS MAY

MAKE THE EXPRESSION a + Va3 — 2a2# -f ax2 A MINI
MUM.
Here it is evident that a is a constant given quantity, and
consequently Va3 — 2a2x + ax2 or its cube a3 — 2a2x + ax2
must also be a minimum. Again as a is also a constant given
quantity we must have = a2 — 2## + #2 =
min. which let = r .'. x2 — 2<2# = r — a2. Solving this
quadratic we find # = a + VV, and here it is evident that
when r = min. it must be = 0, .\ x = a. This problem
may be solved without leaving out any constant given quan
tity in the following manner, which is more elegant—
Let a + Va3 — 2a2x + ax2 = r, .-. a3 - 2a2# + ax2 =
1 /* ~— C5 ) 3 "~~~ CI3
(r — a)3 .-. x2 — 2&r = - - —. Solving this quad
( 47 )
ratic we find a? x = a -f A/-

\f -. Here it is evident
V a
that r cannot be taken less than a, because this supposition
makes the root impossible : therefore when r = min. it must
be = a, . • . x = a as before.

(r —~~ aV^ -—~ cfi
In the equation x2 — 2a# = let x = y -\- a
.'. x2 - 2a# = y2 + 2ay + a2 - 2ay - 2a2 = ?/2 - a2 =
.'. (r — a)3 = a?/2, .'. r = a%t + a,, which is
evidently a minimum when y = 0; . \ r = a and # = a as

before.
PROB. (31.) TO FIND THAT NUMBER X WHICH, BEING MUL

TIPLIED BY THE SQUARE OF ANY GIVEN NUMBER a, AND
THE PRODUCT DIVIDED BY THE SQUARE OF THE DIFFER
ENCE OF a AND X, THE QUOTIENT IS THE GREATEST
POSSIBLE.
The product of the square of a and the required number

x = a2x, and the square of the difference of a and x =
(a — x)2. Therefore the quotient which is to become a
ax
maximum is -.I & —■ XJrg. Since the reciprocal of a maximum
must be a minimum, we must have -—5—— aLx = min. which

let = r, . • . {a — x) 2 = aV# or a2 — 2ax + #2 = a2rx, . • . x2 —
(2a -f- «2r) # = — a2 or x2 —- a (2 + ar) % = — a2. Solving
this quadratic we find,
_ a (2 + ar) /4a2 + 4a3r + a*r2 — 4a2
a (2 + ar) /a3(4<r + ar2)

o + A/ 4
( 48 )
Here it is evident that when r is a minimum it must be ==
0, . • . x = 2a
— = a. In this problem impossible roots are not
required at all.
The same solved without impossible roots in another way.

In the equation x2 — a (2 + ar) x = — a2 let x = y +
a^ 2 — .'. #2 — a (2 -h ar) x = y2 -f a (2 -j- ar) y +
a2(2 +
^ ar)2 a (2
/ft + ar). y a2(2
K—+ ar)2
L = y*9 a2(2 +
- ar)2
L
= — a3, .-. we find a2(2 + ar)2 = 4?/2 + 4a2, .\ r =

2vV 4- a2 — 2a ... . . .
2 which is evidently a minimum when y = 0,
a
.-. r = —-
az = 0. and x = a as before.
PROB. (32.) TO DETERMINE THOSE CONJUGATE DIAMETERS OF

AN ELLIPSE WHICH INCLUDE THE GREATEST ANGLE.
Call the principal semi-diameters of the Ellipse a, b, the

sought semi-conjugates x and x', and the sine of the angle
they include = y. Then by conic sections we find
x2 + x'2 = a2 + b2 .-. x' = va2 + 52 — x2 and ##'?/ = a#
.\ y = —- and therefore v = — . z= = min. Here
^ aVa2 + 62-tf2
it should be remarked that when we desire to find the
greatest value of an angle we may proceed to find the least
value of its sine, for the angle is greater and greater as it is
more obtuse, and the sine of an angle is the less the greater
is its obtuseness. It is for this reason that we have put y, or
the value of the sine of the greatest angle, equal to minimum.
( 49 )
Now, omitting the constant given quantity ab> and inverting
and squaring the function, we find {a2 + b2) x2 — x* = max.
which let = r, and therefore x4* — {a2 + b2) x2 = — r. Solv-
a2 + b2 /(a2 + b2)2
ing this quadratic we find x2 = -—-— ± A/ -— r
and here it is evident that r cannot be taken greater than
fft2 _j_ J2\2
7—— and therefore when r = max, it must be =
4
(a2 + b2)2 A fa2 + b2
-—2"— • ' • °° = A/ —9 —•
TI11 ^
tne
w. -,,.
solution of this pro
blem that property of the Ellipse is made use of which has
not been demonstrated in the Introduction.
The same solved ivithout impossible roots.

In the equation x* — (a2 + b2) x2 = — r, let x2 = y +
a2 + ', and .-. x* - {a2 + 62) #» = y2 4- (a2 + b2) y +
2
\ —±- _ (a2 + 62) y - * ^—^ = y2 -g
= — r, and therefore r = -—-— #2> which is evidently
, a2 + 62
a maximum when ^ = 0, . • . x^ = —-— or # = /y/ —-— A/t±E
as before.
Now as ?/ = — — we must have by substitution
xVa2 + 62 - a?8
2/ =
y^V-^-2^ V^V-
a# 2a#
2~
a2 + 62 a2 + 62'
( 50 )
PROB. (33.) GIVEN THE EQUATION y2 — 2mxy + X2 = «2

TO DETERMINE SUCH A VALUE OF X AS WILL MAKE 7/ A
MAXIMUM.
From the given equation, in which m is less than unity, we

find x2 — 2myx = a2 — y2, and solving this quadratic we
find x = my ± \/a2 + (m2 — 1) 2/2 = my =i= \/a2 — (1 — m2) y2.
Here it is evident that y cannot be taken so great as to make
(1 •— m2) y2 greater than a2, and therefore when y = max.
we must have (1 — m2) y2 = a2, . • . y = , —= and x =
V 1 — m2
ma
Vl - m2'
In the equation x2 — 2myx = a2 — y2 let a? = # + my,
.*. <2?2 — 2my<^ = z2 + 2my^ -f m2y2 — 2my^r — 2m2?/2 =
z2 — m2^2 = a2 — y2, .*. (1 — m2) y2 = a2 — z2, .'. y2 =
a2 — ~2 ^2 which is evidently a maximum when z = 0, .'. y =
a _ ma , „
and # = , —r as before.
\/l — m2 \/l
PROB. (34.) IN A GIVEN CIRCLE TO INSCRIBE THE GREATEST

RECTANGLE POSSIBLE. (Fig. 27.)
Let AC be the rectangle, and EF a diameter bisecting

BC, OG~x and radius = a, then (Euc. III. and II.) EH —
OF; also, BO = \/a2 - a?2 .\ -BC = 2.BO = 2 vV - ^2
and HO = 20G = 2# .•. rectangle ^4C = 2a? x 2v/^_ #a
or 4a? Va2 — a?2 = max. and therefore the square of the
fourth part of this rectangle, viz. a2x2 — a?4, = max. which
let = r, . • . x41 — a¥ = — r.
( 51 )
a? /a4
Solving this quadratic, we find x* = — + A/ — — r, and
a4
here it is evident that r cannot be greater than -j and there-
a4 a2,
fore when r is a maximum it must be = -r .*. a?2 = -~
4 2
and a? = ~t=.
a/2
7%e s#me solved without impossible roots.
In the expression a2<#2 — a?4 = max. let x1 — y -f
- .-. <^2 - ^ = «V + 2 ~ 2/2 ~ «V - 4 = J - V

which is evidently a maximum when y = 0, and therefore a?2
«2
= —-.•# = —7=a as before.
2 \/2
PROB. (35.) THROUGH A GIVEN POINT, WITHIN A GIVEN

ANGLE, TO DRAW A STRAIGHT LINE, WHICH SHALL CUT
OFE FROM THE ANGULAR SPACE THE SMALLEST TRIANGLE
possible. (Fig. 28.)
Let P be the given point, A the given angle, and CB the
line required. Draw PF and CE perpendicular to AB3 and
PD parallel to AC: then, since the angled and the posi
tion of P are given, AD, DP and PF are also given.
Let AD = a, DP = b, PF = c, and AB = a? :
then JSD : DP :: Jki : AC ) /T? . _ _rT .
and dp:PF::AC:Ce\ûc'^YL)
From proportion first we find BD I BA :: DP \ AC, or
AC — t~- and from the second proportion AC =
CE x DP BA x DP CE x DP BA CE
PF ' * BD ~ PF ' °r £Z> ~ PF and
( 52 )
hence BA ; BB :: CE : P.F, or #D : PF :: BA ; CE, oxx-
a ; c :: x ; CE = and therefore ABC = -
x— a 2
CX p
= r-r r = min. and — is a constant given quantity,
<w [X — tt) &
therefore X2 is also a minimum, which let == r, . • . x2 =

a? — a
rx — ra, and .\ x2 — rx = — ra. Solving this quadratic we
find x = y =fc /\/~ - ra = ^-±: A^L^^-J±} and
here it is evident that r cannot be less than 4a, and conse
quently when r is a minimum we must have r = 4a . * . x =
r 4a
2"= !T = 2a-
Hence, if AB be taken equal to twice AD, the straight
line passing through B and P will cut off the smallest tri
angle possible.

T
In the equation x2 —• rx = — ra let x = y + -~-, . \ #2 —
ft*2 ™2 »»2 ™2
ra? = y2 + ry + — - ry - y = y2 - — = - ra .'. ~
— ra = y2 or r2 — 4ra = 4«/2 or r = 2a ± a/4^2 + 4a2

which is evidently a minimum when y = 0, . * . r = 2a + 2a
r 4a
= 4a as before, and x = — = — = 2a.
( 53 )
PROB. (36.) THE RIGHT-ANGLE B OF THE RIGHT-ANGLED

TRIANGLE ABC, RESTS UPON THE STRAIGHT LINE DE,
TURNING IN ONE PLANE UPON B AS A CENTRE; RE
QUIRED THE POSITION OF THE TRIANGLE, WHEN THE SUM
OF THE PERPENDICULARS AD AND CE IS A MAXIMUM.
(Fig. 29.)
Let AB = a,BC = b3 and AID = x; then DB = VV-a?2;
also AB : BD :: BC : CE, or a : vV~=~^~2 :: b : CE = -
a
V^TT^ and AD + C# = oc + — Va2 - ^2 = max.
b va2 — #2 = r — a? and a2£2 __, £2^2 =
which let = r3 . • . —
« a2
b2(a2
— -
5 x2) = rla — 2r#
ft + ar or a2 +3—62 xL9 — 2r#
0 = ¥ —
a2 #2
2 2 2aV _ (y „ t2)a2
T °r * - ^2-^2 * - a2 + j* •
Solving this quadratic we find,

aV ^ / (b2 - r2) a2 (a2 + b2) T~a^r2
x =
a? + . # - V (a8 + 62)2
- a*r _,_ / a2b2 (a2 + 62) - flW
~ a2 + y - V (a2 + 62)2
aV _,_ A / a262 { {a2 + 62) - r2 } .
= zt A/ — -. Now it is
evident that r2 cannot be greater than a2 + 62 and conse

quently when r2 = max. it must be equal to {a2 + b2) .\ r
- ^^T^ and . - «2 a/^T^ = ^ =
a* + b2 ^/a2 + 62 ^
a third proportional to ^4C and ^4i?; which determines the
position of the triangle. To find the sum of the perpendi
culars, substitute the value of x; then, CE + AD =
b A / 0 «* , a2 b ab
a V a2 + 62 VV + §2 a v/a2 + £2 ^2 +
( 54 )
a2 _}_ £2 .
= . = v a2 -j- b2 = AC, .*. the sum of the perpendi-
V a2 + 62
culars, when a maximum = the hypothenuse of the original
triangle.

T ,, ,. 2a2r {b2 - r2)a2
In the equation
^ x2 a2- + b2
Tn w — ~—- ~- let w = y -f-
a2 -\- bJ
a2r „ 2a2r n 2a2r a4r2
a2 + 62 •'• ^ ~ a2
„9 +
, 62
i* * = 2/2
y +' -o—to
a2 + b2*2f +] (a2 + 62)2
2a2r 2aV2_ _ 2 ___ fl4r2 (62 — r2)a2
a< + b2V ~~ J^TVb*)2 ~ y ~ (a2 + 4252 - -^-qr^ and
therefore v(g8 + ^ = ^ . • . r2 = a2 + 62 - ^2
which is evidently a maximum when y = 0, .' . r2 ■= a2 -\- b2,

. a2
,\ r = vfl2 + #2 and a? = , as before.
vV + b2
PROB. (37.) TO FIND THE POSITION OF THE SAME TRIANGLE

ABC (see last Fig.) when the sum of the surfaces of
THE TWO TRIANGLES ADB AND CBE IS A MAXIMUM.
It has already been shown that, if AB = a, BC = b3 and

DA = x9 then Z>£ = Va2 - w2 and CE = - (a2 - a?2)*;
Now 2L4 : ^D : : CB : 5^, by similar triangles, ox,a : w.'.
b:BE= —w;
... ^D5 + BBC = ^|-^ + *=^*?
= « W — #2 + ;r- <# x - vV — #2
2 2a a
= U + -^-g) x\/a2 — #2 = max.
( 55 )
b2
and as ^ + ^2 *s a consân^ given quantity, we must also
have x va2 — x2 = max. or its square a2x2 ■— #4 = max.

which let = r, .'. x* — a2x2 = — r. Solving this quadratic
a2 /a?
we find, x2 = — d= A/ — — r> an^ hence it is evident that
a2
r cannot be greater than — and therefore when r is a maxi-
a cl2 a
mum it must be = ~ . ' . w2 = — and x = , = AD. But.
4 2 ^2
#D = A/a2 - #2 = A/ a2 - J = -^ = ^Z>, .'.the angle

^4jBD is half a right-angle.

a2
In the equation a2x2 — x* = r let a?2 = ?/ -f -— and therefore
a2x2 — x4* = a2y -\- ~ —y2 — a2y —•—=■— — y2 which is
Co
evidently a maximum, when y = 0, . • . x2 = •—- and x = —?=.CL
* v2
as before.
PROB. (38.) A STRING ABE OF A GIVEN LENGTH, IS FIXED

AT A, ONE EXTREMITY OF THE DIAMETER OF A CIRCLE,
AND WOUND ROUND PART OF THE ARC AB. THE REMAINDER
OF THE LINE, BEING STRETCHED OUT INTO A STRAIGHT
LINE AND TERMINATING IN THE DIAMETER PRODUCED; TO
FIND THE RADIUS OF THE SEMICIRCLE SO THAT THE AREA
BDE, INTERCEPTED BETWEEN THE PRODUCED PART OF
THE DIAMETER, THE ARC BD, AND THE STRING, MAY BE
A MAXIMUM. (Pig. 30.)
Let I = the length of the string ABE,

x = variable radius BC;
( 56 )
Then from the well-known properties of the circle
a i. Ann mcAB X BC , . . px2
Sector ACB = - ; and semicircle = '—,
where p = circumference of a circle whose diameter is unity.
We therefore find BDE = Sector ACB + triangle (7^^ -
semicircle.
__ ^4i? x x BE x oo poo2
The condition that "" 2 2 2~
the string is intended (AB ^ ™ ^ __ 2
to continue on a tan- = 1 _ 1 ±_ = 1 (^ — ##2)
gent to the circle is 2
omitted.— Ed. ■, / ^ M 2\ -, •
= |j9 ( — at — or) = max. and since
\p is a constant given quantity, we must have also — oo — oo2
— max. which let = r, .\ — x — oo2 — r, and x2 x=

3 p p
I /T2
— r. Solving this quadratic we find x = — dz A/ —2 — r,
2p ' V 4p2
P_
and it is manifest that r cannot be greater than —2, and
4p2
consequently when r = max. we must have r = —,-2 . • . x =
4p2
— or radius = — and therefore I = 2p x radius = p x dia-
2p 2p
meter = circumference; and hence it appears that the radius
is such that, if the circle were completed, its circumference
would be equal to the length of the string.
In the expression — x — x2 = max. let x = y + — and
therefore — x — x2 = — x + —- — v2 x — -r-0 =
p p 2p2 p 4/r
p
j~2 — y2 which is evidently a maximum, when y = 0 and
therefore x = —- as before.
2^
( 57 )
PROB. (39.) GIVEN A POINT A, IN THE RADIUS BC3 OF

THE SEMICIRCLE DEB; TO FIND THE POINT E AT
WHICH, IF A TANGENT EG BE DRAWN, THE ANGLE
AEG, FORMED BY AE AND EG, SHALL BE A MINIMUM.
(Fig. 31.)
Let C be the centre, CA = a, AE = x, CE = 6, the
angle CEA = <£.
Then, since CEG is a right angle, and therefore a con
stant quantity, it follows that, when AEG is a minimum,
AEC is a maximum; and the problem resolves itself into
the determination of E when 0 is a maximum. Now, by
prop. 13th of the 2nd book of Euclid, and by principles of
Trigonometry, we find a2 = b2 + oo2 — 2bx cos 0, and there-
„ b2 -f x2 — a2 _ , . . ^
tore cos <£ = ^7 . But 0 is always less than a
right angle ; hence when <p is a maximum, cos $ will be a
minimum ;
b2 + x2-a2 . ,. , _ .
.'. ?tt = niin. which let = r,
2bx
.•. b2 + x2 — a2 = 2&*T or #2 — 26r# = a? — 62. Solving
this quadratic we find x =z brdiz \/b2r2 — b2 -{- a2, and it is
evident, by inspection of the diagram, that CB is greater than
CA . \ 62 v d2, and a2 — b2 = a negative quantity, which let
= — P2 ,\ x = brd= \/b2r2 — P2. Now it is clear that r
cannot be taken so small as to make b2r2 less than P2, and
therefore when r = min. we must have b2r2 = P2 = b2 — a2
and r = A/ —T2— and x = br = b A/ — 62
Vb2 - a2 or 62 = a2 + ^2 or CE2 = CM2 + ^4£2, and hence

it appears that CAE is a right angle.
( 58 )

In the equation x2 — 2brx = a2 — b2 let x — y -f br
... & _ 2brx = y2 + 26ry + 62r2 - 26n/ - 26V2 = y2 -
7/2 i 7,2 _ „2
#ys _ ^2 _ £2 = _ (£2 _ a2J b.# r2 _. L-Z-j- which
b2 — a2
is evidently a minimum when y — 0, .*. r2 = —^— .*. r =
/^2 __ a2 i
/y/ —-jz— = t v b2 —a2 and a? = br = Vb2—a2 as before.
PROB. (40.) TO FIND A POINT Z>, IN THE SEMICIRCLE

ADB, SUCH THAT THE SUM OF THE DISTANCES AD + DP
MAY BE A MAXIMUM; P BEING A GIVEN POINT IN THE
RADIUS BC. (Fig. 32.)
Let D be the required point : draw DE perpendicular to
AB ; also, let AC= a, AE = x, CP = b. Then by prop. 35,
3rd book of Euclid, we find, DE2=2ax—x2; therefore PD—
\/DE2 + EP2=: \/2ax-x2 + (a+b-x)2 = V\a + b)2-2bx.
Now by prop. 8 of the 6th book of Euclid AD = \/ABxAE
= V2ax .\AD + PD = s/2ax + V{a + b)2-2bx = maxi-
/
mum. Let V2ax y2
= uy .\ x = ■— by2 and there-
2a and 2bx = —a
by2
(a + b)2 — = max. which let = r, and con
sequently {a + b)2 — = r2 — 2ry + y2 .\ —— y2 — 2ry

/ , *\8 a j 2 2^r a(a + #)2 — ar2
= v;
(a -f #r — r\' and .\ ?/* a + bf vu = ~ a +~b .
Solving this quadratic we find y=-—~dz a/ —L-.
and hence it is evident that r cannot be so great as to make

( 59 )
abr2 greater than a {a + b)z, and therefore when r is a maxi-
V(a -f ft)3
—7i— '
and • „ - JUL. - AMfl + ft) and^ - ^ - g(g + 5) •
converting this into an analogy, we have 2J I a : : a + b I x.

Prom this it appears that if from AB we cut off AE, a
fourth proportional to 2CP, AC and AP, and through E
draw -£JD perpendicular to AB, meeting the circumference in
D, then D is the point required. Since x or .42£ = —£7—>

it follows that, as b decreases, x must increase, and that when
a2
b = 0, x = — = infinity. This is no doubt a fair and legiti
mate conclusion, when the value of x is viewed as an abstract
formula ; it is inconsistent, however, with the nature of the
problem before us, in which we perceive that x, so far from
admitting of indefinite increase, can never exceed the diame
ter AB or 2a. This limit above which x cannot ascend, will
naturally fix a corresponding limit, below which b cannot
descend ; to reach this we have merely to substitute for x its
greatest value 2a in the equation x = , ; the resolu-
tion of which will give the minimum value required ; thus,
2a = -L--—- .'.ft = -; that is, the conditions of possibi-
Zb o
lity fix P between B and another point distant from it by §
the radius of the circle.

_In the equation yL — 2ar 7 y — a(a
— + ft)2—
-7 ar2 let y =
^ y a + by a + b J
ar 0 2ar 0 2ar a2r2
* + a + 4 ' • y
t^ ■'• sr - a + ft *
m » = ^ + « + ft
-—7 * + (a + ft)
( 60 )
2ar _ 2aV _ 2 __ a2r2 __ a {a -f bf -■ ar2
a + b* {a + If" * {a + bf "~ a + Z>
, ,, « o a(a
and therefore rL = — + &)3 — ~ (« + S)2^2 .
— which . , is
. evi-
a#
dently a maximum when # = 0, .'. r2 = -—= —— and «/ =
—- - A /fl2(a + b) and « - £ - a(a + &) as before
PROB. (41.) OF ALL THE CONES WHICH CAN CIRCUMSCRIBE
A GIVEN SPHERE, TO FIND THAT WHICH HAS THE LEAST

POSSIBLE SOLIDITY. (Fig. 33.)
Let D mn and AEB be the circular, and triangular sec

tions of the given sphere, and the required cone the solidity
of which is to become a minimum.
Let CD == a = radius of the sphere.
CE == x and Am — y = radius of the base of the cone.
It is evident that the angle EDC is a right angle, and conse
quently the triangle EDC is equiangular and similar to the
triangle EmA .\ Em : mA :: ED I DC or oo + a \ y ::
VV — a2 : a, and therefore 2/ = ) == .* . the area of the
V x2 — a2
circle, which is the base of the cone = py2 (where p =
circumference of the circle whose diameter is unity) =
'DCt^' ( 7) —J~ Ct\2 id, -4- 0P\
-—3 7T- = pa2 x -,—-—r-~- r, and therefore the solid
or — a2 (a + a?) (# — a)
contents of the required cone = pa2 x 7—■— (x r-7
+ a)2 r x
^ ^ (a? + a) (x — a)
a? + a pa2 (x + a)2 . T ,
—-— =-^r- x = mm. Let y = x — a .\ x +
3 3 x —a
a = y + 2a, and we therefore find — x = -«
( 61 )
(y + 2a)2 . , . pa" . * * •
x — = mm. and since Sr is a constant given
y 3
quantity, we must also have — — = min. which let = r,
and therefore y% + 4ây + 4a2 = ry .• . «/2 + (4a — r) y =
4a —* 7*
— 4a2. Solving this quadratic we find y = — -—-— dz
2
a /(4« — rf ~ 4a — r , A /r(r — 8a) ,
here it is evident that r cannot be less than 8a, and therefore

when r is a minimum, we must have r = 8a, and . • . y = —
4a — r = —
—^— 4a = 2a
n and- # = y 4-, a = 3a
0 . • . i^m
_, = ac + a =
2 2
4a = twice the diameter of the given sphere. Hence it
appears that the altitude of the smallest cone which can be
circumscribed about a given sphere, is equal to twice the
diameter of the sphere.

In the equation y2 + (4a — r) y = — 4*a2 let y = z —
^Zl ... f + (4a - r) y = *» - (4a - r) * + if^zZL2
+ (4a — r)2 £ — - —'- = £2 — - ^— = "~ 4a

and therefore 4^2 + 16a2 = (4a — r)2 = (r — 4a)2 .\ r = 4a
4- a/4#2 + 16a2; here it is manifest that when r is a mini
mum we must have z = 0, and therefore r = 8a .*. y = —
4a -— r
—-— = 2a and a? = ?/ + a = 3a and Em = a? + a = 4a
2
as before."*
* Here y has been used in two different senses, but not so as to pro
duce confusion.—Ed.
( 62 )
PROB. (42.) TO FIND THAT NUMBER WHICH BEING ADDED

TO ITS RECIPROCAL THE SUM IS THE LEAST POSSIBLE.
Let x = number required and - = its reciprocal.
Now by the conditions of the problem we have x + - —

x
x2 + 1
min. or = min. which let = r, .*. x2 — rx = — 1.
x
Solving this quadratic we find x = - =b A/ — 1, and
hence it is evident that r cannot be taken so small as to make
— less than 1, and therefore when r = min. we must have
rp* y*
— = 1, .*. r = 2 and x = — = 1.
4 ' 2
7%e âme solved without impossible roots.

T
In the equation x2 — rx = — 1, let a? = y + ~ and there-
n*2 n*2 rt*2
fore x2 — rx = ?/2 -f ry + ry —— = «/2 — -p = — 1,
. • . r2 = 4y2 -f 4 which is evidently a minimum when y = 0,
.• . r = 2 and <# = —- = 1 as before.
PROB. (43.) AC AND BD BEING PARALLEL, IT IS REQUIRED

TO DRAW PROM C A LINE CXY SUCH THAT THE SUM OF
THE TRIANGLES ACX AND BXY SHALL BE A MINIMUM.
(Fig. 34.)
If AC = a, AB = b} AX = #, it is easily seen that the
area of the triangle ACX is proportional to ax, and that of
T^Tr i a(b — x)2 ,, , , / (J — a?)2)
i*AF to ~ -. so that we have a <x + - —> =
x \ x )
( 63 )
minimum, and therefore x + = *nin. which let = r,

x
.-. x2 + #2 — 2bx -\- a?2 = rxy .*. a?2 —- —-— x = —■ — .
Solving this quadratic we find
2b + r ^ /452 + 4br + r2 - Sb2 __ 2b + r
^? =
4 ~ V 16 "~ 4
(4* + r)r — 462
=*= V" 16
and here it is evident that r cannot be taken so small as to
make (46 + r) r less than 4b2, and therefore when r = min.
we must have r2 + 4br == 4b2 .*. r = a/8&2 — 2b and a? =
2b + r 2b-2b + \/W 2bV~2 b
which deter-
4 4 4 ^/ 2
mines the line CXY.
In the equation x2 — «#= — ■7r-letx = y-i :—
2b + r 9 2b -f r (2b + r)2 2b + r
2# _ A —L _. ^2 _ v —J_ _. __ an(j therefore r

u 8 * 16 2
= \/l62/2 + 862 — 2by which is evidently a minimum when
y = 0} .\ r = \/8Z>2 — 2b and a? = —-— = -7= as before.
4 V2
( 64 )
PROB. (44.) TO FIND THE HEIGHT ABOVE THE GIVEN POINT
A FROM WHENCE AN ELASTIC BALL MUST BE SUFFERED TO
DESCEND FREELY BY GRAVITY, SO THAT, AFTER STRIKING
THE HARD PLANE AT B3 IT MAY BE REFLECTED BACK
AGAIN TO THE POINT A, IN THE LEAST TIME POSSIBLE,
FROM THE INSTANT OF DROPPING IT. (Pig. 35.)
Let C be the point required, and put AC = x, and AB =

a ; then the spaces of falling bodies, by the force of gravity
being as the squares of the times, we find CB = gt2, and CA
= gt2, where g = 16 feet nearly, and consequently t = —j=.
v g
VCB and f = -yLVCA, and therefore t - f = -yLVCB
Vg Vg
7= VCA = —j= \/a + oo y= V oo = the time down
Vg vg Vg
AB} or the time of rising from B to A again : hence the
whole time of falling through CB and returning to A is
—j^Va + x j=.\/oc -f —^.\/a-\- #= -y^(2\/a-{-x — Voo)
Vg Vg vg vg
= min. and as —p=. is a constant given quantity, we must
v g
have 2\/a -f- oo — a/^ = min. which let = r .' . 2v# + oo =
r + aA?. Now let \/oo = ^ . • . oo = y* . • . 2\/« + «/2 = r + y
and squaring both sides of the equation we find 4a •+ 4y2 =
r2 + 2ry + y2 and «/2 —— y = —5 . Solving this qua-
r /4r2 _ \2a
dratic we find y = - db A/ — , and here it is evident
that r cannot be taken so small as to make 4r2 less than 12a,
and therefore when r = min. we must have 4r2 = 12a, and
therefore r = \/Za = a/ a a/ 3 and «/ = = A/ ~
and off = 2/2 = —-, that is ^(C = J ^JB.

