Professional Documents
Culture Documents
T.L. Heath - Apollonius of Perga: Treatise On Conic Sections
T.L. Heath - Apollonius of Perga: Treatise On Conic Sections
T.L. Heath - Apollonius of Perga: Treatise On Conic Sections
CALIFORNIA
LIBRARY OF THE
UNIVERSITY OF CALIFORNIA
fY
OF
CALIFORNIA
LIBRARY
OF
THE
UNIVERSITY OF CALIFORNIA
9se
?
uNliEHSITY OF
CUIFORHU
LIBRARY
OF
THE
UNIVERSITY OF
CUIfORHlA
UNIVERSITY OF CALIFORNIA
LIBRARY
OF
THE
UNIVERSITY OF CALIFORNIA
APOLLONIUS OF PERGA
TREATISE ON CONIC SECTIONS.
UonDon:
C.
J.
ARGYLE STREET.
lLftp>is:
0fto goTit:
^CNIVERSITT
ejcctus
adjViaMai/u
UtJ ammaScrudh
coMXiXcs ita
Jiatur^hcnc
APOLLONIUS OF PEEGA
TEEATISE ON CONIC SECTIONS
RV
T.
L.
HEATH,
M.A.
^\$
, ' .(',
oh
>
Proclub.
/\b1
txfartl
PRINTED BY
J.
AND
C.
F.
CLAY,
MANIBUS
EDMUNDI
HALLEY
D. D. D.
'"
UNIVERSITT^
PREFACE.
TT
-*-
is
not too
much
to
majority of
is
nothing
Yet
book,
written
some
twenty-one
"
the most
by purely geometrical
The general
neglect
Euclid
for,
Elements of
Euclid are
still,
is,
so far
nil.
Nor
is
it
In the
could
seven
Books
in
Greek
or
Latin
which
contain
387
PREFACE.
propositions in all;
and doubtless the apparently portentous bulk of the treatise has deterred many from attempting to
Again, the form of the propositions is because the reader finds in them none of the ordinary aids towards the comprehension of somewhat complicated geometrical work, such as the conventional appro-
make
its
acquaintance.
an additional
difficulty,
priation, in modern text-books, of definite letters to denote particular points on the various conic sections. On the contrary, the enunciations of propositions which, by the aid of a notation once agreed upon, can now be stated in a few lines, were by Apollonius invariably given in vords like the enunciations of Euclid. These latter are often sufficiently unwieldy: but the inconvenience is gi-eatly intensified in Apollonius, where the greater complexity of the conceptions entering into the investigation of comes, as compared with the more elementary notions relating
circle, necessitates in many instances an enunciation extending over a space equal to (say) half a page of this book. Hence it is often a matter of considerable labour even
to the line
and
paragraphs mark
the formal divisions, printed continuously; there are no breaks for the purpose of enabling the eye to take in readily the successive steps in the demonstration and so facilitating the comprehension of the argument as a whole. There is no uniformity of notation, but in almost every fresh proposition a
different letter is
wonder then
of
even
employed to denote the same point: what there are the most serious obstacles in the way remembering the results of certain propositions?
if
Nevertheless these propositions, though unfamiliar to mathematicians of the present day, are of the very essence of Apollonius' system, are being constantly used, and must therefore necessarily
be borne in mind.
refer to the editions where Apollonius can be read in the Greek or in a Latin translation, i.e. to those of Halley and Heiberg; but the only attempt which has been
PREFACE.
ix
made
in
ApoUouius
is
open to
much the same objections. This reproduction of the Conies in German by H. Balsam (Berlin, 1861) is a work deserving great
praise both for its accuracy
occasional
for
an admirable set
;
number
the
in Wi^rds,
there are few breaks in the continuity of the printing, and the
notation
is
of any
more
real
the
editions.
An
edition
is
therefore
still
wanted which
closely
shall,
while in
same time be
so entirely remodelled
by
the aid
of accepted
modern notation
as
to
;
be thoroughly
readable
it
and
this
want
is
my mind
that any
certain
reproduction
of
it
the
Conies
must
fulfil
should
be
scheme followed by
the author
may be
seen as a whole.
it
Accordingly I considered
should
first
make myself thoroughly familiar with the whole work at hand. With this object I first wrote out a perfectly literal
This was a
it
was not
owing
Of
these editions,
Halley's
is
of its design
for
Books V
vii
it is still
tht
For Books
iv
used
for the
most
new Greek
who
and of
all
still
extant in Greek.
The
is
a great
its
admirably executed
I
me
IV.
The
new and
j)ositions,
real
constructive work of
it
unifonn
notation,
the condensation of
some pro-
iv-arrangements of order
kindred propositions
in cases
The
number
of separate propositions.
When
it
seemed
same
more
field
(2) of exj>laining
fully
and
;is
they stiind
The
first
of these
up
and
have ac-
cordingly coiisidrn-d
worth while
to
make
PREFACE.
Introduction as far as possible complete.
of
xi
Thus
all
e.g. in
the case
Archimedes
the propositions
in conies to
and
historical sketch
for
the
period covered, than any that has yet appeared in English, but
also not less interesting
and the
have
been
constantly
indebted
to
an
admirable work by
(German
edition,
to a large extent
covers the
his work,
here replaced by
the
re-edited
the
Eucl.
]
Book
I.
itself
by Proclus
1873).
The
frontispiece to
this
volume
is
a reproduction
of a
quaint picture
The
less
story
is
Pergameni
ed.
I.
7<;
like,
Avith
iirl
point (cf
c.
Claudii
p.
Galeni
V. 8,
108,
3-8
title
page
Proclus'
Friedlein) in which he
of cases,
and the
Proclus
it
work
think
at least contains
' 7payaetBeipap ^;
I
may
claim for
my
own, that
e^ei
and
Ml
should indeed be proud
it
if,
\(
l)r
\
to
L\st\y,
wish
.
in the
PREFACE.
judgnieuL of competent
it
critics,
express
my
thanks
to
my
brother,
H.
S.
for his
for the
of the work.
T. L.
HEATH.
MarcJi, 1896.
et
Sereni Antis-
de sectione cylindri
(Oxford, 1710.)
Edmund Hallet,
Arahico
J.
versi.
L.
antiquis.
H. Balsam, Des Apollonius von Perga sieben BUcher iiber Kegelschnitte iiebst deni durch Halley wieder hergestellten ctchten ucke deutsch
bearbeitet.
(Berlin, 1861.)
.T.
L.
Heiberg,
Litterargeschichtlicke
Studien
iiber
Enklid.
(Leipzig,
1882.)
J. L.
(Leipzig, 1883-8.)
G. Friedlein,
Prodi Diadochi
in
primum
Eticlidis
eJementorum librum
commentarii.
J. L. J.
(Leipzig, 1873.)
(Copenhagen, 1879.)
L.
omnia
cum commentariis
quae svpersunt.
Eutocii.
(Leipzig, 1880-1.)
F.
collectionis
(Berlin,
C. A.
(Leipzig, 1870.)
M. Cantor, Vorlesungen
iiber Geschichte
der Mathematik.
(Leipzig, 1880.)
Deutsche
(Copenhagen, 1886.)
C/^lifOrnia^--
CONTENTS,
INTRODUCTION.
PART
Chapter
T.
I.
discovery
of
Conic
Sections
Mexvii
NAECHMUS
Chapter U. Chapter
III.
Aristaeus and
Eucmp
xxxi
jij
Archimedes
PART
II.
Chapter
Chapter
his
II.
General characteristics
Adherence to Euclidean form, conceptions and
language
Ixsxvii
1.
Xxxxvii
xc\ni
xcviii
ci ci ci
cii
2. 3.
Chapter
III.
1
.
The
(1)
.methods of Apollonius
Geometrical algebra
(2)
(3)
cxi
cxiii
(4)
2.
The use
of coordinates
3. 4.
5.
Transformation of coordinates Method of finding two mean proportionals Method of constructing normals passing through
a given point
....
oxv
cxviii
L^
cxxv
c.xxvii
Chapter
IV.
The construction ok
tangents
a conic by means ok
isxx
xvi
rONTENTS.
PAGE
V.