( 65 )

t ^ i.- 2 2r r* — 4« , , r
In the equation ^ — y = —- let 2/ = # -f -55
., , . 2r 2 , 2r r» 2r 2r2
therefore r — -g- y = s2 + "3 * + 9" ~ ~3 z ~ ~cf ~
2 r2 _ r2 - 4a _ 3r2 - 12a , _ r^_ 3rf _
^ ~ 9" ~ 8 ~ 9 •'• ^ ~ 9 + 9
^ and 4r2 = 9*2 + 12« .-. r = a/9** t ^ which is
evidently a minimum when # = 0, . \ r = v 3 v « and 2/ =
~ = /y -|- .*. # = ?/2 = -|- as before.
PROB. (45.) GIVEN THE HEIGHT OF AN INCLINED PLANE;

TO FIND ITS LENGTH, SO THAT A GIVEN POWER ACTING
ON A GIVEN WEIGHT, IN A DIRECTION PARALLEL TO THE
GIVEN PLANE, MAY DRAW IT UP IN THE LEAST TIME
POSSIBLE.
Let a denote the height of the plane, w its length, p the

power, and w the weight. Now the tendency down the plane
is = gw sin. of the angle made by the length with the base
of the plane
r = gw w- = -—,
w where a = force of gra-
vity = 32^ feet, and the tendency up the plane = gp .'.
the whole motive power up the plane = gp — -— =
ML IK . but the mass resisting this motion is p + w, there-

w
fore the accelerating force for raising the weight upon the
( rvjQQ , ciijo\ n
plane is equal
x to ML(p + w)w
-r~. Now the space
x ascended =
( 66 )
A9 = ~
w — ff {poo —aw)g
r^- tf2« where/
, * = *force .• . ,o {p + w)^2
f = -r^- •—--
J {p + w)% J {poo — aw)g
flu1
= min. and .-. = min. which let = r, and there-
px — aw
fore 00* = proo — awr . * . x2 — pr# = — «wr. Solving this
pr _^ A /p2r2 - 4<awr pr
quadratic we find, x —- — db A/ 7 == "2"
r^r —^Z^ an(j hence it is evident that r cannot be

^
taken so small as will make p2r less than 4aw, and therefore
when r = min. we must have »V = 4«t<; and r = —r . * . a?
~ £_ = and . • . » : w : : 2a : 00 : : double the height of

2 p
the plane : its length.
In the equation x2 — proo = — awr let 00 = ^ + —, there-
fore #2 — ^n*? = ^2 + ^ri/ + —A pry — -^~ = y2 —

~4~
»V2 = — awr .\
1-— -99^ ^2 or r42
#V2 — 4awr = 4?/z âw
r %«■
r = —
4 -^ * ^2 i?
and therefore r = —7- + \f 1 which is evi-
dently a minimum when «/ = 0, and . • . r = 4$'2#

—g- and # =
»r = 2aw■ as before.
^— , c
*** Here p and w are masses, not weights, as stated ; and ■§■/ should
have been used instead of/.—Ed.
( 67 )
PROB. (46.) A LARGE VESSEL OF 10 FEET, OR ANY OTHER

GIVEN DEPTH, AND OF ANY SHAPE, BEING KEPT CON
STANTLY FULL OF WATER, BY MEANS OF A SUPPLYING
COCK, AT THE TOP; IT IS PROPOSED TO ASStGN THE
PLACE WHERE A SMALL HOLE MUST BE MADE IN THE
SIDE OF IT, SO THAT THE WATER MAY SPOUT THROUGH
IT TO THE GREATEST DISTANCE ON THE PLANE OF THE
BASE. (Pig. 36.)
Let AB denote the height or side of the vessel; D the
required hole in the side, from which the water spouts, in the
parabolic curve DG, to the greatest distance BG, on the
horizontal plane.
It is evident that the velocity of the water descending
from A to D with which it must spout out in the horizontal
direction must be expressed by the equation v = v2gs =
V~2 x V~s Vff = V~2 x VAD x \/~g (1)
It is also evident that the time t in which the water spouting
out from the hole at D must reach the ground, must be the
same in which it may descend from D to B and t2 = -j— =
"2 O
2DB , __ \/~2 x VDB
—— .'. t — y— (Z)
9 Vg
Multiplying the equations (1) and (2) we find tv = hori
zontal space GB = 2vADDB = maximum, and supposing
AB = a and AD = % we find 2\/x(a —• x) = 2\/ax — x2,
— max. and .*. ax — x2, = max. which let = r, and therefore
x2 — ax = — r. Solving this quadratic we find, x = — =b
a
V — — r, and hence it is manifest that r cannot be greater
a
than —, and consequently when r = max. we must have
( 68 )
"2 a
—- = r and x = -. So that the hole must be in the middle
4 2
between the top and the bottom.
In the equation ax — x2 let x = y -f -, and therefore we
find &# — x2 = ay + — — y2 — «y — ■—- = — —- y2 which
is evidently a maximum when ?/ = 0, .• . x — - as before.
PROB. (47.) IF THE SAME VESSEL, AS IN PROBLEM 46,

STAND ON HIGH, IT IS PROPOSED TO DETERMINE WHERE
THE SMALL HOLE MUST BE MADE, SO AS TO SPOUT
FARTHEST ON THE SAID PLANE. (Fig. 37.)
Let the annexed figure represent the vessel as before, and

bG the greatest distance spouted by the fluid DG, on the
plane bG. Here, as before, bG = 2\/ADDb = 2\/x(c — x)
= 2\/cx — x2, by putting Ab = c} and AD = x. So that
2\/cx — x2 or ex — x2 must be a maximum, which let = r,
and therefore x2 —■ ex = — r. Solving this quadratic we
e /~&
find, x = — =fc A/ — r, and hence it is evident that r
c2
cannot be greater than —-, and consequently when r is a
maximum we must have r = —r c2 and therefore x = •—•.

c So
4 2
that the hole D must be made in the middle, between the top
of the vessel and the given plane, that the water may spout
farthest.
The same may be solved without impossible roots, as
problem (46.)
( 69 )
PROB. (48.) TO DIVIDE A NUMBER a INTO TWO SUCH PARTS

THAT IF THE SQUARE OF ONE OF THESE BE SUBTRACTED
FROM THEIR PRODUCT, THE REMAINDER IS THE GREATEST
POSSIBLE.
Let x — one of the parts, and therefore a —- x = the other

part, .\ ax — x2 = product of the two parts, and x2 =
square of one of them, and therefore ax — x2 — x2 = ax —
a t
2x2 = max. which let = r .*. x2 —x= — . Solving
a / Cu 7* a
this quadratic we find x = — db A/ — — = — dz
a? __ %r
a/ ——, and hence it is manifest that r cannot be taken
so great as to make 8r greater than a2, and consequently
when r is a maximum we must have a2 = 8r, and therefore
~~ ~4 '
In the expression «# — 2#2 or its half I- x — <£>2J which is
made a maximum, let x = ?/ + — and therefore — x — <^2=

4 ^
tt ft 0/ ft ft
2"2/"l"8~2/2~2"^""l6==16""2/2 Which iS evidently
a maximum when y = 0, and .•.# = — as before.
PROB. (49.) TO FIND THE POINT IN THE LINE JOINING THE

CENTRES OF TWO SPHERES FROM WHICH THE GREATEST POR
TION OF SPHERICAL SURFACE IS VISIBLE. (Fig. 38.)
Let npA and Dgs be two great circles of the two spheres
in the same plane, AD their common tangent, and C and m
( 70 )
their common centres. Also let Cm = c, Cv = a> wm = b,
and CB = x. Now by similar triangles (prop. 8th of 6th
book of Euclid) we have CB : CA\\CA\ Cb, or x : a :\ a :
a2 a2
Cb — — .* . bv = Cv — Cb = a and dw = mw — dm
x x
b2
= b . The surface of the spherical segment whose
C "~~~ X
height is bv = 2pa x bv (p = circumference of a circle

) and the surface a2
x '
of the spherical segment whose height is wd = 2pb x md
ft j and therefore the sum of the surfaces
cC ~~—
— X'x^
(/v3v
a2 J -f
2p (# *!-) = 2p (a2 + 62 - -• — ) = max.

\ C — XI \ X C — XI
and since 2p is a constant given quantity, we must also have

/ a3 b3 \
a2 + b2 — I 1 ) = max. which let = q> therefore
\x C — XI
a3 b3
j = $ _j_ ft __ qt Here it is evident that when
X C — X
q = max. a2 -\- b2 — q must be a minimum, which let = r,
., ^ „ a3 b3 ca3 -f- (b3 — a3)x . , . .
and therefore X 1 C — X = ex — XL5—— = mm. which
CG,3 -4- (b3 — Cu) X

let = r .-. - ;—— = r. Now let x —
y + i'
therefore ""+(»-«> = «* + <* " *> FTl =

Cc2? —- X2, c£ cl
^(y+l^+ffl*--^) (y + l) _ ay + 2asy + a3+ (b*-a?) y + V—cfi

cy — ey
63 - a3 + 2a3 a3y2 + b3 . _ . b3 — a3 + 2a3
= j -L _ min# anc[ smce
c cy e
( ?1 )
is a constant given quantity, we must have — = min.

cr b^
which let = r, .\ cfiy1 + 63 = cry and s/2 — —y= — — .
Solving this quadratic we find y = — ± A/ —

and here it is evident that r cannot be taken so small as to
make cV2 less than 4a3£3, and therefore when r = min. we
must have c2r2 = 4a3£3, .\ r = —— and y = ■— = -f ;
7 c 2a3 «t
c c ca%
therefore a? =
y+ l *f+1 a* + H'
at-
cr b^ cr
In the equation y2 - -^ y = - — let y = * + ^3
2 cr 2 cr c2r2 cr c2r2 _
•'• y ~ "5^ = * + "^* + 4^~"^~2^ =
2 c2r2 £3 8 4«%2 + 4«363 , . , . . , ,.
— r-g = - .-. r2 = g which is evidently
Tift # C
,
a minimum when „ . • . r = 2af6£ and, y = ^-z
a? = 0 cr = —-
if
c 2a3 «f
c caf
.\ a? = ——= = as before,
y + 1 of- + H
*#* Inaccurate description of the figure : J5D and BA are not in the
same straight line. —Ed.
PROB. (50.) TO FIND THE VALUE OF THE ANGLE X WHEN

m sin. ix — a) cos. x = maximum.
It is evident that m being a constant given quantity, we
must have sin. (x — a) cos. x = sin. a? cos. a cos. x — sin. a
cos.2# = max.
Now let cos. x = y, cos. a ~ b, and sin. a = v 1 — #2 = c
( 72 )
.*. by \/\ — y2 — cy2 = c (— \/y2 — y* — y2) = max. or
— \/^2 — y4* — y2 = max. which let = r, and therefore —g-

c c
V , . a , , S2 + c2 Z>2 - 2rc2
y - 7T 2/* = y* + %y r + ** <>r t -2—
y2 = - r\ But b2 + c2 = b2 + 1 - b2 = 1, and therefore
—]_ ?/4 J2 _2rc2 y2 = — r2 . * . y4 —• (62 — 2rc2) y2 — — r2c2.
Solving this quadratic we find

p _ 2rc2 ^_ A /6* - 46Wr - 4r2c2 (1 - c2) m
Now c is the sine of a given angle, . \ c2 must be less

than unity, and consequently 1 — c2 must be positive, and
hence it appears that, excepting &4, all the rest of the terms
in the numerator of the fraction under square root are
negative, and for this reason we cannot take for r so great
a value as will make 4b2c2r 4- 4r2c2 (1 — c2) greater than b* ;
hence when r = max. we must have
4r2c2 (1 — c2) + 4&2c2r = 54, and from this quadratic
b2 b4*
we find r2 + ^ 5r = A 0 „ ^ and therefore r =
1 — c2 4c2 (1 — c2)
£4^2 + £4 _ £4^2 J2 5!
^ 4C2 (i __ C2)2 2(1 - c2) 2c (1 - c2) 2(1 - c2)
—4; - c2)
2c(l ^r = -——
2c (1 + c)-. Now from equation
H (1)
w we find
b2~2rc2 70 62c J2 cos.2 a
1 + c 2(1 + c) 2(1 + sin.a)
2
1 — sin.2a 1 — sin. a „a n . a a . 9a
———: r == = cos.2- - 2 sin. - cos.- + sm.2-
2(l + sm.a) 2 2 2 2 2
2
cos. Z- 7=
— sin. -V
2/I .•. y = cos. %7ra x V21
—7= sm. - x
•F ' - v. '» •»
( 73 )
cos. — cos. 45° — sin. —sin. 45° = cos. (45 + — ) but y

= cos. aOy .•.#= — + 45°

In the equation */4 - (b2 - 2rc2) y2 = - cV2 let «/2 = 3-
+ ?—-fll ... y* - (i» _ 2rc2) J/3 = *2 + (i» - 2rc2) « +
(^L^£l2 _ (* _ 2rc>) g _ ^_z_^)! = .* -

4 2
(ff „ 2rc2)2 _ _ cV2 #.o 4^2 _ 54 + 4^2c2r _ 4r2c4 = _ 4CV2
4
4cV2 (1 - c2) + 462cV = i* - 4*2, but 1 - c2 = b2 and .-.
54 __ 4^2
r2 -4- r = —,4c26z
070 . Now since r = max. we must nave
£4 _ 4^2
r2 + r = max. or its equivalent — ,2 2 must be = max.
which can only happen when z = 0 ,\ r2 + r =

4c262
V^2C2 _j_ £4 J
—ITVP O
/¥W+^) _ x _ /ff + 1 - a2 _ 1 _ i_ _ 1
V 462c2 2 ~ v 4c2 2 ~~ 2c 2
1 _ c 1 - c 1 + c 1 - c2 62
X
2c 2c 1 + c 2c (1 + c) 2c (1 + c)
62 . b2 — 2rc2
Now from equation ?/2 = £ 4 - where
2c(l + c) ^ * 2
.
z = 0, we find ^2 = 62 -—•2rc2 = bLw 2«W b2
2c (1 + c) 2(1 + c)
as before. 2
This is the solution of the problem to find in what direc
tion a body must be projected with a given velocity, that its
range, on a given plane, may be the greatest possible.
( 74 )
PROB. (51.) TO FIND CC WHEN — ^ IS A MAXIMUM.
Since a2 is a constant given quantity we must have

oc{a — oc) ~ aoo — %% — max. which let = r .' . oc% —• ax = — r
/7 fa2
and solving this quadratic we find, a? = — =t /\/ x ~" r#
of
It is manifest that r cannot be 53greater than —
4 and therefore
we must have r = -7 when it is a maximum. .•.<# = -.
7%e same may easily be solved without impossible roots.

This is the solution of the optical problem to determine
the position and magnitude of the least circle of aberration.
PROB. (52.) A REGULAR HEXAGONAL PRISM IS REGULARLY

TERMINATED BY A TRIHEDRAL SOLID ANGLE FORMED BY
PLANES, EACH PASSING THROUGH TWO ANGLES OF THE
PRISM; FIND THE INCLINATION OF THESE PLANES TO THE
AXIS OF THE PRISM, IN ORDER THAT, FOR A GIVEN CON
TENT, THE TOTAL SURFACE MAY BE THE LEAST POSSIBLE.
(Fig. 39.)
Let ABCabc be the base of the prism, PQRS, one of the
faces of the terminating solid angle passing through the
angles P, R.
Let S be the vertex of the pyramid. Draw SO perpendi
cular to the upper surface of the prism. Join OM, RP, SQ
intersecting each other in N. Then it is easy to see that
MiV= NO and consequently SO = QM, and, as the triangles
POR, PMR are equal, so that, whatever be the inclination
of SQ to ON, the part cut off from them is equal to the part
( 75 )
included in the pyramid SPR, and the content of the whole,

therefore, remains constant. We have then to determine the
angle ON$, or OSN, so that the total surface shall be a mini
mum. Let AB, the side of the hexagon, = a, AP, the
height of the prism, = b, OSN = 0. Then ON = MN = ia,
and SN = \a co.sec. 0, and QM = \a cot. 0. The surface
APBQ = \a {2b - \a cot. 0) . The surface PQRS = PRxSN
= —^— co. sec. 0. Whence the total surface of the solid is
3« (2b - — cot. 0) + ^ co. sec. 0 = Gab - ^ cot. 0 +

3#2 3&2
— 3^ co. sec. 0 = 6ab -f -^- (3* co. sec. 0 — cot. 0) = min.
2 2
and therefore v 3 co. sec. 0 — cot. 0 = min. which let = r.
Also let cot. 0 = x and ..• . co. sec. 0 = \/l + oc2 . • . a/3 -f 3a?2
— <#> = r, and therefore 3 + 3<2?2 = co2 + 2r<#> + r2 and a?2 —
r2 — 3 . r
Solving this quadratic we find cc = — -f
2
- and it is now evident that r cannot be taken so

4
small as to make 3r2 less than 6, and therefore we must have
3r2 = 6 when r — min. . • . r = V/— 2 and oc = — r = —p=.
1 or
2' a/2
cot. 0 = —7=. and tan. 0 = \/ 2 . Hence tan. /KR2V =
and SRQ = 2i
7%e 5«me solved without impossible roots.

r2 — 3 y*
In the equation w2 — r# = —-— let # = ^ + -^ .*. w2 —
y%2i m2 A.2 y.2 _____ Q
too = ?/2 + ry + -j ^/ — = ?/2 -

4'* 2 * 4 2
4w2 + 6
4^2 + 6 = 3r2 and r2 = -^--— which is evidently a mini-
o
( 76 )
mum when w = 0 . \ r = A / — = \/ 2 and # == —- = —r^

V 3 2 y^2
as before.
This is the celebrated problem of the form of the cells of
bees. Maraldi was the first who measured the angles of the
faces of the terminating solid angle, and he found them to be
109° 28' and 70° 32' respectively. It occurred to Reaumur
that this might be the form which, for the same solid con
tent, gives the minimum of surface, and he requested Konig
to examine the question mathematically. That Geometer
confirmed the conjecture; the result of his calculations
agreeing with Maraldi's measurements within 2'. Maclaurin
and S. Huillier, by different methods, verified the preceding
result, excepting that they showed that the difference of 2'
was owing to an error in the calculations of Konig, and not
to a mistake on the part of the bees.
PROB. (53.) TO FIND SUCH A VALUE FOR 00 AS MAY MAKE

co
A MAXIMUM.
{a -f co) (b + oo)
CO
It is evident that when r-^ r = max. we must
[a -f oo) (b + co)
have (a + oo)—(b + oo) = mm.. and, . • . ab + (a + b) co + co2, =
00 co
ab ~\~ co
min. or a -f b -\ — = min. and as a + b is a constant
co
(lb -4- CO
given quantity, we must also have = min. which
co
let = r . • . co1 — too = — ab. Solving this quadratic we find
t* /y
a? = — ± \f ~A ab. and here it is evident that r cannot
2 V 4
t2, ab
be taken so small as to make — less than — and therefore
4 2
( 77 )
when r = min. we must have — = ab, . • . r = 2^/ab and # =

V
In the equation
it: x* -— rx = — ab let x = y 4- -~
.-. ^2 - ra? = ^2 + ry + •— - ry - ^- = y* - ~- = - ab
. • . r2 = %2 + 4a& which is evidently a minimum when y = 0,
. • . r = 2v#6 and # = — = \/«6 as before.
This is the solution of the dynamical problem to find the

magnitude of the body which must be interposed between
two others, so that the velocity communicated from the one
to the other shall be a maximum.
PROB. (54.) THE DIFFERENCE OF TWO NUMBERS BEING

GIVEN, TO FIND IN WHAT CASE THE THIRD PROPOR
TIONAL TO THE LESS AND THE GREATER OF THEM IS A
MINIMUM.
Let a = the given difference of the two numbers, x =

greater number, and therefore x — a == the lesser number.
We now have x — a \ x :: x I x — a = the third propor-
x x
tional required = min. which let = r, . • . x% — rx = — ra.
t =fc A//t^
Solving this quadratic we find x = — 4ya = •—
7*
& V 4 Z
± aJ —T (r —- 4a), and here it is evident that r cannot be
taken so small as to become less than 4a, and consequently
t 4a
when r = min. we must have r = 4a, .\ x = — = — = 2a
( 78 )
= greater number, and the lesser number =a? — a = 2a — a
= a. Hence it appears that the third proportional required
is the least possible when the greater number is double the
lesser number.

T
In the equation x2 — rx = — ra let x = y + — and
A*2 (VtZ yS,
therefore x2 — rx = y2 + ry -f — — ry -r- = y2 —
= — ra, . * . r2 — 4ra = 4«/2 and r = 2a + \Zty2 + 4a2 which

is evidently a minimum when y = 0, . • . r = 4a and a? = —
— = 2a as before.
PROB. (55.) THE CONTENT OF A CONE BEING GIVEN, FIND

ITS FORM WHEN ITS SURFACE IS A MINIMUM.
X the altitude, and y the radius of the base.

Let ioa3 'du so
~- be the given content = . • . ^-%-.
Then & = surface = convex surface + base.

But convex surface = sector of circle, of which the radius
is the slant side, and the arc the circumference of the base of
cone, .• . u = py V x2 + y2 + py2. a3 .'. y2 + x2
But y2 = —
as +
+ x3
tf3 .•. u = »a*33 <^fvV
fva3 + x3 +
+ a?3 + an
at \ = mm.
. XT
JNow
<2? I X J
as pa% is a constant; given quantity we must have
\/ a3 + #3 + «f
= min. which let = r, and ,\
a?
Va3 + a?3- + at
= r, and Va3 + x3 = rx — a%; squaring
x
both sides we find a3 + x3 = r2<#2 — 2r#as + a3, and there
( 79 )
fore x2 — r2x — 2a* and x2, — r2x — — 2ra*. Solving this
quadratic we find x = — -f A/ —-—~ and here it is

evident that r cannot be taken so small as to make r3 less
than &a%, and . " . r3 = 8a^ and r = 2a* . * . r2 = 4a and # =
r2 4a
In x2, — r2# = — 2ra* let a? = y -f -^- and therefore

^4 a»4 y»4
^2 _ r2# ___ ^2 _|_ r2^ _j_ r2^ = ^2 = __
2ra^ .•. r4= 4?/2 -f 8ra* which is evidently a minimum

when y = 0, . • . r4 = 8ra^ and r = 2a* and r2 = 4a ; there
fore x = -75- = 2a as before.

CHAPTER II.
PROBLEMS OF MAXIMA AND MINIMA IN THE SOLUTION OF

WHICH CUBIC EQUATIONS ARE USED.
Before reading this chapter the article on u Reduction of

Equations/' in the Introductory Chapter, must be studied
with great care, for this reduction is effected in almost every
problem which follows.
PROB. (1.) WHAT IS THE FRACTION, THE CUBE OF WHICH

BEING SUBTRACTED FROM IT, THE REMAINDER IS THE
GREATEST POSSIBLE?
Let x — the fraction required, and the greatest remainder

= r, . • . x — x3 = r and x3 — x = — r, . * . x3 — x + r = 0.
In order to solve this problem merely by means of quadra
tic equations, let one of the negative roots of this cubic
equation == — a, and it is evident x -f- a must exactly divide
xs — x + t = 0, and therefore the following process is
obtained.
x + a) x3 — x + r = 0 \^x2 — ax -f «2 — 1 = 0 ... (A.)
x3 H- a^
— ax2 — x
— ax2 — a2x
(a2 — l)x + r
(a2 — 1) x + a3 — a, .*. r must be = a3 — a,
and .*. a2 — 1 = — .•. by equation (A) we find x2 — ax
V T
4- — = 0, and x2 — xa = — — . Solving this quadra-
tic we find x = —- =fc A / —-A and here it is evident

2 V 4a
( 81 )
that the greatest value of r is when a3 = 4r = 4a3 — 4a
.-. a = —7= and a? = — = -7= = the required value
of a?.
In the equation x% — ax = let a? = ^ + -^ .°.#2—■
a2 a% 9 a* r
ax=y*+ax + —-ay- — =:y*--T=- —
a3
,\ r = -: ay2, which is evidently a max. when 2/ = 0, .*.
4
r = — a3 ; but r — a% — a, .•. 4a3 — 4a = a3, .\ 3a3 = 4a
4
and. a = -7= 2 .•.# = — a = —7=. 1 as ,before.
„
\/3 ^ a/3
PROB. (2.) WHAT IS THE FRACTION THE CUBE OF WHICH

BEING SUBTRACTED FROM ITS SQUARE, THE REMAINDER IS
THE GREATEST POSSIBLE?
Let x = the fraction required, and the greatest remainder

= r ." . x2 — x3 = r . • . «2?3 —- x2 = — r, or #?3 — a?2 + r = 0.
In order to eliminate the second term of this equation, let
x = 2/ + J ; and by this substitution we find,
** = (y + i)3 = s/3 + sf + 12/ + -^

2 1
- #2 = - (y + i)2 = ' - 2/2 - -g- ?/ - g-
r = + r.
2 == 0.
.', ^ - f + »• = |/3 - ^ + r - —- Let
one of the negative roots of this equation = — a, .•.
( 82 )
V + «J f - iy + r - J^ = 0 If - ay + a2 - £ = 0...(A.)
27
2/3 + ^2
- ay* - £y
ai/2 — a2y
2
(aa _ i) ?/ + r - _
(fla-i)y + ^-4
, a 2
which must also be greatest.

r'
From this equation a2 — J = — and therefore from equa-
tion (A) we find f* — ay -\ = 0, .'. if — ay = .

a a
Solving this quadratic we find y = - -±: A/ — > an(^
here it is evident that when r' = greatest quantity possible,
4« 4 2
we must have a3 = 4r' = 4a3 — — .'. 3a2 =-.'.« = -•
This problem may be solved without eliminating the second
term of the cubic equation, in the following manner.
Let one of the negative roots of the equation x% — x2 -f
r = 0 = — a, and therefore—
oc + a) x3 -x2 + r = 0 ^2 - (a + 1) a? + «2+ « = 0...(A.)
<2?3 + a<#2
— (a + 1) #2 + r
— (a + 1) a?2 — « (« + 1) #
(a2 + a) # + y
(a2 + a) a? + a (a2
—+ «) . • . a (a2 -f a)
T
= as _|„ a2 = r an(j tt2 _j_ a _ __ anc[ therefore from equa-
' a
tion (A) we find x2 - {a + 1) # + — = 0 or ^ - (a + 1) a?
r a + 1
Solving this equation we find x
a' 2
( 83 )
did -f- I)2 • 4&T
V ~
4a
' and here it is evident that when r ==
greatest quantity possible, we must have a (a + I)2 = 4r =
4a3 + 4a2 .'. a2 + 2a + 1 = 4a2 + 4a, or a = J and a? =
a+
II,
-— d= A //a-i——/
(a + 1 2. - 4r = Js_-—
+ 1 d=
_,_ 0A = I9 as before,
, ,
2 — 'V 4a ""2 2 3
Tfte same solved without impossible roots.

r' a
In the equation f~ ay = - let y = £ + and
a ~2
a2 a2
therefc>re y2, — ay -.= z2 + az + z2 -
4
a2 r' a3 • — a^2
.\ r' which is evidently a max.
a "" 4
when # = 0, . • . rf = a3 but r' == a3 - a t a3 :a3 a
~ 4'' ~3
4a n 2 a
.-. 3a3 = "tt and a — 3« Now y = ^ and a? = y + i
3 ~2 =
= § as before.
PROB. (3.) TO DETERMINE THE DIMENSIONS OF THE LEAST

ISOSCELES TRIANGLE A CD THAT CAN CIRCUMSCRIBE A
GIVEN CIRCLE. (Fig. 40.)
Let OS = the radius of the given circle = a, and DO —
the distance of the vertex of the triangle from the centre =#.
Now the triangles DBG and DOS having the angle ODS
common and the angles at B and S right angles, are similar
.\DS : os :: db : BCov\/x2-a2: a :: a+x : bc.-.bc
= a■■■,(a + x) and the area of the triangle
. = BC x DB
Var — cr
= — which being a min. its square must also be a
V^2— a2
,.
mm., and consequently, (a
x-g -f x)* .
j or its equivalent (a + ^)3
———
x -— a x — a
( 84 )
aii, rt (a 4- %y
is a mm. Also let y — % -{- a ,\ y — 2a — % — a r . —
which let = r .'. y3 — ry 4- 2ar = 0. Let a

y — 2a
negative root of this equation = — b . • . y + b must exactly
divide y3 — ry 4- 2ar = 0 .*. we shall have the following
process—
y 4- bj y3 - ry + 2ar = 0 l^y2 - Z>y 4- 62 — r = 0 ...(A.)
2/3 + by2
— %2 — ry
— by2 — #2y
(62 — r) y 4- 2ar
(fl» - r) y 4- 6(62 - r) r. b3 — br = 2ar
b3
,\ r = - r. Also from equation (A) we have y2 — by —
2a -|- 0
t« __ ^ an(j # . # y _. — ± a/ r —. Now if r be the least

362 63 34* -,
possible, we must have r = -^- or ^——r = -j- or 4o =
6a 4- 36 or 5 = 6a, and y = — = -— = 3a, and x = y — a
= 3a — a = 2a = the value required.
In the equation y2 — 6y = r — 62 let y = 2 4- ^ .*. y2
- 6y == ^2 + fer 4- bz — — — z2 T = r — b2 .-.
4 <s 4
£2 gJ2
r = 22 + b2 — — = ^2 4- -p which is evidently a min.
362 , ■ b3 Zb2 b3
when z = 0, .*. r = — : but r = -——% .*. -r- = r—■—■?
' 4 ' 2a + b 4i 2a 4- b
and 6a 4- 36 = 46, .• . 6 = 6a and y = --=—- = 3a, and
we therefore find # = y — a = 3a — a = 2aas before.