CXXXviu
iiAiTKR VI.
The constriction of
I'OiNTs
a conic
through five
cli
Appendix.
clvii
CONJUGATE
15
22
ITS
31
42 53
.
ASYMPTOTES
TANGENTS, CONJUGATE DIAMETERS AND AXES.
64
84
RECTANGLES CHORDS
UNDER
SEGMENTS
OF
INTERSECTING
95
.
102
109
113
ETC.
119
126
I39
168 ISQ
187
CONSTRUCTION OF NORMALS
OTHER PROPOSITIONS RESPECTING MAXIMA AND MINIMA EQUAT< AND SIMILAR CONICS
PROBLEMS VALUES OF CERTAIN FUNCTIONS OF THE LENGTHS OF CONJUGATE DIAMETERS
I97
209
221
INTEODUCTION.
PART
I.
CHAPTER
I.
of
is
The
Book
find
of
Archimedes' treatise
is
On
the
and the
following
mention of
Menaechmus,
with
whom
begins.
" Eratosthenes to
King Ptolemy
greeting.
"There is a story that one of the old tragedians represented Minos as wishing to erect a tomb for Glaucus and as saying, when he heard that it was a hundred feet every way,
Too small thy plan
Let
it
to
be double
In quotations from Archimedes or the commentaries of Eutocius on his works the references are throughout to Heiberg's edition (Archimedis oprra omnia cum commeiitariis Eutocii. 3 vols. Leipzig, 1880-1). The reference here
*
is ni. p.
102.
Valckenaer (Diatribe de fragm. Eurip.) suggests that the verses are from the
7$ '
7*
'
iv
'
H. C.
XVUl
THE
,
;
HISTORY OF CONICS.
in error
for,
when the
manner one
;
remained in
And
for they starttnl from a culie and sought to double it. While then for a long time everyone was at a loss, Hippocrates of (Miios was the first to ohser\e that, if between two straight lines of which the greater is double of the less it were discovered how to find two mean proportionals in continueil proportion, the cube would be doubled and thus he turned the dilKculty in the original problem*
;
Afterwards, they
to
in
into the
same
difficulty.
And
they sent
to
and
Plato in the
Academy
them the required solution. And while they set themselves energetically to work and sought to find two means between two given straight lines, Archytas of Tarentum is said to have discovered them by means of half-cylinders, and Eudoxus by means of the so-called curved lines. It is, however, characteristic of them
find for
all
make
Menaechmus and that with difficulty." Home verses at the end of the letter, in commending Eratosthenes' own solution, suggest that there need be no resort to Archytas'
except to a small extent
unwieldy contrivances of cylinders or to " cutting the cone in the triiuls of Menaechmus t." This last phrase of Eratosthenes appears
Poli/iilus
of
Euripides,
Eratosthenes' own, iind that the verses from the trapedy are simply
y'
tXeioi
if
:
-' ) \^.
\(
(or
^)
are
It
871) gives the three verses as above, but holds that they do not belong to the lOlyidus, adding that they are no doubt from an earlier poet than
no
his difliculty."
think
it is
better to regard
as neuter, and
MENAECHMUS.
again,
wliere,
xix
by way
discovered by Menaechmus.
Thus the evidence so far shows (1) that Menaechmus (a pupil of Eudoxus and a conteniporary of Phito) was the discoverer of the conic sections, and (2) that lie used them as a means of solving the
problem of the doubling of the cube. We learn fui-ther from Eutociust that IMenaechmus gave two solutions of the problem of
the two
oi-iginal
mean
by the intersection
of a certain
and
Assuming that a, b are the two given unequal straight lines and .r, y the two required mean proportionals, the discovery of Hippocrates amounted to the
discovery of the fact that from the relation
!^=i=f X y b
it
(1)
follows that
if
C-Y
a?
(1) are
.-
and,
a -
2b,
= 2x\
xy = ab
The equations
x^
= ay,
bx,
(2),
equations
(2).
Let AO,
at 0,
to y.