( 85 )
PROB. (4.) TO DETERMINE THE GREATEST CYLINDER dg THAT

GAN BE INSCRIBED IN A GIVEN CONE ADB. (Kg. 41.)
Let a = BC, the altitude of the cone

b = AD, x the diameter of the cylinder, considered
as variable; j? = f— -j :). Now it is evident that the
area of the circle frgs = px2, and by similar triangles AC :
£C :: ^^ : 4f or -^- : « :: —^~ : d/ = —^— (A.)
And the solid content of the cylinder = ~^-— =
j- x {bx2 — x3) which is a max. .\ bx2 -- #3 is a max. Let

bx2 — x3 = r, . • . a?3 — te2 + r = 0, and # = y -f —, and
making this substitution we shall find x3 — bx2 + r = ^3 —

Z>2 2#3 £3
— y + r — — , also let r — — which is a max. = r' and
.'. ys b2
5- 2/ + r' = 0; and proceeding as in prob. (2) this
o
problem may easily be solved. We however subjoin the
process.
Let a negative root of this equation = — c} .\y + c must
exactly divide y3 —— b2 y + rr = 0.
. , . b2 . , „ , 0 . . 62 ^ _%
y 3 ^ L^ w...V
* 3 "
«/3 +~cy2
&
~ cy 3" y
- q/2 - c2#
/ 62\
62£
3
62C
4 . / = c3 -
3
( 86 )
b2 r'
and c2 — = — .\ from equation (B) we find y2 — cy +
o c
/ / c _^_ A /c3 — V
Now in order that r' may be the greatest possible, we must

bs b2c
have 4r' = c3; but r! = r — ^- = c3 — .-. c3 = 4c3 —
— or c*= 4C* - — or 3c- == —, .-. c = ~,.-.y = - =
— and oo — y + — = —. Also from equation (A) we have

o o o
b _ — 2a
df= 3=05 — = —y and hence it appears that
o
the inscribed cylinder will be the greatest possible when the
altitude thereof is just \ of the altitude of the cone.

Tr C
In the equation y2 — cy = let y = z + — and
therefore y2 — cy — z2 + cz + — a? = z2 —
c2 =
— 7*' and therefore rf = —c3 cz2 which is evidently
c3 ; but r' = cs
a maximum when z = 0, . • . r' = —r b2c
— and
4 o
.*. c3 = 4c3 — ,\ 3cz = — and c = — . Now y = —
o o o &
= — and therefore a? = y + ir=:"^~ + ir:::::~^~as before.

o o o o o
( 87 )
PROB. (5.) TO DETERMINE THE DIMENSIONS OF A CYLINDRIC

MEASURE ABCD OPEN AT THE TOP, WHICH SHALL CONTAIN
A GIVEN QUANTITY (OE LIQUOR, GRAIN, &C.) UNDER THE
LEAST INTERNAL SUPERFICIES POSSIBLE. (Fig. 42.)
Let the diameter AB — x, AD = y, p = 3.14159 &c. and

c = the given content of the cylinder. In this case it is
evident that px will be the circumference of the base, and
consequently, by multiplying it by y, the altitude, we shall
find pxy = the concave superficies of the cylinder. It is
also evident that since ~- = half the circumference and -
Z A
t)X
= half the diameter of the base,9 we shall have -~r
4 — the area
of the base, which, being multiplied into the altitude y, we
shall have —~- — solid content of the cylinder = c, .*.?/ =
4
4c 4c
—2 •' • poey = — and consequently the whole surface of the
px X
'. ., == 4c hf px*
cylinder 1 . , is
—r which . a minimum.
. . T , 4c h x—-
Let px*
J x 4 x 4
4t 16c
= r, and .*. 16c + px3 = 4r<# .*. ^3 a? -\ = 0. Let
p p
one of the negative roots of this equation = — a and there-
fore x + a must exactly divide xs 4t* x H 16c = 0.
P P
x + a\ x3 -x + —- = 0 I x2 — ax + a2 — — =0...(A.)
P P P
x3 + ax*
— «<ar x
p
2 2
~~""' il/Ju ~~~ CO Jb
/ 9 4r\ loc
4r\ 0 4«r
—
pi
<# + a p
~—— — . We therefore
( 88 )
find «a _ ___ = I_£ and ... r =**=™°. From equa-

p p 4«
4w a
tion (A) we find x2 — ax = a2, and x = — =fc
V4? 3#2
_ Now in order that r may be the least pos-
^? 4
sible we must have 4r
— = 3a2
— or ■*pa3 - 16c = 3a2
—r or _/?#*, =
p 4 #/? 4
64c and « = 4x \f ~ and ^ = "f" == 2 x \/ ~Z* Now
because px3 = 8c and j_w22/ == 4c , * . _p#3 = 2px2y . • . x = 2y
and 2/ = A/ —, hence y is known, and from this it appears

that the diameter of the base must be just double of the
altitude.
In the equation x2 — ax = a2 let x = y + -^- and

i.i. *
therefore ara — a# = ?/z9 4- ay + —
ft2 — ay a2
^- = y2 a2
j-
4r . 4py2 + Spa2 . . . . . -. , ,
= a2, .'. r = *„ , which is evidently a mim-
p 16 '
mum when «/ = 0, .• . r = -~r- ; but r = ——_ , therefore
* > 16 ' 4a
3â2 pa* -16c , *A * A VT ,
T6~ = "—4 °r $ = an a ==: X A/ —
a #—
a
^ _____ 2 x /w — as before.
2 'V _»
( B9 )
PR0B. (6.) TO FIND THE LEAST PARABOLA WHICH SHALL

CIRCUMSCRIBE A GIVEN CIRCLE. (Fig. 43.)
Since the parabola and the circle touch at P . * . CP is a
normal to the parabola, and Cm is the subnormal = | latus
rectum. Let Cm = z .*. equation to the parabola is
y2 = 2z.x (A.)
w& <y2
Pm2 = r2 — z2 .*. Am =
2z
AD = Am + w*C 4- CD = ——- + z + r= —J"2--
4
Now the area of the parabola EAF = - AD.DE and DjE =
3
/ID v A/
\/2z.AD .-. area jE^LF =%ADx \/2z.AD = ^ + ^
(r + z)3
r. u = = minimum. Let r+2r = y.'.j2r=y —
r . • . —-—j = minimum = u, and . * . y3 — uy + z^r = 0. Let
one of the negative roots of this equation = — a, and there

fore y + a must exactly divide the equation y3 — uy-\~ ru = 0.
y -\- aj y3 — uy + ru ~ 0 \^y2 — ay ~\- a2 — u = 0... (B.)
!/3 + ay*
— ai/2 — uy
— ay2 — a2y
[a2 — u) y + n^
(a2 -r- u) y + a3 — au
_ ___—__— ailc[ therefore we
must have a3 — au = ru .' . u
a + r
Now solving this quadratic (B) we find y = — dtz
v w Z~ > an<^ *n orêr îa* u may become a minimum, we

4
must have u
3a2 a3
.
Sa2
— . or
a
4 ' a 4- r 4 r£ -f r
( 90 )
3a -\- 3r = 4a ,\ a = 3r .'. y = — ~ ~ .' . z = y — r =

3r
Y~r = -2r aKI)-
In the equation y2 •— ay + a2 —- u = 0 or y2 — ay = u — a2
a a2
let y = w + —, and therefore y2 — ay = w2 + aw -{- --.
«z# a2
—- = vr9 a2— = u — aL .'. u = wL + 3a2 . . , is
——, which .
2 4 4 '
3#2
evidently a minimum when iv — 0, . • . u ~ — ; but w -=
—■a?— . • . a3 = •—
3a2 or 4a — 3a -f 3r and a = 3r, and
a -f r a + r 4
therefore y^22
= — = — and z — Jy — r — —
2 as before.
PROB. (7.) THE FOUR EDGES OF A RECTANGULAR PIECE OF

LEAD, a INCHES IN LENGTH AND b INCHES IN BREADTH,
ARE TO BE TURNED UP PERPENDICULARLY SO AS TO FORM
A VESSEL THAT SHALL HOLD THE GREATEST QUANTITY OF
WATER, HOW MUCH OF THE EDGE MUST BE TURNED UP ?
It must be observed that the piece of lead is a rectangular

sheet, and consequently when x = breadth of edge turned
up: then x(a — 2x) (b — 2x) = content of vessel = maxi
mum .'. 4x* — 2 (a + b) x2 + oboe = 4r = maximum, or
x3 a +—b or9 + ab ^Ti
-j- x ■— r = 0. Let x = ?/ + —-— a -\~ b
3 , a + b 9 (a + bf (a + b)»
a + 6 , _ a + b , (a + 6)2 (« + 6)3
~ ~~%~ ^ ~ 2 r 6 y ~ 72
( 91 )
ab ab ab(a -f b)
+ -T*= • + -T* + 24
— r = — r.
'''^ 12 ^~r+ 7g 216 24 *

Now r is a maximum ; and besides r the remaining terms of
the second member of the equation are constant and given
quantities, and consequently the whole of the second member
must be a maximum when r is so, and therefore when we
suppose / = the whole second member, we must have r' =
maximum.
(a + b)2 - Sab a2 - ab + b2
Let n = ^ ^ = ^ .-. if - ny
— r' = 0. Suppose that one of the positive roots of this
equation is = c, and therefore y — c must exactly divide
?/3 ■— ny — r' = 0.
y — cj y3 — ny — ?%l = 0 [j/2 + q/ -J- c2 — w = 0 ... (A.)
V3 - cy*
cy — ny
cy* - cV
(c2 — n) y ~ rf
(c2 — n)y — (c3 — era)
#«t 6.3 _ £ft —; r> anC|
= c2 — n and consequently from equation (A) we

y*f c / c3 4ct'
find */2 + cy + — = 0 and .-. y = - — ± /y ^ .
Now it is evident that when r' or 4rr is maximum, we must
have c3 = 4/ . • . c3 = 4c3 — 4*cn or
/w \/a2 — ab + b2 _ a -j~ b 1
V-8 = 6 and * = y + -3- = g
<[« -f # —■ vV — ab + //)>. We have here taken the nega
( 92 )
tive value of y, because on this supposition only can tbe
equation y3 — ny = — r1 be a maximum.
In tbe equation y2 -\- cy -\ = 0 or if + cy = — let

c c
9
f*£i fy%'119
/yv /)"
—
4 = c .*. —
c = ~4 z2 wbicb is evidentlyJ a max.
r! c2 , , r' c3 — en ^
when z — Q, .'. — = — ; but
c 4 ' c __ c 4
C2 / 71 C
— or 3c2 = 4n . • . c = =!= 2 A/ — and y = o" = ~~
v _ = 6
/« + & — xâFâb + 62"} as before.

and * = 2/ + -g- - g
GIVEN PARABOLA BPAqD. (Fig. 44.)
Let Am — x .' . Pm — 2\/mx and Pq == 4^/mx .- . Pn =

mc = Ac — Am — b — x .*. area rag of tbe required rect
angle = 4(6 — a?) Vmx = max. .*. (6 — x) v # = max .'.
(b — a?)2 x = b2x — 2bx2 + <^3 = r = max. Let a = one
of tbe positive roots of this equation, . • . x — a must exactly
divide b2x — 2bx2 + x3 — r — Q
x3 — ax2
(a - 2b) x2 + Sa#
(« — 2&) ^__ a (« - 2b) x
(a — b)2x — r
(a — by- x — a (a —- by-
a {a — 6> -•
( 93 )
T = (a — by1 . • . from equation (A) we find oc*1 + (« — 26)
and —
y> t fa 2b)
X _j — o. Solving this quadratic we find w = — -—-—-
V —^
4a
and here it is evident that when r =
max. then a(a — 2b)2 = 4r = 4a («■— 6)2 .'. a — 2b ■= dfc
2(a-6).
1st. a — 2$ = 2a —- 26 .\ a — 0 and # = 6.
2nd, a — 2b = 26 — 2a .*. a = —- and a? = -.
o o
By a reference to the annexed diagram, it is evident that
w — - corresponds to max. and oc = b to min.
o

In xl + (a — 2i) a? + — = 0 let a? = y — ^-~—- and
... ^ + (a _ 2b) w = y2 - (a - 26) y + ^^ 4- (a - 26)

„ - (g - 2ft)2 - .2 _ («* - ^)2 __ _r . __ a(a - 26)2
y 2 ~ y 4 ~ a 4
— ay2 which is = max. when y — 0, . • . r = —-—j-~— ; but
r = «(« - 6)2 .-. a (a - 6)2 = °(g ~ 2^ or a - 6 = ±
a- 26 _
—-— and
1st. 2a — 26 — a — 26 . * . a = 0 and oc ■=. b.
46 6
2nd. a — 26 = 26 — 2a . * . a = — and ^ = - as before,
o o
94
PROB. (9.) TO DIVIDE A GIVEN LINE INTO TWO SUCH PARTS

THAT THEIR PRODUCT MULTIPLIED INTO THE DIFFERENCE
OF THEIR SQUARES SHALL BE A MAXIMUM.
Let 2a be the given line and a 4- x and a — x the required

parts. Now by the problem (a2 — x2) x 4<ax = max. ,\ x
(a2 __ ^ __ max# wnich let = r or xs — a2x + r = 0.
Also let 5 = one of the negative roots of this equation.
w + ft J xs - a2a? + r = 0 ^2 - Ja? + ft2 - a2 ...(A.)
x3 + to2
— ft^2 — a2x
— bx2 — b2x
JjF~^dF)x + r
(ft2 - a2) x + 6(62 - a2) ... r = b{b2 - a2)
.' . b2 — a* =: ~j- and from equation (A) x2 — bx + — — 0

r , ft ± /«3 - 4r
or x2 — bx = r and <*? = —- =i= A / —--.— . Now it is
ft 2 V 4ft
evident that when r = max. we must have ft3 = 4r = 4ft
2a and
(ft2 _ a*) ... j2 = 4^2 _ 4fl2 #.# 4a2 _ 3ia ^ ^ ._ _^
V 3
2 VT
7* ft
In the equation x2 — bx — 7- let x = y + — and
ft2 ft2
therefore we find x2 ~-- bx = y2 + by + -^ - by - y =
ft2 V ft3
y% - -4 = - T' and ■. • . r = — — ft?/2 = max. when y = 0,
•. »• = ^ j but r = 6(63 - «a) .-. ^- = 6(/>2 - «2) or i2 =

( 95 )
4{b2 - a2) .-. 3b* = 4a2 and b = -^ .-. x = A = -^

V 3 ^ v3
as before.
PROB. (10.) TO INSCRIBE THE GREATEST ELLIPSE IN A GIVEN

ISOSCELES TRIANGLE. (Fig. 45.)
Let Da = 2x, cb = ?/, ^4D = #, DJB = b. Now by the

property of the Ellipse we have en = -^ = —* . \ an =
ax —
■ 2x2 Zto
_. — «# ._ , —-?
But _BD2 x An2 — Pn2 = -^
?/2
a — x a — x AD1 x2
rrb
(«. x nD) ... ^(^-^) =y"___...nfa._^_.
a? v# — 2# = max. .• . x2 (a — 2x) = «#a — 2<2?3 = max. =
2r or xs — 4- a?2 + r = 0. Let 5 = one of the negative
roots of this equation.

* +JJa*_!*» + r = 0(jBi-(fi + !)a? + a(& + !) = 0,(A.)
a?3 + #a?2
(4 + |) a* + r
db\ r to . ab
bib2 + -~-) .'. -7- = ft2 + -o- an^ hence from equa
tion (A) we find x2 — ( b + -) a? = j- and .*. # = —-—

( 96 )
+ i \f \° + 9) j- when r = max. then bib + ~j = 4r
= 4?/ (& + |) .-. 6 + |- = 46 .-. 6 = |- and 26 = |- .-.

a 4a
_ 25 + a __ "3 a_^~_4a_a „a
* ~ 4 ~ ~4~' ~ T ~ 12 ~~ "8 " *' * ~ 3"'
In the equation «^2 — lb + —J a? = 7- let a? = y +
h + -o- I6 + "o")
_rJL,...-(, + £)._,+(* + ±)r + L_iZ.
- (» + T) » 8_ = r 1" = - T
.*. r =
»(» + t)' — %2 = max. when y = 0, .-. r =
bib ' —^
V ^ 2 / and, 46
A .. = b, + —
a .-. b7 = —
a .-. a? - 2b
—^p-
+ a
= -77- as before.
o
( 97
PR0B. (11.) WITHIN A GIVEN PARABOLA TO INSCRIBE THE

GREATEST PARABOLA, THE VERTEX OF THE LATTER BEING
AT THE BISECTION OF THE BASE OF THE FORMER.
(Fig. 46.)
Let ABC be the given parabola of which the axis BD =
a and 4m the latus rectum are known. Let Br — x .• . rJD—
a — x and mr = y = 2\/mx ,\ the area of the required
i t ^ 2y(a — x) 4*\/ m , N /—
parabola mDr = ~^-^ = —5— {a — x) V x = max.
o o
. *. (a — x) V x or (a — x)2 x = max. Now let (a — a?)2^
= r . ' . #3 — 2ax2 + a2#? — r = 0 ; also let 5 = one of the
positive roots of this equation, and consequently x — b must
exactly divide it.
x-> I) x* - 2ax2 + d2x - r = 0 l^2 + (ft - 2a) a? + O-6)2=0,(A.)
3?3 — 5<^2
(6 — 2a) x2 -f a2#
(J - 2a) x2 - fr(ft - 2a) x
(a — b)2x —• r
{a-byx-hlfl-b)* .-.r=b{a- by
T
or {a — b)2 — ~j~. Now from equation (A) we find x2 +
^ o^ r ^ 6 -2a /5(5„2a)2-4r
{b - 2a) a? = - ^ and a? = — + /\/ ^ -,
and in order that r may be a max. we must have b(b — 2a)2
4itt
= 4r = 4*b (a — b)2 or b — 2a = 2a — 26, or b = — ,\
2a
-
X~
& ~2a - JL - -
"~%~ - 2" ~ 3'
( 98 )
The same may be solved without impossible roots,

r
In the equation x2, + (b — 2a) x — j- let x = y —-
ZL^ ... ^ + (6 __ 2«) co = */2 - (6 - 2a) y + -^-^^
-— by* = max. when y — 0, and . \ r = ——7 ; but r =

b(a - 5)2 .-. 4S (a - 6)2 = J (6 - 2a)2 or 4a2 - Sab + 4fta
4a
= 62 — 4aZ> + 4a2 and 4aZ> = 3Z>2 .-. b = — and a? = —
o
2a
b — 2a 3 a
as before.
2 2 3
PROB. (12.) TO INSCRIBE THE GREATEST CONE WITHIN A

GIVEN SPHERE. (Fig. 47.)
Let ArcB be the required cone inscribed within the sphere
AmcB. Let the diameter Bm of the given sphere = 2a, BD
— x .'. Dm = 2a — x}p = 3.14 &c. Now by the property
of the circle AD* = 2a<2? - x2 .'. 4^D2= 4(2a^ - #2) .-. the
area of the base Arc of the required cone = ■£- 4(2a<2? - a?2)
P x (2ax — <z>2)
— p (2ax — x2) .-. content of the cone == ~
= 1L (2ax2 — ^3) = max. .*. 2ax2 — x6 = max. which let

3
— r . • . $ _ 2a<9?2 + r = 0 ; also let & = one of the negative
values of this equation, and consequently x + b must exactly
divide it.
( 99 )
x + 6 J w3 -2aa?'i + r = 0 (j*?2--(6 + 2a)<2? + 6(6 + 2a) = 0, (A.)
x3 + bx*
- (b + 2a) oc% + r
- (6 -f- 2a)x2-b(b + 2a) oc
6(6 + 2a) x + r
b(b + 2a) x + 62(6 + 2a) ,\r = 62
(6 + 2a) or 6(6 + 2a) = -r-. Now from equation (A)

y 6 + 2a
we find ^?2 — (b -\- 2a) x — — -y- or # = =±=
a/ _i ^ ^ and in order that r or 4r may become
a max. we must have 4r = 6(6 -f 2a)2 or 462 (6 + 2a) =

6(6 + 2a)2 .-. 46 = 6 + 2a or 6 = ^ .-. x = 5jL?? =
-Q- + 2a
3 4a
T%e same solved without impossible roots.

T
In the equation x2 — (6 + 2a) a? = ~- let oc = «/ +
—-—
2 and therefore we find by
J substitution
a?2- (6 + 2a) a? = if + (6 4- 2a) y + ^ "*42^2 - (6 + 2a)
(6 + 2a)2 __ 9 (6 + 2a)2 _____ _r _ 6(6 -)- 2a)2
y 2 ~r 4 "" 6;,',r~ 4
— by2 = max. when y = 0 . • . r = —-— ; but r = 62
(6 + 2a) and .-. 62 (6 + 2a) = *(L+i__! .-. 6 = ^ and

4 o
6 -h 2a 4a , „
# = ——— — -— as before.
100 )
PROB. (13.) GIVEN THE SURFACE OF A CYLINDER TO FIND

ITS FORM, THAT ITS VOLUME MAY BE A MAXIMUM.
Let the whole surface of the cylinder = s and x = diame

ter of its base. Now it is evident that the areas of the two
*DX
opposite circles of the cylinder = ^— where p = 3.14 &c,
the circumference of the base = px, and the convex surface
TJX2 2s —- TJX
= s — ~- = q~" which divided by px} the circum-
O q __ 7)00*
ference of the base, gives the altitude = — . Now
TJX
multiplying this value of the altitude into —-, the area of the
base, we find the content of the cylinder = —- x —-

2SX — px3 -i « * 2s
= 8k—— = max. and . * . 2sx -— rpx3 = max. or —
p x —
x3 = max. which let = r. . • . x3 2s x -f- r = 0. Now let
P
a = one of the negative roots of this equation and conse
quently w + a must exactly divide it.
x -{■ a J\ x2, 2s x + r = 0 [ x2 — ax -\~ a2 2s = 0... (A.)
p ^ P
xz + ax2
9 2s
— ax* x
P
— ax2 ■— a2x
(a2 )x + r
2*x
. r = a(a2
( ioi )
T 2s
or —
a = a2 p. Now from equation
i ■ (A)
v / x2 — ax = -—
—
a or x = —
2 =t AV/ —4a '. and in order that r mayJ be a
(2s\
a2 j and . • . a =
—7==- and a? = — = A/ — . Writing this value of x in

2$ —■■ 79X
the equation altitude = —-—— we shall find altitude =
2px
V2s
— and hence it appears that altitude = the diameter
of the base.
In the equation x2—ax= let ^?=v+ tt and therefore

2 a
2 9 , a2 a2 2 a* r
x2 - ax = y2 + ay + -^ - ay - — = y2 - -g = - —
'"• r = T "~ ay* = max# wnen ^ == fy •"• r = ~J > û^ r ==
«(«* _ 21) and ^. I = «(«3 - j) and a = 2 /\/| .-.

a A /2s
x=y=
2 = VS
"VWp as before
PROB. (14.) TO PROVE THAT THE ALTITUDE OF THE GREAT

EST CYLINDER WHICH CAN BE INSCRIBED IN A GIVEN
SPHERE, IS EQUAL TO 2r \/ \ \ T BEING THE RADIUS.
(Fig. 48.)
Let the altitude mn of the cylinder required = 2x, and r
being the centre of the sphere rn = x .• . Bn = vr2 — x2
= the radius of the base of the cylinder, and . • . the area of
( 102 )
the base — p (r2 — x2) where p = 3.14, &c. Now since
altitude of the cylinder = 2x, its contents must be = 2p
(r2x — x3) = max. which let = 2pq, . * . x3 — r2x + q = 0.
Let one of the negative values of this equation = a, and
consequently x + a must exactly divide it.
go + a J ^3 — r2x + g = 0 1^<2?2 — «<# + a2 — r2 = 0, . . . (A.)
x3 + ax2
ax°
(a2 — r2) # + g
(a2 — r2) x + &3 — «^2
—— .•. q = a* — ar-*
.•. — = a2 — r2. Now from equation (A) we find a?2 — ax
Ci CL / Ct3 —~~ 4(7

= — — .-. x = —- =fc A/ —7—-, and in order that q may
be the greatest possible, a3 must be = Aq = 4a3 — 4#r2 .*.
a2 = 4a2 — 4r2, .-. a = 2r /y-^- -"• # = -«- — r /y "a" •'•
2x = altitude required = 2r A / — .
T%e same solved without impossible roots.

o a
In the equation <#2 — a<#> = let x = ?/ -f — ,\ a?2 —
a2 - «y --_
a^ = r + ay + _ «2 = r9 --—■=--—
a2 # or g
= — ay2 = max. when y ~ Q, ,\ g = — ; but q -j= a3 —
«r"9 and-, .*. —

&3 = a3o — ar*2 r . a — , 2r .•.# = —
a = ———
r =r
4 a/3 2 v/3
v — as before.
( 103 )
PROB. (15.) A CANDLE STANDS ON A HORIZONTAL TABLE

DIRECTLY OVER A POINT, AT A GIVEN DISTANCE FROM A
SMALL OBJECT ON THE TABLE ; WHAT OUGHT TO BE THE
HEIGHT OF THE FLAME WHEN THE OBJECT IS ILLUMI
NATED THE MOST POSSIBLE ? (Fig. 49.)
Let A be the object on the table, B the point under the

candle, and C the flame, considered as condensed at a point.
The intensity of the illumination on the object A depends on
its distance from C, and on the angle which the rays make
with the surface (supposed to be horizontal) . By the princi
ples of Optics, the intensity at different distances, the angle
of obliquity being the same, will be inversely as the square of
the distance ; with different degrees of obliquity, the distance
being the same, as the sine of the angle which the rays make
with the surface. Therefore the intensity, as depending on
both obliquity and distance, will be expressed by -J™ sin.
EC
CAB = ACS
~~. But aÂB, n= sin. CAB, then the illu-
mmatmg power on the surface at A — -~r^ x -j™ X -7™
= sin. n COS
—'■?—7b = max. .'. sin. n cos.2 n = sin. n (1 — sin.2?z)
a2,
= sin. n — sin.3 n = max. = r. Now let sin. n = oo, . * . x — ocs
= r, .*. x3 — x + r = 0. By problem (1) when r = max.
1 1
then os = , .*. sin. n = —7=. By the trigonometrical
tables n = 35° 16' : this Sgives BC = AB x a/2

—==. = AB x
7
•71 nearly; so that the height of the flame must be about —
of the distance AB.
The same may be solved without impossible roots as in
problem (1).
( 104 )
PROB. (16.) TO DIVIDE 12 INTO TWO PARTS, SO THAT THE

LESSER MULTIPLIED BY THE SQUARE OE THE GREATER
SHALL BE A MAXIMUM.
Let x = greater part .'. 12 — x = lesser part. Now it

is required to find such a value for x that (12 — x) x2, or
12<2?2 — x3 may be a maximum. Let 12<#2 — x3 = r .'.
x3 — 12#2 + r = 0. Suppose that a = a negative root of
this equation, and consequently x + a must exactly divide it.
x + aj x3-12x2+r = 0 Ix2-{a + l2)x+a{a + l2)=0, (A.)
^3 + «^2
- (a + 12) a;2 + r
- {a + 12) x2 -a(a + 12) a:
a (a + 12) # + r
a (a + 12) x + a2 (a + 12)
T = a (a + 12).
.•. r = a2 (a + 12) .*. — Now from equa-
tion (A) we find x1 —■ (a + 12) x = — — or a? = ——— dfc

65 <o
v ^ — y and in order that r or 4r may be a

max. we must have a (a + 12)2 = 4r = 4#2 (a 4- 12) or a
. . a + 12 Q
= 4 ana <# = —-—- = o.
T%e s#me m#y be solved without impossible roots.

T
In the equation x2 — (a + 12) a? = let x = y +
a~^} and .-. x*- {a + 12) a? = y2+ {a +12) y + (^J_i^!
- ,{a +, lox
12) y - ^(fl+^—L
12)2 = f, --*—^--^
(a + 12)2 = - —
r .-. r =
a(«
_j—+ 12)2
^ ^9 — max. when
t. y — 0,
a andi . • . r = « (« + 12)2 ;
( 105 )
but r = cfi (a + 12) .'. ^-±i^ = a2 (a + 12) .-. a - 4,
and a? = ——— = 8 as before.
x3 3x2
FROB. (17.) WHAT ARE THE VALUES OF X WHEN — g"
-f 2x BECOMES MAXIMUM OR MINIMUM?
Multiply this expression by 3, and let the product = r, . ' .

x» — x2 + 6x — r = 0, also let a = one of the roots of
this equation.
%-a) x*-^x2 + 6x-r=0 ^2+(«-|)^+«2-~ + 6=0,(A)
a?3 — ax2
( 9\
(a — -) x2 -j-6x
(a - |) ^ - a(a - ?) a?
(a2 — — -f- 6) <2? — r
{a2 - y + 6) # - a(a2 - ^ + 6)
,\ r = a(a2 ~ -f 6) .*. — = a2 — + 6. Now from

2 a 2
(9\ y 2a - 9
a — -) a? = - or ^ = —
9\
(a — ~j X a — 4r
— and in order that r or 4r may be a
/y 4a
max. we must have a( a — -J = 4r = 4a (a2 — + 6)
( 106 )
or a2 - 9a + ^ = 4aa - 18a + 24 or 3a2 - 9a = - ^

4 4
5 3 5
o\. a2--3a = \ a = dbl = or land 00 = —
~ 4 *' 2 2
2a - 9 4 2a - 9 —
= 1 for maximum ; 5? : =
4 4 " "-"IT
1 - 9 8
= 2 for min.
2 ~~ 4

(9 \ ?*
a —J # = \et x = y —
9 / 9\2
——.-.x*+{a--)x=f-{a--)y + 4
9\! / 9'
, 9n (a"|-)2
2") ,
, (a"D
(a ~ 2.
+ («- -a) y —=r
g— — = --
—
9
°(c-t)' . , „ °(a-T)"
r = ««/2= max. when y = 0, . • . r =
4 » — * ' 4
/ 9\a
but r = a(a2 - ^ + 6) .-. .—^1 = «(a2 - ^ + 6)

£ 4 2
or a2 — 3a = — — and a = -—• =fc 1 = — or — and x =
4 2 2 2
—-— = 2 or 1 as betore.
( 107 )
PROB. (18.) WHAT NUMBER IS THAT FROM THE CUBE OF

WHICH ITS SQUARE AND TWENTY-ONE TIMES ITSELF
BEING SUBTRACTED, THE REMAINDER IS THE GREATEST
POSSIBLE ?
Let x = number required ; then according to the question

x3 — x2 — 21 x = max. = r . • . x3 — x2 — 21 x — r = 0.
Also suppose a — one of the roots of this equation.
x-aj x3-x*-2lx-r=:0 [_x*+ (a-l)# + a2-a-21 = 0;(A)
x3 — ##2
" (a - 1) a;2 - 21a?
(a — 1) a;2 — a (a —■ 1) a?
(a2 — a — 21) x — r
(a2 - a - 21) a? - «(a2 - « -21)
a(a2 — a — 21) ,\ — = a2 — a — 21 .'. from equation (A)

r __ a— 1 i*A/!/&(« — I)2— 4r
a;2 + (a — 1) x = — - or a? = rt — , .,
Now in order that r or 4r may become a maximum we must
2 a =
have a(a — l)2 = 4r = 4a (a2 — a — 21) or a2 ■— —
o
— and .•.«=— 5, . • . a? = — = 3.
o <^
In the equation a?2 -f (a — 1) # = let a? = ?/ ——
and .*. a?2 + (a— l)a? = 2/2— (a — 1) y + -——— + {a —\)y

(a ~ I)2 2 (« - 1)2 r ^ __ a(fl - 1)2
2 * 4 a * - 4
«y2 = max. when y = 0, and . • . r = —L__L • DUfc r
( 108 )
a(a2- a -21) .\ ^LZL.i}! -a (^ __ a _ 2l) .\ a2 ~ -|-a
85 15 „ . a- 1 -5-1
= _ or a = - — = - 5, and x ~ —= ——
= 3 as before.
PROB. (19.) TO CUT THE GREATEST ELLIPSE FROM A

GIVEN CONE. (Kg. 50.)
Let ABB be the cone, PB the elliptic section, AC= a, Cn=z
x, major axis = 2m = PB, BC = b, nP— y, minor axis = 2n
= to. Now the area of the Ellipse = wmn (see the Integral
Calculus, or my treatise called an Insight into the Nature of
the Integral Calculus). It is evident that PB : IB : : PQ : El
or 2 : 1 : : PQ : El, and BP : PZ :: BD :IF or 2 : 1 : : BD :IF .\
PQ = 2.EZ, and PD = 2/Por PQ x BD = 4E/x^=4W =
ro2=4w2 .-. 2n — \/PQ xBD = V2x x2b = 2\/Tx, and 2m =
a/P^2 + P^2 = a/ (b + ^)2 + P^2 but Pn= CA x ^ =
~L—^—- .• . 2m ■=. a/ (b + x)2 + -jo {b — x)2 and .-. area =
irnm = ^-~— A/ (i + x)2 -f -Tg (* — ^)2 = max. and .'.

a2 a2 + £2
a? {b + ^)2 + -Tg os (b — x)2 = max. and therefore —^—
oft 2(a2
i— — b2)L ofi _j_ (#2 _|_ j2j ^ __ max# Dividing this
a2 + b2
expression by the constant quantity —^— we have x3 —
—V———' x2 + b2x = max. To shorten the calculation let

a2 + b2
2b{a2
—^ - r—^-b2) = n .\x3 — qxl2 + gt^ = max. = r, . • . x332^
— tfar +
a2 + b2 * *
#2^ _ r _. q^ Now suppose v = one of the roots of this
equation.
( 109 )
w—vj x3—qx2 + b2x—r~0 [^x2+(v — q)x + v2-vq + b2=Q(A)
x3—vx2
(v — q) x2 + b2x
(v — q) x2 — v(v — q) x
(v2 — vq + b2) x — r
(#2 — #£ + 62) x — v(v2 — vq + b2)
T
,\ r = v(t;2 — vg 4- 62) and — = v2 — vq -\- b2. Now from
equation (A) we find x2 + (v — g) # = T and . • . x = —
- ± A/ ~ V- 3 and in order that r or 4r

2 V 4v
may become = max. we must have v(v — q)2 = 4r =
2 <72 — 462
4# (v2 — vq + b2) or v2 — — qv = 5 .\ v =
o o
g d= V^g2 — 1262 _ v — q__q — v__q v
3 ~> and * ~ 2~ " ~¥~ ~ ¥ ~~ "2
- 3ff _ y — v/^::'T2F _ 2g dz v7^^"!^2 __

""2x3 2x3"" 2x3
q d= iv^g2 - 1262 g d= Vq2 - 362 2%^ - 62)
3 ~ 3~~~' bUt q ~ a2 + b2
2b (a2 - b2) ±
. • . * = —i 8 b\/a*
+ - \â2b2 + 64 . _, . problem
This ,. .
is
possible so long as the altitude a and base 2b are such as

make a4 — 14a262 + #4 a positive quantity. The limit of
possibility is when the radical disappears ; then we have the
following equation a4 — 14W -f- 4964 = 4864 .*. a2 = 7b2 db
v ; 3 8±4a/3 3 2+^/3
( no )

T
In the equation x2 + (v — q) x = let x = y
(v - of
.'. ^2 + (v - g) a? = y1 - {v - g) y + —~- + (t> - g) y -
(v ~ QY 9 (v - 9Y r v(v-qf 2
2^4 v 4 *
max. when y = 0, .•. —— = r = i>(ir — vg + oj).
#=h\/4tf2 — 12&2
Prom this equation as before we may find v~-
2b(a2 - V) -±z h\/a* - 14a2i2 + 64 , ,
and hence a? = — —-—0/ 9 , ,9N as before.
3 (a1 + b*)
PROB, (20.) THE CORNER OF A LEAF IS TURNED BACK, SO

AS JUST TO REACH THE OTHER EDGE; FIND WHEN THE
LENGTH OF THE CREASE IS A MINIMUM. (Fig. 51.)
The full leaf is mnAB, and when its corner A is turned

back and touches the other edge mB of the page at the point
«, the triangular piece QPA of the leaf falls upon its remain
ing piece mBPQn, and each of the angles QaP and QAP is
= 90°, and consequently the figure QaPA may be inscribed
in a circle.
It is also evident that aP=^PA and aQ~AQ and by the
property of the circle aA x PQ=2AQ x AP (1.)
Now let PA = x and AB—a .'.by Prop. 12, 2nd book of
Euclid Aa? = aP2 + AP2 + 2BP x PA = 2x2 + 2(a - x)
oc = 2a?2 4- 2ax — 2a?2 — 2ax .• . Aa = \/2oi. Now AQ* =
QP2 - AP2 .\ from equation (1) «^2 x PQ2 = 4.4 Q2 x ^P2
= 4^4P2 x PQ2 - 4>AP* .-. 4^4P4 = (4^JP2 - ^2) PQ2 .'.
4** = (4*2 - 2«r) PQ .-. PQ - —^ = min. .-. *^
( 111 )
= max. Let 2oc — a — y>

y, . * . oc = -—■—
2 and 2a?3 = 4.
2a? — a 4w
: -—f-~- = max. '. . f .3. Now let y =
(y + «)3
ab
ab #•
2/
^ —
c be*
_____ \f
1 ——■ VVIQjX
c ' ' {y + «)3 ~ «3(# + <03 " {c + 6)3 «2
1 is a constant given quantity, . • . -,

—2 6c2 ,.3 = max. It is
evident that 7—--- = —-—= x 7——-ro = (1 r-r)

(c + 6j3 c + # (c -f 6)2 \ c -\- W
°2 XT 1 i. c l-x c \ &
x 7—TTTa-
(c + £)2
Now let ——-7 = £, . * . ( 1
c + b V
-—7 x 7——-r-2
c + W (c + &)2
2
= (1 — #) ^2 _= £2 — ^3 = max. .\ by Prob. 2nd, z = — =
c
rn •'• T3 = —
c + 6
= „J + t6 or t6 = t1 •"• y = T
06
_j_ ^
a , w + a 2 3a
_, _ and .= —-.=: _-^— =-.
The same may easily be solved without impossible roots.
PROB. (21.) TO FIND THE POSITION OF THE PLANET VeUUS

IN RESPECT OF THE EARTH, WHEN HER LIGHT IS THE
GREATEST. (Fig. 52.)
The planet does not appear brightest when her disc is per
fectly round ; she is then too remote to produce that effect ;
and besides, she is seen in the direction of the sun. In her
inferior conjunction her crescent is too narrow, almost the
whole illuminated part being turned towards the sun. It is
therefore in some intermediate position, which is to be deter
mined, that she is brightest. Let S be the Sun, E the Earth,
( 112 )
and ABCD Venus, ABD its illuminated hemisphere, which is
turned towards the Sun, and CBD its hemisphere towards
the Earth : produce SVto F.
The portion of the illuminated surface towards the Earth is
contained between two planes DV} BV, perpendicular to the
plane EVS ; and this surface will manifestly be projected
into a crescent, the breadth of which is the versed sine of the
angle BVD, which is equal to EVF, because if BVE be added
to both, each is a right angle.
Now the area of the crescent is always as its breadth;
therefore, the whole disc being taken as a unit, the illumi
nated part will be expressed by the versed sine of the angle
EVF, or by 1 -f cos. EVS. Again the brightness of the
planet is inversely as the square of the distance, therefore the
brightness depending on its position, in respect of the Sun
and its distance from the earth jointly, will be proportional
to ipTfl • -^ a = -®^ *ne distance of the Earth
from the Sun, b = VS the distance of Venus from the Sun,
x = VE, the distance of Venus from the Earth. Then
x% + b2 — a2
cos. EVS = —; . and therefore the brightness
2ox
, , 1 + cos. EVS x2 + 2bx + b2 - a2
of the planet = ^ = ^ =
x2 + 2bx + b2 — a2 . . , 1 , 9 ,
max. or ; = max. which let = r, . * . x-1 +
x3
2bx + b2 — a2 = rx3 or rx3 — x2 — 2bx + a2 — b2 = 0. Now
let x = — .-. ^ - 4 ~ — + a2 - b2 = 0, and .-. {a2 - b2)
y y3 2/2 y
y3 — 2by2 — y + r = 0, and dividing this equation by a2 — b2
we find t - -^j.y* - ^rs>y + ^=rj, = o.

Now since r = max. and -g j-2 = constant quantity, .*.
( 113 )
26
tz = max. which let = v ; also let -a—75 = m an(i
— b2 a2 — ¥
n •a)
y3 — my2 — ny -f 0 = 0. Suppose that c = one of the
negative roots of this equation, and consequently y + c must
exactly divide the said equation
y + cj y3—my2~ny + v~0 \^y2—(c + m)y + c2-\-mc--n=0(A)
y3 + q/2
— (c + m) y2 — ny
— (c + m) y2 — c{c -h m)y
(c2 -f- cm — n) y + v
(c2 + cm — n) y + c(c2 + cm — n)
— »(<&
c(c2 + cm — n)j .-. — = c2 + cm — w, and from
7J C ~f" m
equation (A) we have y2 — (c + m) y = or y == —5—
v C (c ~4~ TYl) 2 —— 4lJ

J± J. m Now in order that 4v or v may be=
max. we must have c(c + m)2 = 4v = 4c (c2 -{- cm — n) or

Omi 9% -4- 4%
c2 -f 2cm + m2 = 4c2 4- 4cm — 4rc .\ c2 + — c = 5
o o
m , /4m2 + 12% 2\/m2 + 3n — m,
orc=-T + Y 9 = § • ^T
N™
c+m A/m2+ 3% + m , ,, . /1W , .
y — —^^— -= .^ ^ and from equation (1) taking
the values of m and ft we find \/m2H-3ft= A/ —, 2-^fl^g—

V/3or+~^2 V?TS + m \/%a2 + 62 + 26
~~ tf-W- ' ' 3 3(«2-62) ~
VSa2 + b% + 2b 1 and • ■! - * =
3a2 + 62 - 462 ~~ \/3«2 + 62 -26 2/
V8«? + & - 2b.
( 114 )
In numbers a = 10,000, b = 7,233, therefore so = 4,304,
the angles ^=39° 43' 30", F = 117° 55' 20", £=22° 21' 10".
— (From the 7th edition of the Encyelopsedia Britannica.)
In the equation ^ — (c + m)y ~ let y = # H —

(c -f- m\ 2
■•" • 2/2 — (c + m) 2/ = ^2 + (c + m) z + -—i—" (c + m)
4
(c -f- m)2 „ (c + m)2 v • c(c + mY
z _ i— -—L. = z2 — .—'- = or v == ——r—-
2 4 c 4
— c^2, which is evidently a maximum when z = 0, . " . v =
c(c + m)2 c(c + m)2
——-—'—. 4 But v = c (c2
v + cm — n)J .-. ——-—- 4 =
/ a +. cm — n)\ . • . c = 2\/m2 + 3^ — m and_ therefore
c (<r . r ?/ =
o
c + m \/m% -f 3^2 + m , p
—_— = as before.
PROB. (22.) REQUIRED TO DETERMINE WHAT MUST BE THE

DIAMETER OF A WATER-WHEEL, SO AS TO RECEIVE THE
GREATEST EFFECT FROM A STREAM OF WATER OF 12 FEET
FALL. (Fig. 53.)
In the case of an undershot-wheel put the height of the
water AB =12 feet = a and the radius BC or CD of the
wheel = <#, the water falling perpendicularly on the extremity
of the radius CD at D. Then AC = a — w, and the velocity
due to this height, or with which the water strikes the wheel
at D will be as \/a-—w, because the squares of times or velo
cities are as the spaces, and consequently velocities are as the
square roots of spaces, and therefore the effect on the wheel,
being as the velocity and as the length of the lever CD, will
be denoted by w\/a — os or \/aoc% — #3, which therefore must
( 115 )
be a max. or its square ax2 — xs = max. Let ax2 — x'6 =
r or x* ■— g^2 + r = 0 ; also let b = a negative root of this
equation .*.#+# must exactly divide it.
x + b) x3— ax2 + r = 0 (jz?2 — (6 + a)x + 62-i- ai = 0, ...(A.)
— (6 + a) <2?2 + f
— (b + a) x2 — b(b + a) x
(b2 -f ab) x + r
{b2 + ab)x + b{b2 + ah)
.-. r = b {b2 + ab)
.\ -j- = b2 + ab .'. from equation (A) we find x2 — (b -f a)
#= f ... x ~ —
* +—a ^
_,_ a/
A /b{b
— + a)2
-^ — 4r which,
, . , when,
b 2 V 46
r or 4r = max. must give b{b + a)2 = 4>r = 46 (62 -J- «6) =
462 (a + 6) .-. 6 + a = 46 and i = -£ •'• # = —^ =
— ; but 05 = 12, .\ a? = —- — = 8 feet radius.

o o
But if the water be considered as conducted so as to
strike on the bottom of the wheel, as in the annexed figure
(Pig. 54), it will then strike the wheel with its greatest velo
city, and there can be no limit to the size of the wheel, since
the greater the radius or lever BC, the greater will be the
effect.— (From the 3rd vol. of the old edition of Hutton's
Course of Mathematics.) In the case of an overshot- wheel
a — 2x will be the fall of water, va — 2x as the velocity, and
x\/a — 2x or Vax2 —■ 2xs the effect, then ax2 — 2xz is a
maximum. Here instead of x we must put down 2x .'. 2x
= — ,\ x = — = 4, the radius of the wheel.
6 6
But all these calculations are to be considered as inde
pendent of the resistance of the wheel, and of the weight of
the water in the buckets of it.
( 116 )
In the equation x2 — (b + a) x = — .-=-, let x = —-— + y

o £
... ^ _ (6 + a) x - y% + (p + a) y + (6 + ^ - {b + a)
(b + a)2 , (b + af r bib + «)2
y 2 y 4 b 4.
— %2 == max. when 2/ = 0 .". r = ——-r——. But r =

6(62 + a*) .-. bJL+j¥i = j(ja + ab) and 46 = ^ + a..-.
, a n b +a 2a . x __ 12 x 2
0 = - and a? = —-— = — ; but a = 12 .' . x = —5—
= 8 feet as before.
PROB. (23.) TO DETERMINE THE STRONGEST ANGLE OF POSI

TION OF A PAIR OF GATES FOR THE LOCK ON A CANAL OR
RIVER. (Fig. 55.)
Let AC, BC be the two gates,, meeting in the angle C,
projecting out against the pressure of the water, AB being
the breadth of the canal or river. Now the pressure of
water on a gate AC, is as the quantity or as the extent or
length of it, AC, and the mechanical effect of that pres
sure, is as the length of lever to half AC, or to A C itself.
On both these accounts then the pressure is as AC2. There
fore the resistance or the strength of the gate must be as
the reciprocal of AC2. Now produce AC to meet BD, per
pendicular to it, in D ; and draw CE to bisect AB perpendi
cularly at E; then by similar triangles AC : AE : : AB I AD;
where, AE and AB being given lengths, AD is reciprocally
as AC, or AD2, reciprocally as AC2; that is, AD2 is as the
resistance of the gate AC. But the resistance of A C is in
creased by the pressure of the other gate in the direction
( 117 )
BC. Now the force in BC is resolved into the two BD,
DC; the latter of which, DC, being parallel to AC, has no
effect upon it, but the former, BD, acts perpendicularly on
it. Therefore the whole effective strength or resistance of
the gate is as the product AD* x BD. If now there be put
AB = a, and BD = x, then AD* = AB* - BD* = a* -
x* ; consequently AD* x BD = (a* — x*) x x = a*x — x3
for the resistance of either gate t and if we would have
this to be the greatest, or the resistance a maximum, we must
find such a value of x which will make a*x — a? = max.
= r. Let b = one of the negative roots of this equation,
and consequently oo + b must divide it exactly.
x + bj x3 - a*x + r = 0 [_x* - boo + b* - a2 = 0, ... (A.)
#3 + bx*
— &2?2 — &2<2?
— bx* — #2<2?
(#2 — a*) x + r
(b*-a*)x + b{b2-a*) .-. r = b{b* - a*)
= b* — a* .• . from equation (A) we find x* — bx = —
*«- , • . x = -^ ± A// —-jt—•

b3 4/* Now in order that r or 4r may
be = max. we must have b3 = 4r = 4b (b* — a2) or b =
-r= and # = — = —7= = a \f — = "57735a, the natu-
ral sine of 35° l& ; that is, the strongest position for the lock
gates is when they make the angle A or B = 35° 16'; or
the complemental angle ACE or JBCE = 54° 44/, or the
whole salient angle ACB = 109° 28'.— (Prom Hutton's
Fluxions.)
( 118 )

In the equation x2, — boo = t let ^ = ^ + —
j- b and . • .
#2 yi £2 r
^ _ to = y« + Jy + — - Jy _ _ = ^ - — = - —
b3 bs
.•. r = — %2 = max. when y = 0 . \ r = -^- ; but r =
J(J2 _ a2) . &? _ j/j2 _ a«v and . j = 2a or ^ _ | =
-7= as before.
V 3
PROB. (24.) IT IS REQUIRED TO DETERMINE THE SIZE OF A

CUBICAL SOLID, WHICH BEING LET FALL INTO A CONICAL
VESSEL FULL OF WATER SHALL EXPEL THE MOST WATER
POSSIBLE, FROM THE VESSEL; ITS DEPTH BEING = a AND
DIAMETER OF THE MOUTH = 2b. (Fig. 56.)
Let ABC he the given vessel, the diameter of its mouth

= 2b and its depth HC = a. EmnD = the required cube.
Let FC — oo. Now by similar triangles we find HC I AH : :
FC : JSJPor a : b :: x : EFov EF = — and .\ED = 2EF=
a
, and consequently the area of the base of the required
cube = I J = —2~- which being multiplied by HF

(= HC — FC — a — oo = the height of the immersed part
4&2#a
of the cube) the product = —^- (a — oo) = the solid con
tent of the immersed part of the cube = quantity of water
452
displaced. Now since —$■ is a constant quantity, therefore
x% {a — x) = ax1 — x3 = max, = r .\ a?3 — a#2 + r = 0.