*
BO
and of length
Comm. on End.
p. 177.
b respectively
Produce
BO
to
and
AO
The passage
is
quoted, witli
the context, in the work of Bietschneider, Die Geometrie nnd die Geometer vor
Kuklides,
t Commentary on Archimedex (ed. Heiberg, in. p. 9298). X It must be borne in mind that the words parabola and hyperbola could not
on
is
One
it
to
make
The
the second solution, with the sole addition of the portion of the rectangular
first
solution.
62
XX
solution
now
consists in
its
vertex
parameter
equal to
BO
or
h,
and a hyperhola with Ox, Oy as asymptotes such that the rectangle under the distances of any point on the curve from Ox, Oy respecbe If tively is equal to the rectangle under 0, BO, i.e. to ah.
MENAECHMUS.
vertex
is
0, axis
Oy and parameter
is
equal to
a.
The point
where
given by
ar
.
= ay
wlience, as before,
a X y - = - = !f X y b
in these
We
have therefore,
two
and the
and
it is
was
due to
to
determine
loci
possessing
these
characteristic
This supposition
is
confirmed by
way
originally
indeed
the
special
method
difficult
to
explain
on any other
after
assumption.
It
is
that,
the
that of
resulting
of
the
equations would
searching investigation.
The form
a fixed point, and the extremities of the lines are found by means of
a rectangular frame, three sides of which are fixed, while the fourth
side can
move
form
(2) of
determine the
corresponding thereto.
known
that the
locus represented
by
y^
= ..,, where y
is
is
and
x^, x,
are the
it would be natural to assume that the equation y' = bx, from the other only in the fact that a constant is substituted for one of the variable magnitudes, would be capable of
and
differing
The only
difficulty
and
it
introduced.
If
an explanation
is
XXll
mus should
we
find it in
the fact that solid geometry had alreivdy reached a high state of
development,
jus is
shown by the
itself
mean
proportionals by Archytas of
This solution, in
two Tarentum (born about 430 B.C.). perhaps more remarkable than any other,
solution of the problem of the
whose base
is
ivs
of the cone, (3) the surface formed by causing a semicircle, whose diameter is the same as that of the circular base of the cylinder and whose plane is perpendicular to that of the circle, to revolve
al)out the
the semicircle nujve.s always in the plane of the circle, in other words, the surface consisting of half a uplit ring whose centre
(he cone and whose inner diameter
(3) is said to
be a certain curve
(h)
double curvature,
the proof, and
(c),
(
is
is
indefinitely small.
We find that
and
rira),
used in
cone*.
The
solution of Archytas
mean
and
AC
circle,
AD
A
i
Bcmicircle
is
AC
AUC, and
.1
MENAECHMUS.
Thus the introduction
Concerning
ties
of cones
in itself
method
cone
of
the
we have no
his probable
definite
information
procedure
ABC as base: this will cut the APC in a certain curve. Lastly let CD, the tanjicnt to the circle ABC at the point C, meet Alt produced in I); and suppose the triangle ACD to revolve about AC as axis.
A
half-cylinder (right)
now
erected with
moving semicircle
will and the point ABC and having ita diameter BE at right angles to AC. The surface of the cone will meet in some the curve described on the cylinder. Let APC be the conesponding point position of the revolving semicircle, and let meet the circle ABC in M. Drawing PM perpendicular to the plane of ABC, we see that it must meet the circumference of the circle ABC because is on the cylinder which stands on ABC as base. Let AP meet the circumference of the semicircle BQE in Q, and let meet its diameter BE in N. Join PC, QM, QN.
will generate the surface of a right circular cone,
This
describe a semicircle
BQE
AC
AC
is their
QN is perpendicular to BE. Hence QN-=BN NE = AN NM. Therefore the angle AQM is a right angle. But the angle CPA is also right therefore MQ is parallel to CP.
QN.
Therefore
.
. :
It follows,
C'A
That
is,
AP = AP AC AP^AP
:
AM^AM AM=AM
if
AQ.
AB,
AC
AC
is
perpendicular to
parallel to
AC
in the plane of
z,
ABC
is
PM
the axis of
then
surfaces
x-
+ U- + '''=^i^'
.c--fi/-'
(1).
= rtx
(2), (3),
where
+ y- + z'^=ajx'- + yAC = a, AB = b.