( 119 )
Let c = one of the negative roots of this equation, conse
quently x + c must exactly divide it.
x + c) x* — ax2 + r = 0 \jxP — (c + a) x + c2 + ca... (A.)
a?3 + ex2
— (a + c) x2 + r
— (a + c) x2 — c(c + a) #
c(c -f a) <2? + y
c(c + «) # + c(c2 -f ca)
.«. r = c(c2 4- ca) .'. — = c2 + ca. Now from equa-
tion (A) we have x2 — (a + c) x = .'. a; = —— =fc

c <4
/c_(a H~ c)2 — 4r
, and in order that r or Ar may become
a max. we must have c(c + a)2 = 4r = 4 c(c2 + ca) = 4c2
/ . \ . * j a c + « 2« -,
(c + a) .*. c + a = 4c and c = — .\ # = —;r— = -5- and
o <o o
consequently one of the equal sides of the required cube =
2b x —
^^ 2bx 3 Aba

y c -\- a
In the equation x2 — (c + a) x = let # = y -\ —
.*. re2 - {c + a) a? = y2 + (c + a) 2/ + v(c +4 a)2

; - (c + a)
(c + a)2 9 (c + a)2 r c{c + a)2

y 2 ~~V Ac A
c#2 = max. when y = 0 .*. -^—7—- = r = c(c2 + ca) .•.
a
c + a = 4c or c = -—•.-.# = c +a = —
2a as ,before.
e
( 120 )
PROB. (25.) IT IS REQUIRED TO DETERMINE THE SIZE OF A

BALL, WHICH, BEING LET FALL INTO A CONICAL VESSEL
FULL OF WATER, SHALL EXPEL THE MOST WATER POSSIBLE
FROM THE VESSEL; ITS DEPTH BEING 6 AND DIAMETER 5
inches. (Fig. 57.)
Let ABC represent the cone of the vessel, and DHE the
ball, touching the sides in the points D and jEJ, the centre of
the ball being at some point F in the axis of the cone. Put
AG = GB = 2\ = a, GC = 6 = b .-. AC = VAG* + GC2
= 6i = c, DF=zFE=:FH = x the radius of the ball. The
two triangles ACG and DCF are equiangular; therefore
AG lACy.DF: FC; that is a : c :: x \ — = FC; hence GF
1 a
= GC - FC = b - — and GH = GF + FH = b + x -
a
ex
— = height of the segment immersed in the water. Then
(by Hutton's and other authors' works on Geometry,—see
Introduction,) the content of the immersed segment will be
(2ex\
6# _ 2# - %b + —)
oc + b J x *5236 = maximum, and therefore

(CP\ / cx\
2x — b H J lx + b j = max.; but 2x — b +
ex ■= 2a + c x — b7 and, a? +, b7
—. ex = a — c x +, b, = 6, —
a a a a
c —
-a? where c is greater than a, because c is the hypothe-
nuse and a the perpendicular of a right-angled triangle. Let
7 c — « (b •— y) a j 2a -\- e
b a x = yu .*. x = c ———
a and consequently
^ J a
* _ 7 - (h — V) a @a + g) _ t __ 3a2Z> — g(2« + c) y __
~ a(c — a) ~~ «(c — a)
( 121 )
Sab — (2a + c) y # /2a + c ,\ /a — c + ,)--
,.(-±_'._,)(«z-.
c — a
^2a 4- g _ ^ ^ __ c - a\* __ Sabif - {2a + c) if
V a / \ a<#
aw /1 c— a
2a + c / _ Sab yt9 ~. y3\
,\ = max. XT
JN ow as 2a + c = a con-
c — a \2a + cy u I c —a
stant quantity, we must also have ^—— y* — y3 = niax.
= r ; also let = A, and .• . y% — Ay21 + r = 0. Let

,â -f- c
ft = one of the negative roots of this equation ;
y + nj ys—Ay2+r=:0 [_y* — (n + A) y + n(n+A)=:0, (B.)
y*+ny%
— (n -\- A) y2 + r
— (n + A) y2, — n(n + A) y
n(n + A) y -f- r
ft(ft + -4) ?/ + ^2 (ft + -4)
(ft + A) and - = ft (ft + -4) .*. from equation (B) we have y2

it
i , as
-(M+^)y r
= --ory = n-y-±A/A-^
+ -^ _. A /n(n+A)2—4<r and,
hence it is evident that 4r cannot be greater than ft (ft + -^)2

and therefore when it is a max. we must have ft (ft + A)2~ 4r
= 4ft2 (ft '+ -4) and .\ ft + A = 4ft or ft = — ; and hence

o
ft + A 2-4 2 x Sab 2ab
y = and a? =
2 3 3 (2a 2a + c
2a5
(»- -)x«a
(b- so
c — a ~~
a 2a +
c — a
abc ■ ^
~~ (c — a) (2a + c) ~ ui ¥lp
the radius of the ball; consequently its diameter is 4ii inches,,
as required.
( 122 )
The same solved without impossible y*oots.
In the equation y* —■ (n + A) y = — let ^ = z -f
— (w + A) y = ^2 + (w + î) 2r +
4
— (n + A) z — (w + ^)2 = 2r29 — ^(w + —-
v—-—L ^)2 = r .'. r
v ; 2 4 ^
w(n + AY , A w(» + -4)2
= __^— —-^ ^ — max. when # = 0 .'. r = r ,
4 4
o / o / ^ ^(^ + A)2 A
but r = n* (n + ^) .*. n2 (n + A) = -*—-—- .'. » = -3 •
2^\
-vVa (ft-T)
n + A %A „„* „ _ (6 - y)
Also ?/ = —~— = -tr- and a? =
* 2 3 c — a c — a
, , , Sah ahc . r
but A = .\ # = 7 r-7s—;—\ as hetore.
2« + c (c — a) (2a + c)
PROB. (26.) TO FIND SUCH A VALUE OF X AS SHALL

(x - I)2
MAKE 7- ~v5 A MAXIMUM.
(# + 1)
Let x + 1 = — .-. (x + I)3 = -1, a? - 1 = 2 =-
———- and (x — l)2 = 5-— and therefore we find
(x - I)2(1 - 2yf
(^TlT = ~F~ x y*
1 = n( ~ 0 y),2 * y = y-4y*
. 2
1 ?/) = max. and .-. ys _ y2 + 1 y

+ 4?/3 = 4 (?/3 - ?/2 -f —
= max. Now let ?/ = ^ + —
...y* = S* + Sj* + ±Z + _L
1 11
4y= +T2r + I2
( 123 )
•'•y3-y* + iy = zS-i2z + T2-T7= max- and
12. 1
To ~ kz ig a constant quantity, and .\ z3 — —- z = max.
— r, . • . z3 — — z — r = 0.
Let one of the positive roots of this equation = «, and

consequently z — a must exactly divide it.
z - a J z3 - — s - r = 0 |^2 + az + a2 - ^ = 0,...(A.)
^3 — az*
az2 — a2z
(fl. _ ^), _ r
(a*__^_a(a*_^)
■••r=a(a*-A)
12/
1 v . /•
and a2 — —
12 = —
a .*. from equation
^ • (A)
\ / z2 -f az = a
^ / /y3 Aai
or £ = — —:+: A / — where 4r cannot be greater than
a3,, ,\ when r = max. we must have a3 — 4r = 4a la2 — — j
or a2 — 4dl — — .• . a2 = — and a = — . Also 0 = — —

3 9 3 2
= 1 ' 11 1
— and y = #-f — ==— •.#+! = — = 6, . \ a? = &.
6 3 6 y
The same solved without eliminating the second term of
the cubic equation y3 — y2 + — 2/ — r = 0.
4
Let a = one of the positive roots of this equation, and
consequently y — a must exactly divide it,
( 124 )
y-aj ys-tf+±y-r=0 (y+(0_i)y+ («_i)2=03 (A.)

if — ay2
(a-l)y2 + -jy
(a — 1) y2 - a(a - 1) y
(a-\)y~r
-)2
(a - ^)\ ~ a(a -
r =
/ 1 \2 r t 1 \2
a( a - — ) or — = (a — ) } and from equation (A)
2 , / -n »' «-l . . /a(a-l)2-4r

y«+(o_l)y = __ory = --x-=fc Y 4^ •
Now in order that r or 4r may become a max. we must have
a{a - I)2 = 4r = 4a (a - y) .-. a2 - 2a + 1 = 4a2 -
j.i 2.. a — 11 .-i1*

4a + 1 or a = — and ^ = -— = — .\ a? + 1 = — = o
and a? = 5 as before.

r a -•
In the equation z2 + az = let z = w ^ and .*<,
a2 + ^
^2 -j- a# ■ = w2 — aw + -r a2 a2 = """"a"
a* = w* ~~ "i" f
a3 — aw2 = max. when w = 0 .*. -r-

... r = — #3 == r =
4 4
. • . a? = 1 = 5 as before.
2/
( 125 )
PROB. (27.) TO SAW OUT OF THE TRUNK OF A TREE A

RECTANGULAR BEAM THAT SHALL HAVE THE GREATEST
POSSIBLE POWER OF SUSPENSION. (Fig. 58.)
Actual experiments lead to this result, that in a parallelo-
pipedon of uniform thickness, supported on two points and
loaded in the middle, the lateral strength is directly as the
product of the breadth into the square of the depth, and
inversely as the length.
Let ACBm be the circumference of the trunk and the rect
angle AB the base or top of the beam cut out of the trunk.
AB = diameter of the trunk = a, AC = breadth = x, and
BC = depth of the beam = \/a? — x2. Also let/= strength
of the wood of which the tree is composed, and Z=the length
of the beam which is in this problem = a constant quantity.
"We have before observed that the power of suspension =
/ x breadth x depth2 _ fx{a2 - x2) f
ISSitE - I - T («*-#) -
max. .*. a2x — x3 = max. = r .*. x3 — a2x + r = 0. Let
one of the negative roots of this equation = b, and conse
quently x + b must exactly divide it.
x + b) x3 — a2x + r = 0 {_x2 - bx + b2 — a2 = 0, ... (A.)
x3 + bx2
— bx2 — a2x
— bx2 — b2x
(b2 — a2) x + r
{V-a*)w + b{b2 -a1) .-. r = b{b2 - a2)
— a2 = -j-. From equation (A) we find x2 — bx = —

y* 7\ /b3 4t*
-t- or # = ~ =!= A / —T7— and hence it is evident that when
b 2 V 46
r or 4r = max,, b3 = 4r = 4J (Ja — a2) or 36a = 4a2 .'. b ==
( 126 )
.•.# = - = —jzz = breadth and v a2 — ocl = A / a2 — —

2 ^3 V 3
=«V4 3
depth of the beam. Now from the points m and
C draw mr and C/z perpendiculars to the diameter AB, then
by prop. 8, 6th Book Euclid, we have AB I AC :: AC : An
0 9
oy a I m :: m : An — 00 CL
— = -pr €b
= ■;;. Also AB I Bm : : Bm I
a 3a 3
o
^r or a I % : : x I — = Br = - ,\ ?zr = ^4B — ?'Ii — ^f^ =

a 3
# == _. hence the following construction.
o o o
Divide the diameter of the trunk into three equal parts,
and from the two points of section draw the perpendiculars
and complete the rectangle, which will be the base or top of
the rectangular beam required.

r ft
In the equation oo1 — boo = j- let w = y + -~- . ' . oc1
52 52 52 r
boo = if + 6y + — - % - — = ^ - — = - T .-.
ft3
== — ft3 or ft (ft2 — a2)
%2 = max. when y = 0 . e . r = —
ft3 , 2a ft a , „
—r or ft = —7= and a? = — = —7=. as before.
^ a/3 3 a/3
CHAPTER III.
PROBLEMS OF MAXIMA AND MINIMA IN THE SOLUTIONS OF

WHICH EQUATIONS OF THE FOURTH, FIFTH, SIXTH, AND
SEVENTH DEGREE ARE USED.
Section 1.
PROB. (1.) WHAT FRACTION IS THAT THE FOURTH POWER
OF WHICH BEING SUBTRACTED FROM ITS CUBE THE RE
MAINDER IS THE GREATEST POSSIBLE?
Let x = the fraction required, .'. Xs — a?4 = max. = r
. • . x4* — xs + r = 0. Now let the product of the two values
of this equation = x% — ax + b, which must consequently
divide it exactly, and .*. we find,
x*—ax + bj x*—xs + r=0 l_x2+(a — l)x + a2—a — b = 0, ..(1)
x* — axs + bx2
(a — 1) x3 — bx1 + r
(a — 1) xs — a (a — 1) x2 + b(a — 1) x
(a2 — a —■ b) x2 — b(a — 1) x + r
(a2 — a — b)x2 — a{a2 — a — b)x -\-b{a2 — a — b.)
Now it has been proved in the introductory chapter that
when any equation is divided by two factors of the form x— c,
x — d, successively, or by their product of the form x2 — ax
-f 7; at once, then the remainder R must be equal to zero and
entirely independent of x in the case when c and d are
supposed to be the roots of the given equation. We therefore
find b {a — 1) = a (a2 — a — b) and r = b (a2 — a — b) ... (2) ;
v
.'. -j- — a2-— a — b. Also we have b{a — 1) =■ a{a2— a — h)
( 128 )
or ab - b = a3 - a2 - ab, .\ 2aZ> - & = 6(20 - 1) = a? -
a3 — a2 a?(a — 1) „ 7 o
tt2 ... i = = _^ -1 ... a2 - a - & = a2 - a -
2a — 1 2a — 1
2a~=T - (2«~- 1) • ' r ~ b{a a b] ~ 2a - 1 x
0(0-1)* a3(«-l)3 , , 4a3 (a - l)3 ,„,
Now from equation (1) we find, x* + (a — 1) a? = t- and
a-1^ /b{a-\Y-4r
solving this quadratic we find #= —HpT""— /v ZZ
r(g-l)Baa(g- 1) _ 403(0 - I)3
a-l _^ / 2«-l (20 - l)2
■j Here
2 /V 46
4ta3(a — l)3
it is evident that 4r or -7—^ 7—- cannot be taken so great
(20 — iy
as to make it greater than b(a — l)2 or - =
and consequently when r=max. we must have -—^—^

= 403(0
(2a -— 1)«
l)3 °r 1 = 2^T
40 A .•. a = -
or 2a - 1 = 4«
1, a- 1 J-13
-and. = __=__=_.
T%e S0me solved without impossible roots.

r n* ft 1
In the equation w2 -f (0 — 1) # = t- let # = y —
~b
(0 — I)2
and .\ a?2 + (0 — 1) 00 = «/2 — (0 — 1) y 4- -—^ -f
(a - 1) y - v(« -
^ !)2
y = y*9 - (« ~ 1)2
4 r and
= - -j ,1 r =
6(0 — l)2
~—7— by2, which is evidently a maximum when y = 0,
b(a _ 1)* . 03(0~ I)3
and consequently r = ——-—'— ; but r is also = , 'i\2'
( 129 )
b{a - I)2 _ a3{a - 1)
and therefore we find —— = —r^ 7T0- or b
4 (2a - l)2
4a3 (a - 1) a2 (a - 1) __ 4a3 (a - I) , __ 4a
(2a - I)2 °r 2a - T" ~ (2a - I)2 ''* ~ 2a - 1
1 a -I i-l 3 , ,
a = - — .-. x = « -— = — =—t— = -r as before.
2 2 2 4
PROB. (2.) TO FIND SUCH A FRACTION, THE FOURTH POWER

OF WHICH BEING SUBTRACTED FROM ITSELF, LEAVES THE
GREATEST REMAINDER POSSIBLE.
Let x = fraction required, then by the problem we find

x •— x4* = max. which let = r .'. x41 — x + t = 0. Let
x2 — ax + # be the product of the two values of this equa
tion, which must consequently be exactly divided by it,
. • . x2 — ax + b J x4* — x + r = 0 (^2 + a^ -f a2 — 6 = 0, . . . (1)
<#4 — ax3 + &#2
a«£>3 — fe?2 — <# + r
ax3 — a2x2 -f #&a?
(a2 — b) x2 ~ (ab + 1) x + r
(02 - 5) ^ _ a(ft2 __ £) ^ + J(a2 _ b)
. • . ab + 1 = a3 — ab .' . b = a3
— ]_ and a3 — b — a2 — a3
~-r„ ]_
2a 2a
a3 + 1 , ., 9 ,. a3 — 1 a3 + 1
= -2T-- and r = b<r ~b^ = -ST" x -25- =
~ -. Now from equation (1) we find x2 + ax
r a / ff~h 4y
—r— . Hence it is evident
44
that 4r cannot be greater than a2b and . • . when r or 4r=max.
i. i 2* ^ i i. z. a3 — 1 -, («3 — 1) (a3 + 1)
we must have arb — 4r : but b = ——— and ~
2a 4a
( 130 )
_ m a2(a3 - 1) _ 4(a3 - 1) (a3 + 1) 2a3(a3- 1)
~ r -'• 2a ~ 4a2 °r 4a2
4 (a3 - 1) (a3 + 1) .-. «s= 2«3-f 2 .-. « = y~ 2 = - ^2
4a2
a ^2 72 1
2 2 V 8 «/"4
7%e same may be solved without impossible roots.

7* Oj
In the equation x2 + ax = =- let x — y ^- and
0 A
a2 a2
therefore we find x2 -{- ax = y2 — ay + — + ay — =
a2 t a2b
y2 — = — or r = — by2 which is evidently a
a2b or 4r = a2b.
maximum when y = 0 . • . r = —- But 4r =
* 4
4(a» - 1) (a3 + 1) 8/TT j a ^T
-^ f-^ -* r. a = — v 2 and #= — —== ——
- a'/* X
as before.
4
PROB. (3.) TO DESCRIBE THE LEAST TRIANGLE TCt ABOUT

A GIVEN PARABOLIC ARC APB OF WHICH C IS THE
focus. (Fig. 59.)
Let AN = x, AC = a, and therefore tC = a + oc. Also
by similar triangles we find, tN \ NP :: tC : CT ov2x : 2VW
:: « + * : CT = (a
v- +■ a?)
~; x- \/«
v ~ X
- \/x % ~
v ~ .-. CT ___
= {a + x)\/a
v^_^/«
oo 2 2V<^
(7J7 x *C _ (a + x)2 x \/"
and therefore the area tTC =
2 2\/ ^?
( 131 )
= mm, and . • . (a + wY = mm.

. and .• . x — = max. Let x
x \a-\-xy
ab
__ ab u x _ c _ abc3 __ 1
~~ T ' (a + xY "~ JaFT~acY ~~ a*(c + 6)4 ~~ a5"
"c4
or = max. It is evident that
c + by {c + 6)4 ~ *- — — * ^ + ^4
b c3 _ / c
c + a X (c + by = ~ (\ 1 — T+b)) x 7 + bf
~ (c 7T^. Let v =
c be3
—=- . . 7Va = (1 — y) v3 = V3 — V* = max. In this
3 c 3
case by problem (1) we find y = —, but 7 = y =—
c + b 4 b 4 ab % x b
or .-. 1 H = — . But a? = — .*. — = —
c 3 co c a c
X 4 ., 1 x 1 .. a
•"• 1 + = "
a
The same may be solved without impossible roots as pro
blem first.
The same may be solved by the following more direct and
common way by which the two first problems have been
solved.
ay
Let a + x — — . • . the y 1 - ay w if
y (a + x) 4 1
*/4
=y3 — m/4=max. and .\ — 2/3— 2/4=max. which let =r .*. y*
y3 + r = 0. Let y2 — by + c = product of the factors

of the two values of this equation and consequently we have
ya-Jy + CJ y4_ly3 + r=0 (jf + (* ~ ^ + 42- ~ - C = 0, (1 .)
2/4 — fo/3 + C2/2
(4 ) y3 - g2 + r
( 132 )
(b"--a-c)f-c(h-l}y + r
(4»_i_c)y._i(y_|_c)lf + e(i._|_c)
and therefore r = e(b2 c) (2.)
and c(b --) = b{b2- - - c) (3.)
c #2
From equation (3) we find be = b3 — — — be and con-
a a
sequently 25c
c =
- #hs __ A . g(2a& - !) _ ÂLzA
a a a a
.*. c — _ 7 —^ and from equation (2) we find r =

lab — I
'\° a ~ > " 2ab-l X V a 2ab - 1 )
_ b2(ab - 1) /aft2 - b __ &2(a5 - 1)\ _ b2{ab - 1)

~" 2ab - 1 V a 2a& — 1 / ~ 2a6 — 1
/a2*3 - 2ab* + S\ _ 52(a* - 1) (a£ - I)2 x b _ {ah - I)3*3
\ 2a2b - a / ~ 2ab - 1 2a2* -a " [2ab - 1) V
From equation (1), we have y2 + ( b — —J ?/ H = 0, .'.
5-1
?/ = h / (6 J x e — 4r and hence it is
evident that r or 4r cannot be taken so great as to become

/ 1 \2
greater than lb ) x e and consequently when r = max.
/ 1 \2 Mob — l)3*3
we must have 4r = q# — — ) . But 4r = ,rt , T^—
\ «/ (2a* — l)2a
( 133 )
(k lia - bâb ~ *) lh ~ —\* - bâb ~ 1)S
•'•4r~cr a) - 2ab- 1 \ o/ ~~ «a(2«6-l)
and .'. 2a6 = — 1 .*. b = — — ; but «/ =

2a' * 2
JL _ JL
2a a 3 1 4a a ..
= — .\ a + a? = — = — = & + — and
2 4a ' «/3 3
#? = — as before.
o
The same may now easily be solved without impossible roots.

(1 \ T
h ) V — let ^ = # —
b- —
_îL and .-. jf + (6_i) ,, = *•- (»-4)' +
\ a' , /, 1\ V a/ = *20 V a1
(4-1
V a
= and consequently r = ~~ cz<2> wîch is
evidently a max. when # = 0 ; . • . r — «('-t)'

^ •"• 4r =
/, 1 \2 c[ab - l)2 , 62(a6 - 1) ,

c(i ) = -^ s }- but c = -^-= y1 .*. 4r =
V a/ a2 2a6 — 1
J«(aj __ 1)3 . 4(a6 - l)sjs #(aj - l)3
—- - -- and 4r is also ■—■ — — * — -
a2{2ab - 1) (2a6 - I)2 x a * * a2(2a6 -1)
- 4(*ft - 1)3 >< ^ • 1 - 4ft • j-_i but*,
~ (2a5 - I)2 x a " ' a ~ 2a6 - 1 2a y
( 134 )
1 1 1
b
a 2a ~~ a 3 , 4a
= — = — and a -f a? = —-
2 2 4a 3
a a
a + .-. X = as before,
3~
PROB. (4.) LET AB BE THE DIAMETER OF A CIRCLE, IT IS

REQUIRED TO FIND A POINT, C, IN THE DIAMETER, SO
THAT THE RECTANGLE FORMED BY THE CHORD DE, WHICH
IS PERPENDICULAR TO AB, AND THE PART AC MAY BE
THE GREATEST POSSIBLE. (Fig. 60.)
Let AB = a, AC = x, and C.B = a — x, then [a — x)x

= CD2 and CD = \/«# - <r>; therefore D^ = 2\/^ - x2,
and the rectangle EG = x x 2vax — x2 = max. . * . its square
4<2?2 (aa? — <a?2) or 4ax3 — 4<2?4 = max. .'. ax3 — x* = max.
which let = r . • . ^4 — 0#3 -f- r = 0. Let <#2 — bx + £ =
product of the two values of this equation, and therefore
we find ;
x2—bx-\-c) xi—ax3 + r=0 [_x2 + (b— a)xJt-b2—ab— c=0, (A.)
<2?4 — &2?3 -j- C<2?2
(b — a) x3 — ex2, + r
(b — a) x3 — #(# — «) <^2 -f- c(b — a) x
(b2 — ab — c) x2 — c(b — a) x + r
(tf-.ab-c)x2-b(b2--ab-c)x + c(b2-ab-c)
.-. c{b — a) = 6(62 - a£ - c) (1.)
T
and r = c(b2 — ab — c) .\ 62 — ab — c = — (2.)
From equation (1), c =

~ ^ .• . b2 — ab — c = b2 —
&b — a
. b\b-a) bib -a)* r ^T .
a6~ IS—;T=4r—r = ^r- Now from (A.)
2b — a 2b — a c
( 135 )
r b — a
we have x% + (b — a) x = — — or x =
c 2
v (ft
-—:—
4
q\2
(b - af
y
and it is here evident that when ror- =
c
r
max. we must have -—-—- = — = -77
bib - ay 1
c
• '• ~t =
y*
4 c 2b — a 4
6 , ^,q__^"~"tt_ 2
2b — a ' ' 2
"" ~4 '
7%£ sawze solved without impossible roots.

7* b -~~" CL
In the equation a?2 + (b — a) x = let x = y — ——
(ft — a)2
.-. a?2 + (* — a) x = «/2 — (J — a) y + -—j—^- + (J — fl)y
(ft _ fl)« (ft - «) » _ r _ c(ft - tf)2

2 V 4 c *'* r 4
,
— cyz = max. when r = c(& - a)» .\ -—-.—-
——-.—— (b-ay = —
r =
4L 4e C
bib - ay 1 b , a b- a
"771 •'• -r = <n or ft = — and x = —
2b — a ^ 2b — a 2 2
= — as before.
4
PROB. (5.) TO DIVIDE 12 INTO TWO PARTS, SO THAT THE

LEAST MULTIPLIED BY THE CUBE OP THE GREATEST, SHALL
BE A MAXIMUM.
Let x = greater part .'. 12 — x = lesser part and 12^3 —

x41 — max. = r .\ x* — 12<#3 + r — 0. Let the product of
the two values of this equation = x2 — ax + b.
( 136 )
-ax+bj x4~-l2xs + r=0 lcc2+(a-12)cc+a2-12a-b=0,(&.)
xA — ax3 -f bx2
(a — 12) x3 — bx2
(a - 12) x3 - a{a - 12) x2 + b{a - 12) x
(a2 - 12a - b) x2 - b{a - 12) x + r

(a2-l2a-b)cc2-a(a2-12a~h)% + b(a2-12a-b)
.-. r = b(a2 - 12a - 5) .*. -£ = a2 - 12a - * (1)

Also % - 12) = a[a2 - 12a ~ b) = a3 - 12a2 -oft >.-.
**> io\ s io 2 / a* -12a* a2 (a - 12)
b{2a - 12) = a3 - 12a2 ... * = -^-—- = -^zig-
.-. from (1) -^ = a2 - 12a - 6 = a2 - 12a - ^ " jff- =

o <â — i^
a3 - 24a2 + 144a a (a - 12)2 . . a(a - 12)2
2a — 12 2a — 12 2a — 12
a\a - 12) a{a - 12) 2 a3 (a ~ 12)3
Now from equa
2a -12 2a -12 (2a-12)2*
tion (A) x2 + (a — 12) x = r or x = ~—
b(a — 12) 2— 4r
a/ —Za ™~% êre ^ *s evident that when r or 4r=
max. we must have b(a — 12)2 = 4r or -~ — (a — 12)2

â — i<£>
= 4a3(a-12)2 4a
(2a - 12)' °r J = 27^12 •••«=- 6„ and^
, = -
2 _y>
7%e same may be solved without impossible roots.

T
In the equation a? + (a — 12). a? = j- let x = y
- ^v^ ••• ^ + (» ~ 12) x = y* - (« - 12) y +

( 137 )
^(a -£—i-
12)2 + (a
, - 12)
10. y - ^(a-Tz—+~
12)2 = y22 - ^—^—^
(a - 12)2
r ••• r =
- T J (a -4 12)2 . 2 .but. r = a^(a
by W^T2T- 12)3 and,
_ ^2(g - 12) . a2(a- 12) g3(a-12)3 .

2a - 12 ' " 4(2a - 12) l j ~ (2a - 12)2 ' '
a — 12
a = —. 6 and x = — = 9 as before.
PROB. (6.) TO INSCRIBE THE GREATEST ISOSCELES TRI

ANGLE IN A GIVEN CIRCLE. (Fig. 61.)
Let ABC be the isosceles triangle required, and suppose

BD = x and BE = diameter == 2a .-. DE = 2a — x r.
half the base = AD = V2ax — x2 and area of the isosceles
triangle = AD x BD = <# \/2a<2? — a?2 = \/2a^3 — a?4 =
max. .'. 2a<#3 — x4 = max. = r .*. x* — 2a«^3 + r = 0.
.*. let x2 — bx -\- c = product of the two values of this
equation,
ap—hx+ej xA—2a%s+r=0 {x2 + (b— 2a)x + b2—2ab-c~0, (A.)
x41 — bxs + ea?2
(6— 2a) x3 — ex2

(b-2a) xs-b(b-2a) x2 + c(b-2a)x
(b2—2ab—c) x2—c(b—2a) x-\-r

(h2-2ab-c)a;2-b(b2-2ab-c)x+c(b2-2ab-c)
,-.r= c{b2-2ab-c) .-. — = 62 - 2a6 - c (1.)
Also c(b - 2a) = 63 - 2a62 - 6c .\ c = §^|^ •'•

r A2 o a m o * 63-2a62 6(6 -2a)2
— = b2 — 2a6 — c = b2 — 2a6 —?—^r— = -^——-
c 2b — 2a 2b — 2a
( 138 )
6(6 -2a)2 _ 62(6 - 2a) b(b - 2a)3 _
26 - 2a ~ 2b -2a X 26 - 2a ~
63(6-2a)3 . 463(6-2a)3 XT „ 4. ...

"(26^2^ °r 4'' = (26 -2a)2' N°W fr°m eqUatl0n (A)
#3 + (5 — 2a) <2? = — — or a? = — dz
c 2
Vc(b — 2a)2 — 4r
— ^ and here it is evident that when r or 4r
= max. we must have c(b — 2d)2 = 4r or -^7 S-^~ =

463(6 - 2a)3 4b
(26 — 2a)2 26 — 2a
'—-— = —. Hence AD = \/2ax — x2 =. A/ 3a2 —
= 2^ .-. yiC = 2Î> = «V"3". ^££ = a/^Z>3 + BD*
= A/ — + —- = \/3a2 = a\/ 3 ,\ the triangle required
is equilateral.