.x-
From
the
first
two equations
and from
this equation
and
a
(3)
we have
^ Jx^+y^+z" ^
V^+I/-
y/x-'+y'
l^
or
AC:AP=AP:AM=AM:AB.
xxiv
if
.
in
HISTUUY
(F
COMCS.
manner
in
we bear
mind
(1) wljat
we
which the
earlier writers
kinds of
rii,dit
on conies produced the three curves from particular circular cones, and (2) the course followed by Apol-
any
circular cone,
whether right or oblique. Eutocius, in his comnientaiy on the Conies of Apollonius, quotes
with approval a statement of Geminus to the effect that the ancients defined a cone as the surface described by the revolution of a rightangled triangle about one of the sides containing the right angle, and
that they
knew no
otlier cones
Of these they
dis-
was
less
than, equal
to,
Further
they prcKluced only one of the three sections from each kind of cone, always cutting it by a plane perpendicular to one of the generating
lines,
was their name for a parabola, the " section of an acute-angled cone" for an ellipse, and the "section of an obtuse-angled cone" for The sections are so described by Archimedes. a hyperbola.
cone
"
Now
clearly the parabola is the one of the three sections for the
be a
point on the diameter JiC of any circular section in such a cone, and
if
7'
I'y'-^BN.NC.
Draw
generator
AM
OB
in
OBC
meeting the
at right angles in
.1,
and draw
meeting
in
OC
D;
let
DEF,
in
/'.
perpendicular to
and
AN produced
AD
is
Then
is
AF
likewise bisected by
Draw CG
perpendicular to
BC
meeting
produced in G.
Now
the angles
iV,
BCG
are right
therefore B, A, C,
are
(oncyclic,
and
B.V.NC ^AN.NG.
But
AN=CD = FG-
MENAECHMUS.
tlierefore, if
.1
F meets
Hence
and,
if
is
'2AL
is
constant.
./
XXVI
proljiil.le
Mcnapclinius was aware of these general propositions. It is more that he obtained the equation referred to the asymptotes
ref'rred to tlie axes;
and
Thus,
if / Ite
PK'
be perpendicular
to the iusyniptotes
CH'
of a rectangular
hyperbola,
and
if
ii.syniptotes,
PK' = the
= the
CKPK'
CKPE,
quadrilatei-al
since
aCEK'=
APJiA'.
Hence
PK
FK' ^ A RON -
PEN
= h{CN^-PN')
Nslicrc
.',
//
an the coonlinates of
/'
liyjKTbola.
We
obtuse-anglt'd cone,
have then U> sljow iiow MeiKWJchnius could obtain from an by a section perpendicular to a generator, the
H'ctangular hyperlntla
a:'
y* ^ (const.)
=4
say,
or
y _ avr.
./,,
when
y from the
jMiints yl,
respectively,
and
A' -a.
MENAKCHMrS.
Take an obtuse-angled
circular section of
it.
through
A' and
A draw
in
AN
cone, and let BC be the diameter of any Let A be any point on the generator OB, and -Ai right angles to OH meeting CO produced in
BC
N.
line
drawn from
perpen-
OBC
Then y
will be
f^BN.NC.
Let
AD
BC
and meeting
OC
in
DF,
CG
be drawn perpendicular to
BC
meeting
AX
G
produced in Z, F,
are coney clic
respectively.
BAG, BGG
/>,
A, C,
..
y-
But
similar triangles,
^A'N
Hence
AA'.
AN.
2AL
AA
AF '^,.'
- AA'^'-'
and the locus of the extrenuty of y fur different positions
of tlie
by a
-.
.,
^^
XXVMl
This
AOL
is
equal
to half the
the i)osition of
apex
is
(f,
determined, and draw the circle circumscribing meet LO produced in some point K, and OA' will le its diameter. Thus the angle A'KO is right .. _ = complement of .ALK= ^AOL = ^ LOO - _ A'OK, A' are equal, and whence it follows that the segments AK, A'
For suppose
;
AOA'
this will
therefore A'
lies
on the
line })isecting
^'-
is
drawing a
semicircle.
line
bisecting A A' at
A'
angles
Thus,
A' Z joined,
determined.
lii.illy
an
fllipse '
AA
-,-
will
be ditlerent
from
unity.