T
In the equation x2 + (b — 2a) x = let # = y —
ill*? . • . * + (6 - 2a) « = </3 - (b - 2a) y + {±^

/» « n (6 -2a)3 , (b -2a)3 r
+ (6 _ 2a) y - ^ 2 ; = */3 - 4 =-—.-. r =
c(6 - 2a)3 , . _ c(b - 2a)2
——-—- cy1 = max. when y = 0 .*. r = -r .
But r= ¥{l - 2a)2

{2b- 2a)3 and, C = 63(6
-2b=W- 2a) ••• b*(b - 2a) » =
(26-2«)2
^ZlM x iiziM! or 1 = ** .-. i = - « and

26 - 2« 4 26 - 2a
( 139 )
x = - h-=-— = *jff. Hence AD = y/Zax - ^ =
2 2
a/^T"^= 2^1 ... AC=2AD=2 x *£l = aV^B
V 4 2 2
as before.
PROB. (7.) TO INSCRIBE THE GREATEST PARABOLA IN A

GIVEN ISOSCELES TRIANGLE. (Fig. 62.)
Let AGF be the given isosceles triangle and CHPME the

required parabola. Let AD = b, GD = a, and GP — x.
Now KPG being a subtangent to the parabola, we must have
by conic sections GP = PK = x .' . GK = 2x ; also PjBl :
PD :: flic* : CD2 (A.)
Now by similar triangles GD : AD : : GK : flif, or a : Z> : :
2<2? : 2bx = HK
Trr7 . • . ,by proportion
' . , A , , x I a — a? : : 46V
(A) —^ .•
CD2 .-. CD2 = ~ (a - a?) a? .-. CD = ?V(a - a?)a?. Now

2 26 2
the area of the parabola or — PD x CD = — x -^ {a — x)
o do
\/{a — x)x = — \/ [ci — oofoo = max. or (« — ^)3<^ = max.

o(l
Let a — x — y .*. x — a — y .-. (a — #)3# = */3 (a — ?/)
= a?/3 — y4 =: max. = r . • . «/4 — «z/3 + r = 0. Proceeding
exactly as in the solution of Prob. (4) we shall find y ■ =
3a , 3a a
— and x — a — y = a — = —.
4 ^44
The same may be solved without impossible roots as
Prob. (4) was.
This problem if solved by the common method given in
works on Diff. Calc. must ultimately produce a cubic equation,
to solve which is generally tedious.
( 140 )
PllOB. (8.) TO DETERMINE THE GREATEST PARABOLA THAT

CAN BE FORMED BY CUTTING A GIVEN CONE ACD.
(Fig. 63.)
Let nv, parallel to CA, be tlie axis of the parabola rvm
and rm the base (or ordinate) thereof. Putting DC — a,
bx
CA = b, and Dn = x ; then, by parallels, a l b : ; x I — =
a
nv; moreover by the property of the circle, we have rn2—
nm2 = Dn x Cn — ax — x2, the square root of which multi
plied by — x — (because every parabola is — of a paral-
lelogram of the same base and altitude) gives —— v ax — x2

for the contents of the parabola = max. .\ ax3 — x4 = max.
= r . • . x4* — ax3 + r = 0. Now by proceeding exactly as in
Prob. (4)
v ' we find x = —-
4 when ax3 — x4* = max.
The same may be solved without impossible roots in exactly
the same manner in which Prob. (4) was.
PROB. (9.) THE CORNER OF A LEAF IS TURNED BACK, SO

AS JUST TO REACH THE OTHER EDGE OF THE PAGE, FIND
WHEN THE PART TURNED DOWN IS A MINIMUM. (See
Fig. 51.)
It has been shown in Problem (20) Chapter 2nd that aA x
PQ = 2AQ x AP and that aA = V2ax. PQ = A/- 2^

V 2x — a
V4?ax^ — 2x x AQ ,\ the area of the part
( 141 )
., , x x AQ 2 /— A / x4*
turned down = —-— = - V a ^ ^—^
^ X41
V 2<a? — a
= min. .\ -
y + a = x .'. r x*
2# — a
(y +
= min. Let 2x — a = y
*—— = vy ^ a)4 = (y + a)4

^-.^ '■ = mm. or
2 2# — a 16 Joy
y
16y
u y
= max. or 7—f—— = max. Aiii.
Also tt&
let y = —
(2/ + a)4 (y + «)4 c
^ c _ c4^^ __ c'3^ _
" ' (y + «P == a4(& + c)4 "" a*c(b + <?)4 ~" a3(6 + c)4 ~
e4
1 be3 be3 _ , 6c3
max. — -, = max. But
a3 (b 4- cy ' (b -f c)4 (b + c)4
#3 — £4 = max. Proceeding exactly as in problem (4) we

., j2r = —-
find 3 or - c = — 3 .•. 6 + c = & ^ _1 = — 4 =.
4 6 + e 4 c c 3
1 ^ h 1 ^ ah a , V + a
T + x •'• T = ¥ and T = T = y and * = ~2~ =
a 2a
3" + a = 3-
2
The same may be solved without impossible roots as
Prob. (4) was.
Section 2.
PROB. (10.) TO FIND SUCH A VALUE OF X AS MAY MAKE
MX* — X5 = MAX. = r.
"We have now the equation x5 — mx4, -f r — 0, and let the

product of the three values of this equation = x3 + ax2 -f
bx + c and . • . we have
( 142 )
x?+aoc'2+hw-\-cJ x5-mx*+r=0 \<x2-(a-\-<m)x-\-a2 + am-b=0,(l)
och + ax4 + bx3 + ex2
— (a + iT^x^—bxP — cx2
- (a + m)xA- (a2 + am)x3-(ab + bm)x2- (ao + cw) a?
(«2 + aw — &)#3 + («5 + to -c)x2 + (ac + c?7£)# + r
(a2! + am-fyx8;+ (a? -{■ a2?n-ab)x2 + (a2b -{• abm~b2)x
-\-c(a2 + am—b)
.-. a2 + am-6 = —
c (2.)
v '
Also ab + bm — c = a3 -\- a2m — a6 (3.)
ttc + cm = a2& + abm - 62 or c = ... (4.)
a -\- m
Prom (3) and (4) we have ab -{- bm ■— c = ab + bm —
a2b + abm — b2 __ a2b -f a5m -f «3m -f 6m2 — a2b — abm + 62
a + m a -\- m
abm + &m2 + b2 7 2 2
a + m
a4 + 2ma3 + a2m2 — a2b — ma£ or b2 + (2ma + m2 + a2) 6
= (a2 + am)2 or b2 + (a + m)2Z> = a2 (a + ^)2 or b
(a + m)2
= - Hr + V/(a" + m)4 + ^
4a2 (a + m)2
0r h_ =
(a + m)2 + a/ (a + m)4 + 4a2 (a + m)2
2 '
_ /-.v o / n r a + m
From (1) og2 — (a + m) x = or #. = —-— +
(a
v +-—m)2 r .•. when
, (a + m)2 = —
r = aL
V 4 c
r = max. -—-.
4 c
2a(a + m) + (a -f m)2 — (a + m)V(a + m)2 + 4a2
+ am — 6 = — ~ —
Z
or a + m = 4a + 2a + 2m — 2a/ (a -f m)2 + 4a2 or
5a + m = 2a/ (a + ^)2 + 4a2 or 25a2 + 10am + m2 = 4a2 +
8am -f 4m2 + 16a2 or 5a2 + 2am = 3m2 .* . a2 + -z- & = -=-
5 o
( 143 )
m . A /15m2
••• a = ~ T ~ V -w + ml 4m
25 = T m
~T = 3m
T' and,
3m
-^
a + m 5 h ^ 4m TP _ ,. 4
a? = —— = = —. If m = 1, then a? = — .

mx3 — x5 = max. = r.
We have now the equation x5 — mx3 + r = 0, and let the
product of the three values of this equation = x3 + era?2 -f
Zw + c, we therefore find,—
cc6-\-ax2-\-b3C+c) ccb—mx3 + r=0 {^%2—acc-\-a2—ft — t»=0, .... (A.)
^5 -f- ##4 + &E3 + £#2
— aoc^ — (Z> + m) #3 — ex2
— aotfi — cPofi — dbx2 — cax
(aP—b—m) a?-\-(ab — c) x2 + cax + r
(a2— ft— m)xs -f (^3 —aft— mn)x2 + (a2£ — ft2 - ftm)x
-\-ca2 — be — cm
T
.*. r = ca2 —- 6c — cm or a2 —- 6 — m = —
c (1.)
v '
ab — c = a3 — ab — am (2.)
flh — u --— riwi
ca = a2# — b2 — bm or c = ... (3.)
-, 7 _ a2Z> — b2 — #m Z>2 + #m
and r . ab — c =■ ab — = =
a a
a3 — ab — am .\ b2 + bm = a* — a2b — a2m or i = -
a2 + m /a4 + 2a2m + m2 -j- 4a4 — 4a2m __ a2 + m _^
_____ ,y __ _ __
V5a4 — 2a2m + m2 __ [a2 -f m) + VW - 2a2m + m2
4 ~" ~ 2
and equation (A) gives x2 — ax = —7* . • . x = -d=
CL
A// -jfl 7*
( 144 )
a2 r , 2a2—- 2m + a2 + m-— \/5a4— 2a2m+m2

. \ —- =z~=za2 — m — 6 =
4 c 2
and a2 = 6a2 — 2m — 2\/5a4 — 2a2m + m2 or 5a2 — 2m =
2\/5a4— 2a2m + m2 . • . 25a4 — 20a2m + 4m2 = 20a4 — 8a2m +
1 2<m- a
4m2 or 5a4 — 12ma2 = 0 .'. a2 = ——> and a? = — .*. x2 =
5 <^
a2 12m 3m A /3m ., /3
= — .-.# = A/ —. If m = 1, then # = A/ -~-.
4 4x5
mx2 — x5 = max. = r.
"We have the equation x5 — mx2 + y = 0, and let the pro
duct of three values of this equation = xs + ax2 + fo*? + c .'.
xs + ax2 + bx + cj x5—mx2-\-r—Q \^x2—ax + a2—b = 0, ... (1.)
a?5 + a«2?4 + i#3 + c#2
— ax41 — Ja?3 — (c + m) <^2
— ax* — a2xs — a&a?2 — acx
(a2 — 6) x3 + (a& — c — m)x2 + ac# + r
(a2- 1) a* + O3- al)a? + (a2b - W)x + c{a2- h)
.-.a2-b = — (2.)
c
ab — m — c = a3 — a& (3.)
#c = a2b — b2 or c = (4.)
a
a% _ 52 a2b _ am _ a2b + b2
.• . ab — m — c = ab — m — =
a a
b2 — am „r 7o OJ 70
= = a3 — ab or ¥ — am — a4 — aô or cr +
a
a26 = a4 -f- am .•. b = — — a* + A// ——|-a4 a4 + am
2
— a2-f \/5a4 + 4am ,\ a2—
0 0=
, 3a2- v//5a4 + 4am and, from
n (1)
,,x
( 145 )
and (2) x = — + A/ -—- — ,\ when r = max. we must

2 V 4 c
, a2 r 3^2 _ /v/sa*.].^^
have -— = — = a2 — & — '—— ,\ a2 = 6a2 —
4 c 2
2v 5a4 + 4am .e.25a4 = 20a4 + 16am .\ a3 = —- and# =
PROS. (13.) TO FIND SUCH A VALUE OF X AS MAY MAKE

mx — x5 = max = r.
We have xs — m# + r = 0, and let the product of three
values of this equation = x3 + #<#2 + hx + c, and therefore
we have,
#s-f-a#24-&#+eJ #5--m#-hr=0 [^x2—ax + a2—b=z§, .... (1.)
x5 + a<#4 + &#3 + C<2?2
— ax41 — bx3 — ex2 — wz<2?

— a#4 — a2x3 —- a&#2 — acx
• \ c(a2 - 5) = r .-. a2 - Z> = — (2.)

Also ab — c • = a3 — ab (3.)
ca — m ~ a*b — b2, .*. c = a ......... (4.)

x '
7 7 a2b — b2 + m b2 — m
r. ab — c = ab = = a3 — ab or
a a
b2 — m = a4 — a26 .-. 62 + a2b = a4 + m and .\ & =
— a2 + A/5a4 +~4m XT a /a2" JT
— ^ —. Now from (1), # = — +^/___
u
( 14G )
, a2 r 0 _ 0 ■— a2-f\/5a4-f m
.*. when r= max., -7- = — =-flr — 6 = cr—
'4c 2
1 fim ft2 y*
= 20a4 -f 16m. ,\ a4 = -— . Now since — = — we must
5 4 c
1 a A a* 16m m
Vt-
If m = 1, then ^? = a/ — — tt=-
VT 1
V b ^5
Section 3.
mx5 — x6 = max. = r.
Since we have x6 — m«#5 + r = 0, let the product of four
values of this equation = #4 + ace1 -f bx2 + ex + <# . * .
.T4 + a^3 + ^2 + c^ + 6?J #6— m#5 + r = 0 l#2— (a + m) x + a2 + am — 5 = 0, (l.)
a?6 + ax5 + 6<#4 4- c^3 + da?
— (« -f m) x5 — bx41 — ex3 — da?
— (a + m)#5 — (a2 + am) x4— (ab -f- Jm) a?3
— (c« -f- cm) x2 — (ad + 6?m) a?
(V + #m— 5) xA + («& + bm — c) xs + (c« + cm — d) x2 + («J + dm) x
(a2 + am — b )x4 + (a3 + a2m — ab) a? + (a26 + «5w — b2) x2 +
(ca2 + cam — bc)x
4- r
+ a7 (a2 + «w& — b)
T
.• . r = d (a* + am — b) .*. a2 + am — b — -j (2.)
Also «Z> + 5m — c = a3 + a2m — ab (3.)
ca + cm — d == c?b + abm — b2 (4.)
( 147 )
* . r 9 , 7 ca2 4- acm — 6c /r s
ad + am = cyt + aem — be, .*. d = ... (5.)
a +m
T v 7 , N c(a2 + «m— 6)
.\ ca 4- cm — a = c(a
v + m)J — d = c(a
v -\- m)7 a + m
__ c(a + m)2—c{a(a-\- m) —b] __ c(a + m) (a + m— a) + bc
a + m a +m
c(a + m) m + be , . 72
-_ _! ^ __ a2^ _|_ a^m — b2 = ab (a + m) — ft-
a + m
b(a + m) {a(a + m) — 5} ■ , " . ,0,
.\ c = ; 4- m)r + b=
m(a ': and from equation
H (6)
■ 7/ , i(« 4- m){«(a + m) — 6}
ab -{- bm — c = b(a v + m)' -, + w&); 4- o7
m\a =
mb(a + m)2 + b2(a -{- m) — ab (a -\- m)2 -}- b2 (a -\- m) __

m(a + »&) 4- 6
2b2(a
-—i + m)£—!+ Z>(a
i— 4- ^)2
-L—±(m — a)L = a2(a
9, +, ■ m)\ — ab*
m[a + m)+ b
or 2b2 (a 4- m) + b {a + m)2 (m — a) = ma2(« 4- ^)2
+ ab(a — m) (a + m) — «62 or {2(« 4- m) + a} b2 +
{ (a 4- ^)2 {m — a) — a (a + m) (a — m)}b = ma2 (a + m)2
or (3a 4- 2m)62 4- {(a 4- m) (m — a) {2a + m)}b = ma2(a + m)2
' ■ (« 4- m) (2a + m) (m — a) 7 ma2(a 4- m)2 ,
or o2 4- -^ —-^ b = -~ -— , . • . o =
3a 4- 2m 3a 4- 2m
— (a + m) (2a + m) (m — a) + \/(a + m)2 { (2a + m)2 (m — a)2 + ±ma\Za + 2m) }
2 (3a + 2m)
— (a + m) (2a + m) (m—a)+ \/(4a4 + 8a3m + 5a3m + 2am3 + m4) (a + m)2
= 2 (3a + 2m) ~"
— (a + m) (2a + m) (m — a) + (a + m)2 v4a2 + m2 . , N
—!> — '-— '-—-\ , since m — a= — (a — m)
2 (3a + 2m) *
.*.. 67 = i(a + m)^(2a + m)^-7^——rr\

(a — m) + (a + m)2 v 4a2 + m2 .
2 (3a + 2m)
^
From ,n\ r , \
(2)
w a(a + m)
h — 2<?(3^ + 2yn) (a + m) ~ (^ + m) (2a + m) (a — m) — (a + m)2 -v^4a2 + m2

~" ~ 2 (3a + 2m)
( 148 )
It is evident that 2a (3a + 2m) (a + m) = 6a3 + 10a2m + &am2
and (a -f m) (2a + m) (a — m) = 2a3 -f a2m — 2am2 — m3 .'.
2a (3a + 2m) (a -b m) — (a -f w&) (2a -f m) (a — m) =
4a3 + 9a2m + 6am2 -f m3 = (a2 -I- 2ma -f m2) (4a + m)
= (a + m)2 (4a + m), therefore we find a2 -f- am — 6 =
(a + m)2 (4a + wi) - (a + m)2 \/4a2 + m2 ^ /1N
— —-—
2 3a + ~—r—
2m) . From v(1),
;' a? =
a + m /(a + m)2 r # „T-u_ „ r_ (« + m)2

2 +' \f
V —~a—4 T
tf •"•
' ' wnen r == raaxv* 4
__ (a -j- m)2 (4a -f m) — (a + m)2 \/4a2 + m2 __
"" 2 (3a + 2m) '"' "~
2(4a + m) - 2v/4aTTm2 K 0 /T-T-l 5 OK a
— ^ .'. 5a = 2v4a2 -f m2 or 25a2- =
3a -f 2m
2m
,16a4
^ „ +, 4ma
. o or a = 2m
—, and oo = —jr— = 3 5
. a + m = 5m
-^
5
If m = 1, then # = —.
PROB. (15.) TO FIND SUCH A VALUE OF 00 AS MAY MAKE
m#4 —- #6 = max. = r.
Let y = a?2 .\ m?/2 — ?/3 = max. By Prob. chap. 2nd, we
, , 2m 2m /2m
must have y = — or or = -5- .\#= \f -^-. If m = 1^
o o ▼ o
then a? — A/ —-0
( 149 }

mx3 — XQ = MAX.
Let y = x3 .-. my — y2 = max. then by Prob. chap. 1st,

i. t m ^ m A fm TjP
we must nave y = — or x3 = — .\ a? = A/ -^-. It m = 1
then # = V7=*

m#2 — X6 = MAX.
Let x2 — y .\ W2^ — ?/3 = max. then by Prob. chap. 2nd>

_ or #2 = /y/ —
3^W~V 3 . • . x = /y/
V --;3 If
m = 1, then a? = "47=
v 3

mx — x6 = max. = r.
Since we have the equation mx —• x* = r or a?6 — <m«2? +
r + 0, let the product of four values of this equation =
x41 + ax3 -f bx2 + ex + d,
:.tf*+aa?+fta? + cx+dj ccQ—moc+r=® {jx?— aoc+a2— 5=0.. .(1.)
a?6 + a#5 + bx4* + c«#3 + ^2
— «^?5 — 6<2?4 — CX3 — ^2 — W2<2?
— ax5 -— a3#4 — abx3 — aca?2 — «d#

(a2— b)x4+ (ab—c)a?+ (ac— d)a?+ (ad—m)x
(a2-l)o£+ (a?~ab)a?+ (a%~l2)x2+(a2c-*'bc)a!
+ d(a2 - b)
,\ d(a2 ~~ b) = r or a2 - & = ~ ........................... (2.)
( 150 )
Also ab — c = a3 — ab (3.)
ac - d= a2b - b2 (4.)
ad -— m — ca2 — be (5.)
CCv ——~ be ~\~ TYh
Equation (5) gives d = .\ ac — d — ac —
ca2 — be + m be — m OT 70 a3b — ab2 + m
= = a2b — b2 . * . c = =
a a b
a b — ab2 + m 2ab2 — a3b — m
.' . ab — c = ab — , — , —
b b
a? — ab, . • . 2ab2 — a3b — m = a3b — ab2 . \ 3a62 — 2a36 =
70
m .-. cr 2a2
—6, = —m and1 .• . 67 = a? + va6 + 3am ,\ az„ —
3 oa 3a
7 c a3 + Va6 + 3am 2a3 — \/aP + 3am _ ,_ x
b = a2 = . From (1)
3a oa
a / a2 t a2
we find, x = — + A/ -; r> •'• when r = max. then —
' 2 V 4 d 4
r 2a3 — \/a6 + 3am
= -7- = a4 — 6 = and therefore 3a3 =
d oa
8a3 - 4vV + 3am . • . 25a6 = 16a6 + 48am, . • . a5 = — , and
o
a _ a5 16m m 5/m _„
then * = ^L
It may be remarked here that all the problems of the two
last sections of this chapter may be solved without impossible
roots, in the manner laid down in preceding chapters.
CHAPTER IV.
PROBLEMS OF MAXIMA AND MINIMA IN WHICH TWO OR MORE

VARIABLE QUANTITIES ARE USED.
If there are two variable quantities,, find the value of each

in terms of the other, according to the conditions of maxi
mum or minimum, and it is evident that by this means we
will find two equations by the comparison of which the values
of the two variable unknown quantities will be found in
terms of known constant quantities. If there be three vari
able quantities, find the value of each in terms of the other
two, and thus make three equations, by means of which the
values of the three unknown quantities will be determined.
The same method may be adopted when there are four or
more variable quantities.
The reason of this rule is obvious. When the being of
maximum or minimum of any function depends on the values
of two variables, for instance, then it is evident that the
value of a single variable in terms of the other, found on
the function being a maximum or minimum, will, itself, be a
variable quantity, since the other variable is not yet deter
mined; and consequently there will be infinite maxima or
minima of the function proposed. Now, in order to find the
required maximum or minimum out of these, we must solve
the function with regard to both for their maximum or mini
mum values, then compare these two values, and thus deter
mine them. The same reasoning may be applied in the case
of functions of three or more variables.
( 152 )
PR0B. (1.) TO INSCRIBE THE GREATEST PARALLELOPIPEDON

WITHIN A GIVEN ELLIPSOID.
Let 2x, 2y, 2z be the edges, 2a, 2b, 2c the principal diame
ters of the ellipsoid . • . the contents of the parallelopipedon =
Sxyz = u, and by what is shown in the Introduction we
find the equation of the ellipsoid to be
z2 x2 y2
c2 a2 b2
(x II \
1 — ~2 — j^j and .'. square of Sxyz =
Mx2y2z2 = 64arfy2 x c2 (l - -2 - |g) = 64c2 x
= max. and .•. a2b2x2y2 — 62#4?/2 — a^V = max. First

let x be considered as constant and y as variable . • . a2b2x2y2
- b*xy - aa«y = av pWîJ^ _ ^) = max. ...
aW _ b2x2y3 a2b2 __ £2^2
2—g £- - ?/4 = max. = r .*. ?/4 -^ */2 =
- r .'. 2/2 = aWj-VW

—2 ^ />/fjfP&
=fc ^ - b2x2)2
r^ L - r> ••• when
2a2 ~" V 4a4
(^2 _ 52^2
r = max. we must have -j~i — = r, and .\ y2 —
a2b2 — b2x2 ,, v
2a2 [ j
Now let y be considered as constant and x as variable, . • .
a2b2xy - a2xY - &¥#* = %2 (^ ~ ^ #2 - #4) =

a2b2-a2y2 9 A . a2b2 - a2y2
max. .-. —.—~ x2 — #4=max. = r, .'. ar
62 75——
#2
2 __ gaff _ qy /(a262 - a2?/2)2
r " 2&2 ~ v 4J2
( 153 )
± -Af,L = r, when r = max. .\ ar = ^ ... (2.)
.•.y* = 2 —. Comparing this equation with equa-
aW-bW aW-2bW 2aW-4bW

tion (1) we find ——— = -2 = ^r—
. • . SbV = a262, . • . ^?2 = —, . • . x — —j=- and . • . equation

3' Vs
9,2 __ a2^
(1) gwes ^2 = ^2 = -gy = -a .'.^^ and
V1 «2 b2) " V
3/ ~ 3' ' 3
c 4
, .. If v = volume of ellipsoid, i; = — pabc where p
2v or ^ : v : : 2 : pv , 3 .
3.14 &c. . • . u = —7=r
â/3
PROB. (2.) GIVEN THE SUM OF THE LENGTHS OF THE THREE

AXES OF AN ELLIPSOID, FIND THE LENGTH OF EACH, THAT
THE VOLUME OF THE ELLIPSOID MAY BE A MAXIMUM.
Let x, y, z be the three axes and s their sum ,\ x + y +

z = s.'.z = s — x — y and the volume of the ellipsoid =
%pxyz — ipxy (s — x — y) — ip (sxy — x2y — xy2) = max.
.*. sxy — x2y — xy2 = max. First let x = a constant quan
tity, .* . x (sy — xy — y2) = a? { (5 — x) y — ?/2} = max.
and . • . {s — x) y — y2 = max. = r . • . y2 — (s — x) y = — r,
n rp / (S Xl
and .- . y = —~— =£= A / 7— — r. Here it is evident
that when r = max. we must have -—-—- = r, .' . y =

s-^- no
( 154 )
Now let y — a constant quantity, .*. sxy — x2y — xy2 =
y {sx — x2 — xy) — y {{s — y) x — x2} = max. . \ (s — y)
x — x2 ~ max. = r, r . x2 — (s — y) x = — r, .\ x — * -*-
2
v (*-y)2
4
T~ ^ and. .* . (*-*/)2
-—j^~ — r, when r = max. .' . # =
4
—-—} .* . y — s — 2x. Equation (1) gives y = —-—, .'.
s —• 2x = ——-- — 2s — 4z*? = s — x . • . 3,2? = s, . * . x = — and
y — s — 2x — s — = — and z=s— x — y — s —
6 6 o
s s
— = —} and hence it appears that the axes of the elHp-
o o
soid required, when a maximum,, must be equal to each other ;
that is to say, the ellipsoid required must be a sphere.
PROB. (3.) TO FIND THE VALUES OE X AND y, WHEN

x3 + y3 — Saxy = max. or min.
In the first case let x = a constant quantity and find the
value of y which will make x3 + y3 — Saxy = max. = p . * .
?/3 — 3£m/ = _/? — x3 = r, r . y3 — Saxy — r = 0. Now let
5 = one of the negative roots of this equation, and .' . y + b
must exactly divide it.
y + bj y3 ~ Saxy — r = 0 ^?/2 - by + b2 — Sax = 0, (A.)
y3 + by2
— by2 — Saxy
— by2 — Z>2y
(Z>2 — Sax) y — r
(b2 - Sax) y -f b (b2 - Sax) .'.b(b2 - Sax)
v
\—
L — r .'. b2 — 3a^ = &r
( 155 )
T
.' . Sax — b2 = — and 3abx — bB = r. Now equation (A)
7 r b , A /b*~+~4r b
gives y- -^ by ^ —
j ; •.-.• .y^ - 2 — y 4fi - 2
v^ 46
-. Here it is evident that if 4r -j- 4i3 be
a negative quantity, there shall never be a maximum or a
minimum, and if 4r + 463 be positive, we shall have a
minimum; for in this ease we cannot suppose r so small or
negatively so large as to make 4r -f 4^3 less than 3b3 which is
negative, .*. when r = min. we mnst have 4r + 4Z>3 = 3b3 or
I2abx — 4b3 + 453 = 12«&z? = 363 or b = 2-\/W and y = —
= wax. When ^/ is considered as constant, we can show,
exactly in the manner above stated, that x — w ay, .' . x2 =
a?/ = aw ax, .*. xAt-=.azx) .'. x3=a3, .'. x-=.ay and ^ = vaa; =
\/a2 = a.
v b
In the equation y2 — by = -j-, let y = z -{- — and there
of2 62 £2 r
fore y2 — by — z2 \- bz -\— bz — — = z2 - = —
* " 4 2 4 6
b3
.\ r = bz2 —. Here it is evident that r becomes a mini-
4
mum by being negatively large (for r = p —■ x3) and . * . when
b3
r = min. we must have z = 0, .\ r — r = 3«&i? — b3, .*.
4
— 63 = \2abx — 4S3 .*. 3&3 = 12&&s .'. b = 2\/ax as before.
PllOB. (4.) TO FIND THE VALUES OF X AND y SUCH THAT

a;2?/3 [a — x — y) = max.
First let # be considered as variable and y as constant, . * .
x3[a — x •— ?/) = [a — y) x3 — x4 = max. or ^3 — «z4 =
( 156 )
max. where a — y — A. Now proceeding exactly as in
Prob. (4), chapter 3rd, we find a? = — = ' (1)

Now let x be constant and y be variable, and dividing the
given expression by xs we find y2{a — x — y) = (a — x)
y2 — y3 — By2 — y3 = max. where a — x = B. Now pro
ceeding exactly as in Prob. (4), chapter 2nd, we find y =
2B 2 ( 2 / 3 ^
x) = -T(a- -y! from equation (i)
3 »}
2 la 2« , 6y
y ■
3 It _ 12
"•" 12 ~6
+ y2'" . •£/ = and
3 / 3( 3 2a
X : -y) = X
3/ 4 "3 ~"
The same may be solved without impossible roots as pro
blems in the preceding chapters.
PROB. (5.) GIVEN THE PERIMETER OF A TRIANGLE ABC,

SHOW THAT ITS AREA IS THE GREATEST, WHEN IT IS
EQUILATERAL.
Let 2p = the perimeter of the triangle required, AC = x,

AB = y and . \ CB = 2p — x — y, and consequently, by
what is shown in the introductory chapter, the area of this
triangle = Vp(p — x) {p — y) {x + y — p) = max. and . * .
p(p _ x) (p — y) (x + y — p) = max. Now when x is con
stant, we find (p — y) {x + y — p) = max. . • . — p2 + px +
2py — xy — y2 — max. and since x and p are constants, we
find 2py — xy — y2 = max. = r, .• . y2 + xy — 2py = — r. . • .
y2 — (2p — %) y = — r, and solving this quadratic we find
vu = 2p
JL 2- x =fc AAV//(2p
— - x)2 - 4r
— and, . * . when
, , or r =
4 4r
max. we must have (2p — a;)2 = 4r, and .\ y = —-— , (A.)