Ut
is
AA\
is
^'
(as
Archimedes
called
parabola).
in
The jissumption that Menaeclinius discovered all three sections the manner alx)ve set forth agrees with the reference of
that the
ICratosthenes to
proliJible
cylinder.
Menaechmean triads," though it is not imwas known earlier as a section of a right Thus a passage of Euclid's Phdenomena says, "if a cone
tlie
"
ellip.se
a section of an acute-angled cone which is similar to a showing that Euclid distinguished the two ways of producing an ellipse. Heiberg {Littfrargeschichtliche Studien iiher
%"
It
is
h'liklid, p.
88) thinks
it
probable that
a question whether Menaeclimus used mechanical contrivOvpeov for the cllipBe occur.s several times in Proclus
;
The cxpreHeion
imd particularly in a passage in which ueminus is quoted (p. Ill) and it would seem as though this name for the curve was more common in Geminus' time than the name "ellipse." [liretschucidcr, p. 170.]
MEXAECHMUS.
ances for effecting the coHstruction of his curves.
Tlie idea that he
all
who had
two mean
pro-
portionals had written theoretically but had not been able to effect
a certain extent, Menaechmus and that only with dithculty, (2) upon two well known passages in Plutarch. One of these latter states
that Plato blamed Eudoxus, Archytas and
Menaechmus
for trying
and mechanical constructions (as though such methods of finding two mean proportionals were not legitimate), arguing that the good of geometry was thus lost and destroyed, as it was brought back again to the world of sense instead of soaring upwards and laying hold of those eternal and incorporeal images amid which God is and thus is ever Godt; the other passage {Vita MarceUi 14, 5) states that, in consequence of this attitude of Plato, mechanics was completely diA-orced from geometry and, after being neglected by philosophers for a long time, became merely a part of the science of war. I do not think it follows from tliese passages that Menaechmus and Archytas made machines for effecting their constructions; such a supposition would in fact seem to be inconsistent Avith the direct statement of Eratosthenes that, with the partial exception of Menaechmus, the three geometers referred to gave theoretical solutions only. The words of Eratosthenes imply that Archytas did not use any mechanical contrivance, and, as regards Menaechmus, they rather suggest such
a method as the finding of a large number of points on the curve J.
It seems likely therefore that Plato's criticism referred, not to the
.;' (% ( '
is
:
j).
xviii.
The Greek
';
ets
opyafiKas
KaraaKevas
,
[scr.
vapfiKOi
).
(( ^ , ' . ,
rovs
[scr.
' ']
alairtp
OTepfoO
^' ^
6 debs del
^>,
^/-
(6
1.)
where
partly suggested by Eutocius' commentary on Apollonius t. 20, 21, remarked that it was often necessary for want of instruments to describe a conic by a continuous series of points. This passage is quoted by Dr Taylor, Ancient and Modern Geometry of Con/r-s-, p. xxxiii.
This
it
is
is
XXX
mechanical
and Menaechmus.
Much
solution
is
hivs
been written on
the
difHculty
of
reconciling the
censure on Archytas and the rest with the fact that a mechanical
attril)ute<l
is
The most
proljable explanation
been consulted by
Zeuthen
luus
may have
referred,
was not satisfied to regard a curve as completely defined by a fundamental plane property such as we express by the equation, but must needs give it a geoin the case of
Menaechmus,
metrical definition
iis
make
it
its frm
renli.sable
u.se
by the
in
senses,
though
this presentation of
its
of
pioperties
not so comprehensible
if
applied
where the
cui-ve in
which the
is
a curve
easily represented by
an
equation.
CHAPTER
II.
We come next to the treatises which Aristaeus the elder' and Euclid are said to have written; and it will be convenient to deal with these together, in view of the manner in which the two names are associated in the description of Pappus, who is our authority upon the contents of the works, both of which are lost. The passage of Piippus is in some places obscure and some sentences are put in
'
its
"The four books of Euclid's conies were completed by ApoUonius, who added four more and produced eight books of couics. Aristaeus, who wrote the still extant iive books of nolid loci connected with the conies, called one of the conic sections the section
an acute-angled cone, another the section of a right-angled cone and the third the section of an obtuse-angled cone.... ApoUonius says in his third book that the locus with respect to three or four lines' had not been completely investigated by Euclid, and in fact neither ApoUonius himself nor any one else could have added in the
of
'
least to
of conies only
what was written by Euclid with the help of those properties which had heen proved up to Euclid's time; ApoUonius
is
himself
when he
says that
tiie
theory of
Now
Euclid
regard-
made in conies, and without anticipating him or wishing to construct anew the same system (such was his scrupulous fairness and his exemplary kindline.ss towards all who could advance mathematical
science to however small
in
no vise con-
wrote so
much about
Hultsch), pp.
672 67.
XXxii
Had lie done so he would certainly have deserved censure, but, as matters stand, he does not by any means deserve it, seeing that neither is ApoUonius called to account, though he left the most part
of
tlir
liis
conies incomplete.
ApoUonius,
too,
lacking portion of the theory of the locus through having become familiar iK'forehand with what haxl already been written about it by
Euclid and having spent a long time with the pupils of Euclid in Alexandria, to which training he owed his scientific habit of mind. Now this locus with respect to three and four lines,' the theory of
'
is so proud of having added to (though he should rather acknowledge his obligations to the original author of it), is arrived at If three straight lines be given in position and from in this way.
which he
one and the same point straight lines be drawn to meet the three
straight lines at given angles, and
if
contained by two of the straight lines so drawn to the square of the remaining one be given, then the point will lie on a solid locus given
in position, that is
sections.
And,
if
straight lines be
drawn
given in position, and the ratio of the rectangle under two of the lines so drawn to the rectangle under the remaining two be given,
lie
(? ).
is
It
<;
that
loci
and
are lines, are again divided by Proclus into plane loci and solid loci
( ''
figure,
siirface-loci
( ).
two
tv
classes, line-loci
),{
and the
all
()
as " a position of a
The former, or
( )
which
sections.
and
like
(),
the
cylindrical
helix
Similarly
(I)
the plane
locus
the locus of
from which on a
are
is is
mean
line
distances from two fixed points are in a given ratio (a locus investigat*d
by ApoUonius in
tlu!
>'/),
name from
ARISTAEUS
EUCLID.
wxiii
they arise from the cutting of solid figures, as for instaiice the
sections of the cone
Pappus makes a
loci,
i.e.
fui-ther
(
loci
plane
may
)
of,
and linear
( ).
mean
of the class
loci,
of solid
into
Thu.s, he
lineai- loci
nor
circles,
).
or
is
is
a surface
Thus
more complicated and unnatural origin than straight lines, circles and conies, " being generated from more irregular surfaces and
intricate movements;}:."
It is now possible to draw certain conclusions from the passage Pappus above reproduced. 1. The work of Aristaeus on solid loci vas concerned with those loci which are parabolas, ellipses, or hyperbolas in other words, it was a treatise on conies regarded as loci. This book on solid loci preceded that of Euclid on conies 2. and vas, at least in point of originality, more important. Though both treatises dealt with the same subject-matter, the object and the point of view were different had they been the same, Euclid could scarcely have refrained, as Pappus says he did, from an attempt to improve upon the earlier treatise. Pappus' meaning must therefore be that, while Euclid wrote on the general theory of of
;
;
loci of Aristaeus.
angled,
of
The
three-line
Apollonius, Vol.
p.
X Pappus,
(
.
C.
270
'
ii.
p. 184.
yap
yivtaiv
tiTi-K(ir\(y
^ (^ *, -^..
ras
+ Pappus,
p.