( 157 )
Now suppose y to be constant . • . (p — x) (x + y — p) =
max. or — p2 + 2px -\- py — x2— xy = max. and since y and p
are constants, we find 2px — x2 — xy = max. = r .'. x2 —
(2p — y)% ■= — r. Solving this quadratic we find a; = -—-
2
a/±Jl 41
4
and here it is evident that when r or
4r = max. then (2p — y)2 = 4r, .*. a; = (B.)
Comparing equations (A) and (B) we find y = •—-— =
2 -
- 4 •> ' ' • ' 4y
^ = 2j»
^ +' y,
*' .-.• yy = -^jp,
3 -'.
2 -^
# = — - = == — ». and the third side =
2 2 3 ^'
2 2 2
2p — x — y — 2p — — p — -~-p — -^- p, and hence it ap-
o o o
pears that the triangle required must be equilateral.
The same may be solved without impossible roots, as pro
blems in the preceding chapters.
PROB. (6.) GIVEN THE SURFACE OF A RECTANGULAR PARAL-

LELOPIPEDON, FIND WHEN THE CONTENT IS A MAXIMUM.
Let x} y3 and z = length, breadth, and thickness of the

parallelopipedon and 2a = its surface. Now it is evident
that the whole surface given must be = 2ocy + 2ocz + 2yz =
2a or wyJ + ocz -f uyz = a,' and . ' . z = w + y and the con-
tent = — ———— = max. When x = constant and

x +y
y = variable . *. y x —- == max (A.)
cc "T" y
{ 158 )
and when y = constant and x — variable then x x —

x + y
= max . (B.)
Equation (A) gives —— — = max. — r,.'.ay — xy2= rx +
„ a —r a—r , / (a — r)2 — 4x2r
and hence it is evident that as r is greater, so {a — r)2 be

comes less, and 4a?V greater, and .'. when r = max. we must
have (a — r)2 == 4a?V or a2 — 2ar -\- r2 — 4#V, and . •. r2 —
2 (« + 2x2) r = — a2, . • . r = a + 2#2 dt 2<z,a/« -f #2, .'. y =
a — r
\/a -f <^2 (C). When y = constant,
2x
then from equation (B), —-—— = max. = r ; and exactly
as above we find x — — y — v a f y2 (D.)

From equation (C), ?/ -|- a? = — s/a -\- x2, r . y2 + 2xy + x2 =
a + ^2, . * . <2? = ———^-9 and from equation (D) we find ———
— - y — a/« + #2, .'. 2 = — Va + y2, and .'.

a* + 2$ + yi = a + fora-^l=l,.,a + f = W.,
a
2 a fa a — y2 3 /a
2.. 3
^
«
.. a — xy
A /a 3. /a
and * = t+7 = —75 = V 3' •"• *=*=*= V g
2V3
and .• . #y -f ^2/ + 2/^ = x2 + x2 + x2 = 3<#2 = 3 x — = « ;
o
hence it appears that the required parallelopipedon is a cube.
( 159 )
PROB. (7.) INSCRIBE THE GREATEST TRIANGLE ABC WITHIN

A GIVEN CIRCLE. (Fig. 64.)
Let R = radius of the given circle, a} b, c> the unknown

sides of the triangle required, n = Z. B, m = z_ C. . • . area
oi„ ,,
the xtriangle
. . = BC x—■AD = BC.AB.AC
-^ -, ,because ,by
6 B. Euc. we have 2R x AD = AB.AC. . • . 2Ec sin. n ^ cb
sin w?
,\ b = 2R sin. ft ,\ c = b —r^— = 2R sin. m and # = 2R
sin. ft
sin. ^4 = 21? sin. (m + ft) .*. area required = ■— = 2R2
sin. ft sin.m x sin. (ft + m) = max. .\ sin. ft sin.m sin. (m -f ft)

= max. Now let sin. n = a? .*. cos ft = v 1 — #2, sin. m =
y . * . cosm = v 1 — £/2 . • . xy [x\/\ — «/2 + y \/l — x1) = max.
First suppose that # = constant, . • . y[xv\ — ?/2 + y v 1 — #2)
/——2 , yVi — X? t x v'l — ^2 .
or vr — */ + = max. Let = A .*.
a; a; .
\/y2 — y4' + -4?/2 = max. = r .*. y2 — y* = r2 — 2Ary2 +
,9 4 . 4 _ 2^r + 1 2 _ __ r2 . 9 __ 2Ar + 1 _^
ATf ' -y JJ+I y ~ A2 + 1 " • y ~ 2(^42 + 1) —
v /, .„— ' and here it is evident that r cannot be taken

4(^42+l)2
so great, as becoming greater than A may make 4>r(A -— r)
= a negative quantity greater than one, for in this case the
root becomes impossible, and therefore when r = max. we
must have 4>r(A — r) = — 1 .*. r2 — ^4r = — . \ r =
^ + yîff+T , 2 _ 2^r + 1 _^-f U ^V^2 + 1

2 y ~2(A* + 1) "" 2(^* + l)
,but ^ = 1-■ ■ —2 re2
■■ ,\ ^ = 1 + a/1 — x% or sm/ m =
1 + COS ft ...
2~~ (A°
In exactly the same manner as shown above we may find,
( 160 )
. 9 1 + cos m m.
sm/ n = - ( tJ.j
2
when y = constant and
vT r— is supposed = A. From (B)
we find
1 + cosm 1 — cosm
1 — sm/n = co^n = 1 — = ~ > • •
A <&
but equation (A) gives cos2» = (1 — 2sin.2m)2 =
1 — cos m or cos2„ 2m
rt = 1 — ^cos m .*. cosm =1, — 2n cos220m
2m
2 <»
= — (2 cos2 2m — 1) = — cos 4m . * . cos m = — cos 4m.
Hence it appears that m is such an angle that its cosine is
equal to the negative cosine of its quadruple .\ m = 60°
"XT * X- /TIN ' 2 T + COS60° ! + 4
Now from equation (JB) sin/ m = = —~— =
o \/'"s\
— .'. sin. n = —p—- .*. % also = 60° .*. the third angle =
4 A
A = 180 — 60° — 60° = 60° .-. the triangle required is
equiangular and equilateral. One of its sides = a = 2R
sin. A = 2R x ~~- = 12 v7 3 = 6 = c.
2
PROB. (8.) TO FIND THAT POINT WITHIN A GIVEN TRIANGLE,

FROM WHICH IF LINES BE DRAWN TO THE ANGULAR
POINTS, THE SUM OF THEIR SQUARES SHALL BE A MINI
MUM. (Fig. 65.)
Let ABC be the given triangle, and let BD = a, AC = b}

AD = c, AE = %, EG = y where G is the point required,
.-. DE = oc — c3 FB -= a — y, and EC ~ b — x, and there
fore AG2+ CG* + GjB2 = #2 + 2/2 + (J - xf + if+{x- cf
+ (a - ?/)2= 3a?2 + 3y2- 2(6 + c) w - 2ay + a2 + 62 + c2
2 9 2(6 + c) 2a «* + 62 + c*
= max. .-. ^2 + y- ——'- % - — y + — =
f 161 )
max. = r. First let y = constant and oc variable, x2
2(b + c) a? ~ r
-^-g—^ a2 + b2 + c2 y2 4- 2a
y y and, .• . at
b + c^ / _ a2 + 62 + c2 , 62 + 26c 4- c2 ^ , 2a
3 V 3 +' 9s y2+^-y
J 3
6 + c , A /""" 2fo - 3a2 - 2b2 - 2c2 9 , 2a
—-±/\/r
- 'V ' + 9 r + T*
By inspecting the diagram it is manifest that 2b2 ~7 2bc
.\ 2bc — 262 = a negative quantity = — w, .*. we find
* = b-i~
+ c _,
~ V/r » + 3a2 + 2c2
^" + "8
2o~
9 V
b -\- c / n 2c2 Sa2 9 2a
= -3~~ Vr- 9~ - T " 9" - ^ + "3 y'
Now we say that 3a2 2a y\ if it is not so, 1st, let
— + y2 is "7 -$■
9 o
3a2 . 2a , 2a 3o2 a=»=\/-3a2
T + f = ^y,.-.r--Iy = -T,---y = 8 =
3a2 2a
an imaginary quantity; 2nd, let — 4- y2 Z. — ?/, and .'. let
3a2 0 T> 2a 9 2a 3a2 _,
- + y» + P = -y, ^y»-,-yss---Pf ,-.y =
ad=V/-2a2-9P . . .... TT 3a2

= an imaginary quantity. Hence —
o y
— y2 + 2a
-77 2/ — a negative quantity = — m, suppose \ . * . x =
3
g =fc A/ r — — m, and .' . when r = min. then
o * y y
n 2C2 t 5 4" £ ,,v -nrn
r = — 4- -TT- 4- w* and % = —-— (1) . When a? = a
9 9 o
constant, then from the original equation we find
9 2a a2 4- b2 + c2 s 2(6 + c) ,
y "8" V = r 3 8 * an aS
above it may be shown that when r = min. we must have
r _ 2a2 + 362
_. + 3c2
_ +, a»2 - 2{b
_L_—4- c) X} ... y = _
a ... (2.)
,

Y
( 162 )
PROB. (9.) TO FIND A POINT WITHIN A TRIANGULAR PYRA

MID, FROM WHICH IF LINES BE DRAWN TO THE ANGULAR
POINTS, THE SUM OF THEIR SQUARES IS THE LEAST POSSI
BLE. (Fig. 66.)
Let ACEB be the given pyramid, ABC its given base,

EG — a = perpendicular drawn from the vertex to the base
of the pyramid. Let K be the point required and the per
pendicular drawn from this point to the base = KH and HD
= perpendicular from H to AC = y, and AD = x, GF =
perpendicular from G to AC = b, and AF = c. Let Hn be
drawn parallel to DF, . * . Hn = DF ~ c — x and Gn = GF
- HD = b - y, .-. JIG2 = (c - a?)2 + {b - ?/)2. Also let
AC — d} .*. DC = d — a?.- Draw ii7 parallel to iJG and
. * . HG2 = if/2. Join if, 5 and J9, i7, and now it is evident
the jL KHB = a right angle. By a process exactly similar
to that used in the foregoing proposition, it may be shown
that HB2 = (d — x)2 + (e — y)% where e is the altitude of
the triangular base of the pyramid. Let KH = z . * . .KB2 =
(rf _ xf + (e - y)a + *2 (1.)
It is manifest that AH2 + KH2 = AD2 + flD8 + iffl2 =
x2 + y2 + z2 , (A.)
CK2 = CH2 + ifiJ2 = CD8 + HD8 + KH* = (rf - ^)2 +
y2 + *a = d2- 2& + a?2 + y2 + s2 (B.)
KE2 = iO2 + ffi2 = if/2 + (EG - KH)2
= {c - x)2 + (b - y)2 + (« - *)2
= c2 — 2ca? + #2 — 2by + a2 — 2az + x2 + y2 + z2
= a2 + 62 + c2- 2cx - 2by - 2az + x2+ y2 + *2. . . (C.)
From equation (1) we find
jKB2 = d2 + e2 - 2<fo - 2ey + a?2 + y2 + z2 (D.)
Adding together these four equations we find
AK2 + BK2 + CK2 + ##2 =
4x2 + 4y2 -f 4s2 + a2 + i2 + c2 + 2d2 + e2 - 2^ - 2% -
2«# — Mx — 2e?/ =
( 163 )
.wo.a,^ c + 2d b + e a
a2 + 62 4- c2 + 2^2 + e^
>•
4
JNow as = a constant . \ when z3 or y,
4
or #, are supposed to be constants respectively, we shall have
severally the following three equations, the second members of
which must be such as to become negative when the orignal
minimum quantities are taken very small, for these second
members are nothing more than the difference of the mini
mum quantities supposed and constant quantities taken to
the other sides of the equations.
"9 x — min. = r when y and z are constants,
2
b + e
y = min. = r when w and z are constants,
~2
a
z = min. = r when oo and y are constants, and
from these equations we find
and #=—=!= A / ——- + r, and here it is evident that r

4 V 16
cannot be taken so small or negatively so large, as to make
the roots impossible, and therefore when r = min. we must
(c + 2d)2 + r = 0, (b
have i—ttt— + cf
v 7. } + r = 0, and —a* 4- r = n0,
16 16 16
c + 2d b + c , a
and .*. a? = —-.—
4 ,' yy = —-—
4 and z — — 4.
*#* The symbol r is used in three different senses.—Ed.

( 164 )
PROB. (10.) TO FIND VALUES OF X AND y SUCH AS WILL

MAKE (x + 1) (y -f 1) (z + 1) = MAX (1) WHERE
axb*cz = A (2).
Taking logarithms of the equation (2) we find

x log a + y log b + z log c = log A and let log a = p, log 6
= m, log c = ft and log A = q (3)
q — px — my _
r. px -\- my + nz — q, . \ # = - ,\ # + l =
- ———- ; substituting this value of z + 1 in (1)

ft °
we find
{x + 1) (y + 1) ^g W ~~ ^ "" m^ ' = max. or (x +1)
ft
(y + 1) {q + n — pw ■— my) = max. Now when ^ + 1 =
constant, we have (y + 1) (g + ft — px — my) = (q + ft)
«/ — pxy — my2 — my — px -\~ q + n = max. . * .
, + ft — m — jm?)x y — myz„ —- jpa? + g + ft = <((q + n—m—px)
(q
y —. y*„ — ■£px
px — aq1 — ft)
n\j> m = max. . q + n — m — px
m
px — a — ft ^T px — q ■— n
i/J — uy2 — m = max. JSTow as m =
, x t q + n — m — px 2
constant, we have — y — y = max. = r, .*.
m
y2 — ^~~ — y = — r. Solving this quadratic we
m
,. i q + n — m — px, A /(q + n—m—px)2
here it is evident that when r = max. we must have

(q + n-m- px)2 g + ft - m - px
4m2 -r, . .y- 2m W
Now let y = constant .'. (x + 1) (q + n — px — my) =
max. .'. (g + ft — my) x — jpa?2 — my + q + n — ^ =
f<7 Jr ft — p — my
(g + ft —^ — my) x ■— j?<z?2 + q + ft — my = <
( 165 )
x — a?20 -h q + n — m?/)
-V x p = max. .*. q + n — p — my
^-J
P f P
x — xl„ + q- + n — my = max.. and_ as q + n — my =
P P
xx we mustxthave q + n — p — my~ w
constant, — xl9 — max. = r
P
and therefore «#2 — — x = — r ; solving this
q+n—p—my /(q + n—p—my)2

quadratic we find x-=? ' '" 0^—— =!= A/ "^"'^ '"^ ~~r
~2p ~" v 4$f
i xi q + n — p — my ,„ .
.', when r = max. then a? = ~
2p ~ (5.)
v '
,\px == ~ pp- -. Substituting this value of px

q + n — p — my
q + n — m — ~ ~
in (4)
w we find uy = 2m =
q + n — 2m + p + my # __ g + w -j- jp — 2m , .
4m " " # y "" 3^ * ^ *'
Substituting the values of g, n, p, m from equations (3) we
fi , __ log ^4 + log c + log a — 2 log Z> _____ log (J!ac) —2 log 6
y 3 log 6 3 log b
• • y + * = -3i^6~ (7-}
Substituting the value of y from equation (6) in (5) we
find ff + n + j? -- 2m
^ q + n-p 3 ~_2q + 2n-4p + 2m
X ~~ 2p ~~ 2 x -3p
= £ ^1# Now substituting the values of q, n, m3

to /„x n -, log ^ + log c + log 6 — 2 log a
and *» from (3)
N we find x = —5 —-.
3 log #-— —
= log(^6c) ~- 2 log a . ^ + x = log {Aabc) ^ ^ ^
3 log a 3 log a
Now from equation x log a -\- y log S -I- # log c = log ^ we
find # log £ = log A — x log a — y log Z>
( 166 )
log (Abe) — 2 log a log (Aac)— 2 log b
= log A
3 3
— ^^Qg^-lQg^-lQg(^)+^^Qg^-lQg^-lQg(^)+2log6
~ 3
__ log A — log (5c) — log _(ac) + 2 log a 4- 2 log b
=
__ log ^4 — log 5 — log c — log_a — log c + 2 log a + 2 log 5
=
__ log ^4 + log 5 +_ log a — 2 log c __ log (^4«5)_— 2 log_c
__ _
"* + T = 1^llof? •'• We find ^ + 1} (y + 1} (* + 1}

{log (Aabc)Y
= max. =
27 log a log 5 log c
PROB. (11.) TO INSCRIBE A TRIANGLE WITHIN A GIVEN

CIRCLE SO THAT ITS PERIMETER MAY BE A MAXIMUM.
(Fig. 67.)
Let ABO be the triangle required. The centre of the given
circle is E, and ED, EF} EG perpendiculars let fall from the
centre on the sides of the triangle. Let the z. AEC = 20
.-. each of the angles AED, and CED = 0. Likewise AEF
= FEB = $ and .-. the z. BEG = 360° - 20 - 20 and
.-. BEG = GEO = 36° - ^ - 2» = 180 -(» + *) and
sin. BEG = sin. (0 + 0). Also let the radius of the given
circle = — . Now it is evident that AD — — sin. 0 .\ AC
Z &
= 2^4D = asin.0, and in like manner AB = & sin.0, and jBC
= « sin(0 + </>) .'. perimeter == «{sin0 + sine/; + sin(0 + 0) }
( 167 )
:= max. .'. sin. 0 + sin. <£ + sin. (0 + 0) = max. Now let
sin. 0 = constant .*. sin $ -f sin. 0 cos. 0 + sin. $ cos. 0 =
(1 -f- cos. 0) sin. 0 + sin. 0 cos. <£ = max. Let 1 4- cos. 0
= ^ sin. 0 = x, sin. 0 = c, .*. cos. 0 = vl — ^2 .•.
fta? + cyl — oc2 = max. = r, .*. c2 — A2 = r2 — 2ftr<z>
+ A2 .*. (c2 4- ^2) a?2 — 2^ra = c2 — r2, .*. #2 •—
_ oo = -^ 2 5 solving this quadratic we find oc =

C "T~ ft £ 4~ ft
ftr 4- c\/c2 4- ^2 — r2 # • # when

1 r = max. we must, ,nave *r0 4-,
c2 4- ft
a / o . o w ft
^2 = r2 . * . r = \/c2 4- ft2, . * . oo =
c2 4- ft2 V'c2 + n2
1 4- cos. 0 1 4- cos. 0
a/1 + 2cos.0 4- cos.20 4- sin.20 \/2(l 4- cos. 0)
VI 4- -cos. 0, .*. a? = sm.
. 0 = A//l 4- cos. 0 . In like
manner when sin. <p = constant, we may easily find sin. 9 =
\f * N°w let sin. 0 = y .* . cos.0 = a/1 — y2

and we have supposed sin.<£ = w . •. cos.$ = vl —- oc2 .• . w2=
1 4- \/r^"2*
- £- and, ?/2 = —
14- aA - ^3 and, . ■ . 4a?4
, 4 — 4a?2
. 9 4-, 11 =
2^2
1 -y2 .\y2 = 4x2(l ~ w2) (2.)
Also 4#4 - 4y2 4- 1 = 1 - oe2 .-. 4a?2 = 16/ - 16*/4;
substituting this value of 4a?2 and 1 — a?2 in equation (2) we
find y2 = (16y2 - 16z/4) (%4 - %2 4- 1) = - 6%8 4-
5 15
128?/6 - 80z/4 4- I63/2 and / - 2#4 4- -j- y2 - wi = 0.
5 z — 15
Now let y2 = z, .\ zs — 2z2 + -— —= 0. This equa-
tion is exactly divisible by z — -7- as may appear by actual

o /~~3
division .'. — = a value of z = y2 .*. y = \/ ~k an(^
( 168 )
a* = ——£ 2- = -^ = -^ and a? = /\f — or sm. 0
= /\/ — and sin 0 = /y — .*. 0 = 0 = 60° and hence it

appears that the triangle required is equiangular, and the
sides = —-— each where a = radius.

PROB. (12.) TO FIND SUCH VALUES OF X, y, Z AS WILL
xyz
MAKE ; — ; ;~ ; 7-7 ; r = MAX.
(x + a) (x + y) {y + z) (z + e)
First let y and z = constant quantities . . . • . -(—■—r—■
x —■—r
^ ^ (x + a) \x + y)
(a? + a) (# + ?/) x2 + (a + y) x + ay
= max. .-. -—:—-- — = — = mm.
x x
= r, .' . x2 — (r — a — y) x = — ay. Solving this quadratic
H* si y f (v a ii)
we find x = — —£ =*= A/ - -j —— ay, and here
\7* —~~ a ——" if\
it is evident that when r = min. then ——— = ay .'.
r — g—-
a — y = vay,
/— .'.x = r — a—^
— y = A/ay
/— (1.)
,.. N
Secondly when x and # = constants we find likewise

y = \/xz (2.)
Thirdly when x and y = constants we find # = Vye... (3.)
?/4
From (1) and (2) we find x2 = ay and #2 = ~ and «y =
?/4
~, y3 . • . y3~az2 and from (3) we find —
. • . <z = ^ z2 = y, .' . y3=
Z Z c?
— = az\ .*. £4 = ae3, .• , z = vae3 and y = — = =

( 169 )
\/ae = vê2, x = \/ay, .• . a?4 = #2?/2 = a2 x «e = a3e, .* . x =
y«3e. Hence it appears that a?, ?/ and z are in geometrical
progression and the common ratio is A/ —.

PROB. (13.) IF THE CONTENT OF A RECTANGULAR PARALLE-

LOPIPEDON RE GIVEN, FIND ITS FORM WHEN THE SURFACE
IS A MINIMUM.
Let the content of the parallelopipedon = a = xyz .*.

z = — : and it is evident that half its surface must be
xy
a a x2y2 + ax + ay
= xy + xz -f- yz — mm. or xy H 1 = —-
y x ocy
_. , . , , , w2y2 -\- ax + ay
= mm. First let y = constant .\ — - =
y2[x2 + —5 + — ) ar + -5 + —
47 V *iy2 \L—
y 1 = mm.
. ,\ y y— = mm. = r} .*.
/ a\ a a _. I \ y2> a
\ 2/z/
y2/ y
v
.*. when r = min. we must have Vv-—

v, _ —. Likewise when a? = constant and r = min. we find,
uy — \f
y —
x and z = xy
— = a = v xyu ': therefore x2 = —
y
V xy
1 o a a2 0 a2 a a2 _. a
^ a? y ^2 y y y y
( 170 )
or ii = a% . ' . x2 — — = <$ . \ x ■=■ oh and # = v a?w = vaw
a*
= \Za* = ai Hence it appears that the parallelopipedon
is a cube.
PROB. (14.) TO FIND A POINT P WITHIN A QUADRILATERAL

FIGURE ABCD, FROM WHICH IF LINES BE DRAWN TO THE
ANGULAR POINTS, THE SUM OF THEIR SQUARES SHALL BE
THE LEAST POSSIBLE. (Fig. 68.)
Let AB = b} AB = a, BC = c. From the points D,

(7 and P draw straight lines perpendicular to the base or the
base produced of the given quadrilateral .*. FB = ft sin. A,
FA = 4 cos. -4, GC = c sin. B, BG ^ c cos. 5. Draw ^JP^"
parallel to AB and let AN = x, NP = y .' . EP = FN =
îV + ^4P = ^ + 4 cos. ^ EB = BF - EF z= BF -
PN =, b sin. ^ - y. PH = NG = NB + BG = a - x +
c cos. B, Cff = GC - HG = GC - PN = c sin. J3 - y; we
therefore find, ^P2 = x2 -V y2 (1.)
PB2 = (a - x)2 + y2 (2.)
PC2 = (a - a? + c cos. £)2 + (c sin. B - y)2 (3.)
DP2 = (a? + ft cos. ^4) 2 + (Ssin..4 - y)2 (4.)
Adding these four equations we find ;
AP2 + PB2 + PC2 + DP2 = 2?/2 + ^2 + (« - ®Y +
(a — # + ccos.B)2 -f (csin. J3 — y)2 + (x + bcos.A)2 -f
(ft sin. A — y)2 = min.
First let ?/ = constant and a? = variable, .*. x2 +
{a — x)2 -f (a ■— a? +'c cos. J3)2 + (a? + ft cos. ^4)2 = min.
or 4a?2 — 2 (2a — ft cos. ^4 + c cos. B) x + 2a2 + b2 cos2A
+ 2ac cos. JB ^ = 4 (a?2
/ 0 2a — b cos. ^—EA + c cos. i? # +
2a2 +
Z ft2 cos.2^ + 2ac cos. B\J = 4(^2
^L—. . . 9 __ jioV
D +, Q)
r>\ — mme
( 171 )
R =fc A//BF
. • . x2 — Rx -f Q = min. = r, . * . x = — -j- + y — Q.
Here it is evident that r cannot be taken so small as to

R2
make r — Q a negative quantity greater than —; and .#.
when r = min. we must have —-- = Q —■ r, .•.£? = — =
4 2
2a — h cos. A ~\- c eos. i?
4 ~-"•
Secondly let 3? = constant,, we find 4?/2 -— 2(^sin. ^4 -f
c sin. 5) y -\- b2 sin.2J( + c2 sin.2JB = min. and proceeding
exactly in the manner as shown in the case of y being a con-
„ _ b sin. A + c cos. 5
stant we nnd y = ~ •
The same may be solved without impossible roots.
prob. (15.) let u — ax + by + c#, a maximum and

^2 + J/2 + £2 — 1; find #, y; and z .* . u — ax -i- #y -f
c\/l — x2 — y2 — max.
First let ?/ = constant .*. ax + c\/l — ^2 — y2 — a(x +
—V
a 1 — a?2 — Jy2)J r= max. . * . a; -+ a vl — a;2 — Jy2 = max. = r
-r ^r-—o ?r = ? — 2rx -f ar and . • . -— xl — 2rx =

&z *r az,/ a2
c2 c2 9 _ _ p _ 2#2r c2 — c2y2 — a2r2
-5 9V r" and therefore ar 5 a? = 9——5
a2 cr*7 tt2+ ^ cr + c2
- a*r -»- a / (g2 - g2ya) (fl2 4- g2) -f flV* ~ a2 (a2 + c2)r2
r'X~ a* + c*~ V (a8 + c2)8
and .'. when r = maxv then (c2 — c2?/2) (a2 -f c2) = a2cV2
and r= a/^Ê±Z) and ... „, ^/EEMEM
37 = tfT*
( 172 )
= V^| • (!•)
Secondly, when x = constant, proceeding as above and
putting b instead of a and x instead of y we find y =
VI b% + c2
/?2 _— flf'lp'
(2.)
Squaring equations (1) and (2) we find x2 = —g f- .*.

$ "T" C
» = a?-
" (a2 2+ c>2 and. .'. ya9 = -Trr"—
b2- JV = a2 - ^—
(a2 + c2)^2
a2, a b2 + c2 ar
= 1 - ^ + ^ . (fl2 + c2)^2 _ _ b2 - 6V _ c'jf 6V2
O • • O "~~ •*■ 7 O . O """■ 7 9
a2 ' ' a2 62 + c2 62 + c2
c2 ft2 9 /a2 + c2 b2 *2
#* + ^2 ~T' £2
79. +, C2 a2.9
- '•"•• VI fl8
.9.W 62
19 + cV
_9.) '^'
. C2/ 62 + C2
a262 + aV + 62c2 + c4 - a262 9 c2 iY>" • W" ——
a\b2 + c2) b2 + c2
a2 . • . x = /■ « and . e . yl0 £2 ■— Z>2#2
/ 9 . 79 . 9 ^
/»2/»2
S2-
«2 + £2 + C2 £4 + b2c2 ft
b2 + c2 "~ (62 + c2) (a2 + 62 + c2)
a2 + b2 + c2
b a2
.'. x = . — and ^2 = 1 — x2 — v2 = 1 »—79-:—»
vV+62+c2 a2 + 62+c2
a2 + b2 + c2 a2 + b2 + c2 .'.
• 'z v'rfT^T^"
The same may easily be solved without impossible roots0

( 178
PROB. (16.) FIND THAT POINT WITHIN A GIVEN TRIANGLE,

FROM WHICH IF LINES BE DRAWN TO THE ANGULAR
POINTS, THE SUM OF THEIR SQUARES SHALL BE A MINI
MUM. (Fig. 69.)
This Problem is a more elegant solution of Prob. (8.)

Let ABC be the triangle, and P a point within it; a, b, c
the sides of the triangle. Draw PN, AD perpendicular to
the base; join AP} BP, CP. Let CN = x ; NP = y, then
AD = b sin. C; CD = b cos. C. Then CP2 = x2 + y2, BP2
=:y*+ (a- x)2 = y2+ x2+ a2- 2ax, AP2 = (6cos.<7- x)2
+ (A sin. (7- ?/)2 = b2 + ^2 + y2 - 25(<2?cos.C + ysin.C);
3<2?2 + 3?/2 4- a2 + b2 - 2aa? — 2b (x cos. C + y sin. (7) =
mm. .• . x20 + y20 H a2 +- b2 2ax
-—- 2b-q- ((# cos. n
C +t y sm.
. u)
~x
a o o
= min.
„. 0 a2 + 62 2a + 25cos.C
First let ?/ = constant .*. ar ^ -— 5
o o
2« + 25 cos. C a2 -{- b2
x = mm. = r .'. or ~ a? = r ^r—
a + 5cos.C / a2 + b2 , (a + 6cos.C)2
and a? = g sfc /\J r ^— + g
a + 5 cos. (7
= 8 * V/r + 20*008.0+ A2cos.2C- 2a2 - 3^
9 •
It is evident that if a -7 b, then a "7 b cos. C . • . 2a2 -7 2«#
cos. (7 .-. 2«6 cos. C — 2a2 is negative and b2 cos.2 C is evi-
, , 079 2«6 cos. (7 + b2 cos.2 (7 - 2a2 - 3Z>2 .
dently Z. Sb2 ,\ is nega
tive = — P .". x = J (a + 6 cos. (7) =fc -\/V — P .'. when

r = min. then r = P .\ a? = J (« + # cos. (7).
Secondly when x = constant, then we shall have
0 a2 + 62 25 sin. (7
^2 + —5— 5— y = min' = r •*•
( 174 )
h sin.C -4-
h2 sin. 2 (7
t
— a2 + &3
y 3 9
f h2 sin.2C- 362 a2
= —— ±yr+1 § -g-. Here it is
52 sin 2C — 3b2 &2
evident that '—- = —- (sin.2 C — 3) — a nega-
9 9
tive quantity which let = — Q .' . y = —^— =*= vr — Q^
i i ^ 6 sin. (7
.\ when r = mm. we must have r = Q .\ y = —-— .
o
For ~Q read -Q-^-—Ed.
PROB. (17.) TO FIND A POINT WITHIN A GIVEN TRIANGLE,

FROM WHICH IF PERPENDICULARS BE LET FALL UPON THE
SIDES, THE SUM OF THEIR SQUARES SHALL BE A MINIMUM.
(Pig. 70.)
Let ABC be the triangle as before, P the point within it,
draw PN, PM, PQ respectively perpendicular to CB, CA
AB. Let CN = x; NP = y, PM = p, PQ = q, CB =
a, CA — by AB = c . \ u ~ y2 -}- p2 -}- q2. ' Now it is evi
dent that p2 = 7¥P2 = FP2 x cos.2 MPF = FP2 cos.2
C = {FN - PiV)2 cos.2 C = {x tan. C - y)a cos.2 C =
i(a? tan. C — if)2 = 11

——-dL.
sec.2 (7
ru — <2?tan. C\2 = ^
V sec. C /
, „ . .yNO
w cos. c - x sin. C)2
J
Also g2 = PQ2 = PE2 cos2 EPQ = (.EN - PN)2 cos 2B =
/?/ — (« — ^?) tan,I2\2 r ,_, . . . _ .
r sec.i? —) = {y cos-B - (a ~ *) Mn-B> •'•
u = ?/2 + (2/ cos. C — x sin. C7)2 + {^ cos.i> —- [a — a?) sin. J3}2
= min. or y2 + y* cos.2 C - 2^ cos. C sin. (7 + if cos2 B
+ #2sin.2C — % (« — x) cos. 5 sin. If -f (« — x)2 sin2 B
= min,
( 175 )
First let oc ~ constant .'.if + f cos.2 C ■— 2^?/ cos. C
sin. (7 + ?/2 cos.2 5 — 2y {a — os) cos, JS sin. J3
= (1 + cos.2C + cos.2!?)?/2 - 2y{# cos. Csin. C + (a - a;) cos. B sin, i?}
/ {# cos. Csin. C + (a — z) eos.B sin.B}\ .
JV J i + cos.2C + eos.2i* /
{# cos. Csin. C + (# — a?) cos. I? sin. I?}

= mm. .*. *w- — 2 ^ + cos.2^
1 + cos.2C ^=z Jy =
min. = a negative quantity and .\ as in the foregoing pro-
,, „ . os cos. C sin. C + (# — a?) cos. 5 sin. B
blem we find y =
1 + cos.2C + cos.2.B
(cos. Csin. (7 — cos. B sin.i?) % -\- a cos. B sin. 5
(i)
1 -j- cos.2C + cos.2J3
Secondly let y = constant . • . — 2ocy cos. Csin. C + x2 sin.2C
-f- 2yx cos.-B sin.jB — 2aoc sin.2jB -f oo1 sin.2!?
(sin.2_B + sin.2 C) x1 — 2 (y cos. C sin. G—y cos.J9 sin. !5 + a sin.2!?).27=
/ • 971 , • orn\t 9 o 0/ cos- Csin. G—y cos. _Bsin.!?-j-«sm.2!?) \
v(sm.25 + sm.2C)yV( #2 — 2 ~ t~rr- + sm.2C
sin.2.B r—^ -# /
= mm. .*. 0Gl„ 2(2/cos.Csin.C—t/cos.J5sin.i?

~ = ; f=r :—r— + asin.2-B)
—- x =
sin, iJ-fsin/O
min. = a negative quantity, and . • . as in the foregoing problem,
a a _ 2/(C0S-Csm«C — cos. 5 sin. B) + a sin.2J3 , >
we iind. x — — ;—^== + sm.2C
sin.2JB ;—x-~ ...... v[&)J
Now let cos. Csin. C — cos.jB sin. B = P^

a cos. J3 sin. 5 = S
1 + cos.2C-j- cos.25 = T [ (3.)
a sin.2J3 = Q
Sin.2B + sin.2C= fl
.-.a? = -^—...(4.) and y = —^— or a? = Jp ...(5.)
Comparing equations (4) and (5) we find,

Jig _|_ pQ
2/ = -^7= ^ and substituting the values of R,$,P}Q,T
from equations (3) we find,

( 176 )
(sin.2-Z? + sin.2 C) a cos. B sin. B + (cos. C sin. C— cos, i? sin. B) a sin.2I?
(1 + cos.2j5 + cos.2 O) (sin.2 (7+ sin.2^) - (cos. Csin. C- cos. B sin. _E)2
# sin.2Csin.I> cos.jB 4- a sin.2jB cos. C sin. C
sin.2£ + sin.2C + sin.2C cos.2.B + cos.2<7 sin.2jB + 2cos.C
[sin. C cos. 5 sin.i?
a sin. B sin. C sin. (jB + C)
1 - cos.2£ + 1 - eos.2C + sin.2Ccos.aS + cos.2 C sin.2B
[+ 2cos.Csin.Ccos.i3 sin. 5
_ a sin. -4 sin. 5 sin. C -.
= 2(1 - cos.2jBcos.2C + cos. B cos. C sin. B sin. (7)' an
substituting the values of sines and cosines of A?B,C, in

terms of the sides of the given triangle we find-, y =
ahc sin.A abc sin.B , abc sin. C
P = „2 , ja ,-^2 and ? =
a* + 6* + c2 ' ' ^ ~ a2 + ft2 + c2 * "" a2 + b2 + c2'
PROB. (18.) TO FIND THE VALUES OF XflyZy THAT, (tdchfz^

— xzyzz* — x2y^z* — x2y3z5 may be = max.
First let x, y = constants and z = variable,

. * . ax2yzz^ — x3y3z* — x2y4z^ — x2y3z5
= x2y3{(a — x — y) z4' — z5} = max. and
.*. (« — a? — y) £4 — 05 = max. .'.by Prob. (10), chap. 3,
z = 4{a
-i — —■x —
—^ y) or 4a? + 4y + 5z = 4a (1)
D
Secondly let x, z = constants and ?/ = variable, then pro

ceeding as before we find (a — x — z) y3 — y* = max. and
•'• y = ~^ ~— ••• ^ 4- % + 3* = 3a (2.)
Thirdly let y, z = constants, and a? = variable, then as

before (a - y — z) x2 - <*?3 = max. .'.by Prob. (2), chap. 2,
( 177 )
x = 2(a~y-z) and .•.Sx + 2y + 2z = 2a (3.)
<J
Subtracting (3) from (2) we find 2y + z = a (4.)

Multiplying equations (1) and (3) by 3 and 4 respectively
we find, \2x + \2y + 15^ = 12a
12<# + 8y + 8z = 8a
4y -f 7z = 4a, and multiplying
(4) by 2 we find 4y + 2z = 2a
5z = 2a ,\ z = — ,\ 4y + — = 2a
o o
.-. ^ = JL and .-. 3a? + 2y + 2# = 3a? + -^ + -^ = 2>x +
J.U o o
2? - 2a • 2,x-2a- — -Za • * - —
The same may be solved without impossible roots.
2 A
SUPPLEMENT,
It will be observed throughout this work that a great

many equations of the second degree solved for finding out
the maximum value of r have been reduced to the form x2,
-f Ax = — r or x2 -j- Ax + r = 0, where A is generally
negative, and in like manner the cubic and biquadratic equa
tions have been reduced to the forms, x3 + Ax2, -f- Bx + r
= 0, x* + Ax3 + Bx2, + Cx -h r = 0, where the maximum
value of r is to be determined.
The object of this supplement is to solve these general
equations, and thus to find out general expressions which
may enable us to solve numerous problems of this book in
an instant, without going through long and sometimes
tedious operations.
We will also add in this part of the work a few interesting
problems which we have unfortunately forgotten to put in
their proper places.
1st. Solve the equation x2, + Ax -J- r = 0, where r =
max. A + \f/A*
We have x2, + Ax = — r .• . a? = — -^ -r- — r
Al1 A
.*. when r = max. we must have —4 = r .*. x = — —...
2 (A.)
v '
EX. 20# - x2 = max. = r .-. a?2 — 20^ + r = 0. Here
on
^ _ _ 20 .-. by (A), w= -^ = 10.
In like manner other examples of this kind may be solved
by means of (A).
2nd. Solve the cubic equation, xs + Ax2, + Bx -J- r = 0.
Let a negative root of this equation = a . • .
( 179 )
w+ a) x3+Ax*+Bx+r=0 [_fc2+(A-a)w+a?+B-aA=zO(l.)
x3^-ax2
(A-a)x* + Bx
(A — a) x2 + (aA — a2)x
{a2 + B — aA) x + r
(a2 + B - aA) x + a {a2 + B - aA)
a% + B — aA = — .\ Equa. (1) gives x2 + (A — a) x
r ___ ^4 — « 4. A /^ - «)' ^
Here it is evident that when r = max. we must have

^ 7 a)2 ==— = a»+JJ-flL4, .-. A2 - 2aA + «2 = 4a2
4 a
2.4
+ 4B-— 4cl4 or 3a2 — 2.4a = ^42 — 4jB or a — a ==
o
^2 - 4B A + \/4>A2- 12B . ^ - a
— , .-. a = g and x = ^— =
q - X __ V4A* -"125 - 2,4

(B.)
2 ~~ 6
Ex. (1) a?3 - a?2 + r = 0. Here -4 =■ — 1, JB = 0, .-. x
= 4-^ = |. Ex. (2) a?3 - 5? + r = 0, ^4 = 0, 5 = -

o o
1, ^ * = *LJL = -JL Ex. (3) *»-&?■- 15* + r = 0,
*> V 3
-4 = — 6, 22 = — 15, .•. by (B), x = - =5.
o
3rd. Solve the general equation of the fourth degree, viz.
a?4 + Ax3 + Bx2 + Cx + r = 0.
Let the product of the two values of this equation = a?
+ ax + ft, and we therefore find,
( 180 )
x2+ax+bj x4>+Ax3+Bx2+Cx+r=0 ^x2+(A-a)x+B+a2—
l>te-ft=0...(l.)
x* + ax3 + bx2
{A - a) xs + (B - ft) x2 + Cx
(A — a) x3 + (Aa — a2) x2 -f (^4ft — ab)x
(B+a2-Aa-b)x2+(C+ab-Ab)x + r
(B+a2-Aa-b)x2+(aB+a3-Aa2-ab)x
[+b{B + a2-Aa-b)
,-.B + a2- Aa- b = -^- (2.)

Also C + ab — Ab =: aB + a3 — Aa2 — ab, .-. b =
a£ + ^ - Aa2 - C /0 ,
~ 2^T2 ^
A — «
Now solving the equation (1) we find x = - 1-
a/ —;—
4
r ; and here it is evident that when r =
b
max. then ^7 ^ = -£- = 5 + a2 - ^« - ft, .-. (A - a)2
= 45 + 4a2 — 4.4a — 4ft, and from (3)
L4 — «)2 = 45 + 4a2 — 4â :—
2a — A
__ 4>aB + 4a3 - SAa2 - 4^5 + Â2a +.4C
"" 2a — ^4
or 4a£ + 4«3 - 8^4a2 - 4^LB + 4^42a + 4C = 4-<42a -
3^4
5a2^4 + 2a3 - ^43, and therefore a3 - — a2 + 2Ba - 2^45
+ 20 + ^ = 0 (C.)
Now it is evident that from this equation the value of a
jl a
may be determined, which, when put in x = -— =
a —- A
—-—, we will find out the value of x sought.
( 181 )
Ex. (1). ar* - xs + r _ 0. A = - 1, B = 0, C = 0,
«3 + -|«2 - -| = 0. Let a = -I, .-. «» + |- o« - 1
J_,
8 + 3__l_n
8 _a-^_|
2 -«, •••*- 3 - +2 l_
- 34-
Ex. (2). x* - x + r = 0, A == 0, B = 0, and C = - 1,

„~ ~i A — a a 2%
a3 — 2 — 0> .'. ffl = 2 and a; — -— = — = —
1 1
2* ~ ¥4'
Ex. (3). a* - 8x3 + 22a;2 - 24a; + r = 0. Here .4 = — 8,
B = 22, C = - 24, .-. a3 + 12a2 + 44a + 48 = 0. Let
a = - 4, .-. - 64 + 192 + 48 - 176 = - 48 + 48 = 0,
a —A a+8 —4+ 8
and x - __ = __ - _____ _ 2.
Ex. (4). To inscribe the greatest parabola in a given isos
celes triangle. (Fig. 62.)
Let AD = by GD = a> GP == x, .\ the area of the para-
bola = 77- via — x)sx = max. .*. (a —• #)3#? = a3x — 3a2a;2
3a '
+ Sax3 — a;4 = max. = r, . • . a;4 — 3a#3 -f SaV — a3# 4- r
= 0. Here A = - 3a, B = 3a2, C = — a3. Now substi
tuting these values of A, B} C} and putting y instead of a in
the equation (C) we find,
Vs + 9a 5a3 = 0. By trial the value of y is
"9*" 2/2 + âV + "9"
5a — _4 4- y
found = - —y .\x - ———£
o 5a
3a — —
2 a
2 4'
Ex. (5). In the trapezium ABCD, the base AB = a, _4D
= BC = 6, find CD, CD being parallel to AB, that the area
may be a maximum (m & n are the points where the perps.
cut the parallel line required and mn = x) .
( 182 )
It is evident that Am — nB .-. the area of the whole tra-
Dm x Am _ Cn x nB Dm x Am
pezium = ► -f ran x Dm -{ - = h
_ -h Dm x5- Am = Dm
^ x Am
A + mn x Dm
y^ =
+ mn x Dm
a —x
AA-rf-t+*AA-r-i-*)'
= m-¥ V>- - w- V* r-r-")'- (^
= max. = r, .-.«*- 2(262 + a2)*2 - 8«62a; - a2(452 - a2)
+ r = 0. Here ^ = 0, B = - 2(2^ + a2)3 C = - 8aP,
.-. y3 - 4(262 + a2) y- 16a62 =0. Let y = - 2a,
.-. _ 8a3 + 16a62 + 8a3 — 16a62 = 0, and therefore
,» - 4(2y + oF) y -JLggg = 2 __ _ ^ = Q ...
^ + 265 * ^ ' *
= a + VWTcF and a; = ^— = -|- = ^ .

It may be remarked in this place that cubic equations got
by reduction of biquadratic equations may be solved by
Cardon's Rule, instead of the method of trial as effected in
the preceding examples.
A EEW NEW PROBLEMS.
PROB. (1.) FIND THE GREATEST AREA THAT CAN BE INCLUDED

BY FOUR GIVEN STRAIGHT LINES. (Fig. 71.)
Let a, bj c, d, = four given straight lines, n = the angle

included by a, b and m = the angle included by c, d and D =
... . , cd&m.m ab sin. /a cd
diagonal; .*. area required = -— -f — = 77-
(/ sin.
. m H, ab sin.
. n \) = max. . • . sin.
. m -\ ab7 sin.
. n = max.
V cd I cd
Squaring this expression, sin.2m -f- —=■2ab sin.m mn.n
. 4- a2b2
-^
sin.2^ = max. = r (1.)

But c2 + d2 — 2cd cos.m = D2 = a2 + b2 — 2a6 cos.rc,
ab c2 -f d2 — a2, —- b2 „ ,
.*. cos.m — —^cos.^ = —= = B and .'.
cd 2cd
o 2ab a2b2 „ _. /rt x
cos/wi r cos.m COS.W -f *n» cos/w == xr (2.)
Adding equations (1) and (2), and transposing, we find
r + W-^-l
— eos.fm -f n) = ——7 ; and here it is evident
&ab
cd
that the greatest value for r, or the second member of the
equation, is the greatest positive value of the first member ;
that is to say, we must have — cos. (m + n) = — 1 x cos.
(m -f- n) = 1, which can only take place when cos. (m -f. n)
= — 1 or m + n = 180° .*. sin.m = sin.^, and therefore,
( 184 )
Area = ah * C°'sin.rc = V {P - a) (P - b) (P -c) (P-rf)
where P = ^ as found by calculation.
PROB. (2.) TO FIND SUCH A VALUE OF X THAT (mX + 7l)

{ny + m) = max. and a™*. «/ny = c.
From the second equation we find; mx log a + wy log b =

logc. Let log« = ^ log 6 = J3, and logc = C, .*.
f/ — 7YlA.X
mAx + ftify = C, '.'. ny = = .'. ny + m =
X5
^— , and theretore (m<# + n) {ny + m) =

ml4«a?2 — (mC+
- ^ m?B — ra»^)-x nC +=wm5 = — r, and, there
xl
in JD
n 9 (mC + m?B — nmA) nC + nmB Br
fore a?2 — - — '-% = —-^-A g-r, .\ we
n , , ,, . ~f _ ^ C 4- *w-B — ^
find (as in problems m Chap. 1st) x = —-j =
log c + m log b — n log a . c£m
— a
2m log — ~ lo0** an
log a2m '
PROB. (3.) OM AND OP ARE TWO ARCS OF GREAT CIRCLES

ON A SPHERE, AND THE ARC PM IS DRAWN PERPENDI
CULAR TO OM, FIND WrHEN THE DIFFERENCE BETWEEN
OP AND OM IS THE GREATEST. (Fig. 72.)
Let POM = cc, OP = 0, and Oilf =0, .-. 0 — 0 = max.

= r. By Napier's Rules for the solution of right-angled
triangles (spherical) tanfl = cos. a tan0; .*. 6 = <j> — r .'.
( 185 )
, n 'tan d> + tan r x + rf
tan U = 1 — 7tan (j) tan r = cos. a tan 0r or 1 — r —x (where
r' = tanr = max.) = ax x (where a = cos. a, and a;
\ — a 1 « — 1
= tan (j>) or #2 + x = —
a/ « 2ar'
v n — a)2
,oo
4arr4
i
• Here it is evident that when r = max.
a
then , o o = nun. .*. when r = max. then we must

4«V
(1 - aY 1 a - 1 a - 1 1
have 4oVa
, o a =—,.".
a' r = 2a/
=—7=
# .'. # = ——
2ar = -7=
v/^
= (cos. #)-*.
I had to say something more regarding the Algebraical

theory of Maxima and Minima, but being afraid of enlarging
the wrork too much, I conclude these sheets.
P. S. D'ROZARIO AND CO., PRINTERS, TANK SQUARE.
2 B
yf S^v
c© / / \
fc ^ \
i *
\ Is.
\
£ /
£
S\ V
^^?>'''
^ k > *
r>. JK ^\
^ f N \N
/
n
\A
x^ ^y ^-^-
^ f^
^
^
LONDON :
COX AND WYMAN, PRINTERS, GREAT QUEEN STREET,
lincoln's-inn FIELDS.

A Treatise On Problems of Maxima and Min

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

A Treatise On Problems of Maxima and Min

Hochgeladen von

Copyright:

Verfügbare Formate

This is a reproduction of a library book that was digitized

by Google as part of an ongoing effort to preserve the

Scanned by Imagenes Digitales

Date work Began: _

MAXIMA AND MINIMA,

REPRINTED BY ORDER OE THE HONOURABLE COURT OP DIRECTORS

WLvtizx tfye <8upmtttm&mce o£

PROFESSOR OF MATHEMATICS IM" UNIVERSITY COLLEGE, LONDON.

COX AND WYMAN, PRINTERS, GREAT QUEEN STREET,

In the year 1850, iny friend the late* J. E. Drinkwater-

* John Elliot Drinkwater, the eldest son of Lieutenant-Colonel John

algebra, proceeds as follows :—

" The Brahma Sidd'hanta, the work of Brahmegupta, an Indian

Greece and India stand out, in ancient times, as the countries of

Business brought him to my house one morning, nearly thirty

the demands of his own spirit an impulse towards speculation of a

* It would not at all surprise me if it should be found that some one

"Scribatur x— a=0, x— /3=0, x— y = 0, a?— 3=0, &c. et per

I believe that selections from Kamchundra's work might advan

Having thus given the reader the account which he will

Uniyeksity College, Lohpok,

PBOBLEMS OF MAXIMA AND MINIMA,

For the last four or five years I was desirous of solving

All the problems solved in the second chapter bring out

The third and fourth chapters, and the supplement con

This work contains about 130 problems taken chiefly

I have also given definitions, formulae, and propositions

Owing to the necessity of having the work printed in

Chapter I.—Problems of Maxima and Minima in the

Chapter II.—Problems of Maxima and Minima in

Chapter III.—Problems of Maxima and Minima in

Chapter IV.—Problems of Maxima and Minima in

(1.) REDUCTIONS OF EQUATIONS.

m.= -^o or i^-= -°-=0 or Q is in both

(2.) TO FIND THE EQUATION TO THE PARABOLA. (Fig. 1.)

(3.) TO FIND THE EQUATION TO THE ELLIPSE. (Fig. 2.)

Let two indefinite lines Sm, Hm, revolve, in a given plane,

(4.) TO FIND THE EQUATIONS TO THE ELLIPSOID,

An Ellipsoid is a solid figure, such that sections of it per

(5.) TO FIND THE AREA OE A TRIANGLE. (Fig. 4.)

Rule 1st.—Multiply the base by the perpendicular height,

(6.) TO FIND THE DIAMETER AND CIRCUMFERENCE OF ANY

This may be done by the following proportion, viz. As

(7.) THE AREA OF ANY CIRCLE = RECTANGLE OF \ CIRCUM

Conceive a regular polygon to be inscribed in a circle;

(8.) EVERY SPHERE IS TWO-THIRDS OF ITS CIRCUMSCRIBING

(9.) TO FIND THE AREA OF AN ELLIPSE. (Kg. 8.)

The equation to the Ellipse is y = - n/ a2 — ^2 and to the

circle described on the major axis as diameter is y1 — %/ a2— a?21.

(10.) TO FIND THE SUM OF U TERMS OF THE SERIES

posed to be indefinitely great, ^ and 2 ft may be put instead

(11.) TO FIND THE AUEA OF A PARABOLA. (Fig. 9.)

The equation to the parabola is y2 = 4&aos and consequently

rect. Fw= GF x Fr = GF x —A— = -;—r &c. = &c.

^■D2 AD (n+ 1) (2k + 1) __ r ^ (» + l) (2» + 1)

Put the given number = a, one of the parts required =

than — for if it be so, the value of x becomes impossible ;

r> and .•.# = —-

The same solved without impossible roots.

let x = y + h where the value of y determined by the

condition of ax — x2 being a maximum, will show whether

dently a maximum when y = 0, .•.#=- as before.

PROB. (2.) TO DETERMINE THE GREATEST RECTANGLE