62.
d%
xxxiv
TMK
.
tlie
HISTORY OF TONICS.
same
locus,
was unable
to
work out
and did not jxdd fresh discoveries of his own. 5. The Conies of Euclid was superseded by the ApoUonius, and, though the Solid Loci of Aristaeus was
in
treatise
still
of
extant
it is doubtful whether Euclid's work >vas so. The subject of the three-line and four-line locns will be discussed but it may be in some detail in connexion with ApoUonius convenient to mention here that Zeuthen, who devotes some bril-
Pappus' time,
liant chapters to it, conjectures that the imperfection of the investigations of Aristaeus and Euclid arose from the absence of
any conception of the hyperbola with two branches as forming one curve (which was the discovery of ApoUonius, as may be inferred even from the fulness with which he treats of the doubleThus the proposition that the rectangles under the hyperbola).
segments of
ratio
intei-secting chords in fixed directions are in
a constant
is
and
and parabola.
So far therefore as
ApoUo-
was incomplete.
On
was reduced
Euclid would
have had the means (which we are told that he had not) of
also.
ApoUonius probably
own
it
and defective
have the
as that of the
first
own
A more
important source of
infi>nnalioii,
in
the sense of
XXW
at
liand in the
Thus
fundamental property of the
:
Tlie
PX' AN.
tliat of tlie
'
ellipse,
--
= P'N"
AN' N'A'
.
BC"
AC",
hyperbola,
PN'
-.AN.
A'
= P'N"
AN'
N'A',
PN-=p,,.AN,
are assumed, and must therefore presumably have been contained in
Euclid's work.
(b)
At
the beginning of
the treatise
on
the
area of
If
QVq
ix
QV =
Vq.
in T,
If the tangent at If
Q meet VP produced
PV= PT.
at
and bisected
V,
: :
PV PV'^QV
'^And
these propositions are
Q'V".
of conies"
(i.e.
proved in
the elements
in
The
On
Conoids and
:
If straight
drawn from the same point touch any conic section whatever, and if there be also other straight lines drawn in the conic section
and cutting one another, the rectangles
"
contained by the segments (of the chords) will have to one another
the same ratio as the squares of the (parallel) tangents.
is
And
this
proved in
(d)
the elements
of conies
."
the parabola
In the same proposition we find the following property of If p be the parameter of the ordinates to the axis,
:
and QQ' be any chord not perpendicular to the axis such that the diameter PV bisects it in V, and if QD be drawn perpendicular to PV, then (says Archimedes), supposing to be such a length
that
QV-.QD'^p:p,,
the squares of the ordinates to
and
c'l
of
XXXvi
towards P.
"For
tJie
conies."
In
otlier
words,
if
/),
are
the
parameters corresponding
referred
Archimedes p. liii.) Euclid still used the old names for the three conic sections, but he was aware that an ellipse could be obtained by cutting a cone in any manner by a plane not parallel to the base (assuming the section to lie wholly between the apex of the cone and its base), and
also
quoted above
by cutting a cylinder. This is expressly stated in the passage But it is scarcely (p. xxviii) from the Phaenomena.
had
in
for,
had the cone been oblique, the statement would not have been true
without a qualification excluding the circular sections subcontrary
to the base of tlie cone.
eVi^avcta, Of the contents of Euclid's Surface-loci, or we know nothing, though it is reasonable to suppose that the treatise dealt with such loci as the surfaces of cones, spheres and But cylinders, and perhaps other surfaces of the second degree. Pappus gives two lemmas to the Surface-loci, one of which (the This lemma states, and second) is of the highest importance*.
of,
whose distance
from a
given point
section,
in a given ratio
is
to its
distcmce
from a fixed
than, unity.
line is
a conic
and
an
ellipse,
a parabola, or a
to,
is less
than, equal
or greater
The proof
unity
is
in the case
is
different
from
shortly as follows.
.S'
J^t
SX be
aS"
on the fixed
dicular to
Let
PN
perpene
SX,
so that
SP
'^
is
to
XX
NX'
Let
be
Now
let
SX such
'
that
XK'
Pappus
(ed.
Hultsch)
p.
seqq.
ARISTAEUS
then,
if
EUCLID.
'
NK = NK\ we shall
PN'
have
+ SN'
NX'
XK
XK'
The position of the points N, K, K' changes with the position of I'. coincides falls when If we suppose A to be the point on which with ', we have
KAN
SK'
K'S
It follows that
-^
-T^
are both
known and
Hence
equal,
and therefore
SX SK TTTr SA SN
r, i
> '
are both
is
known and
equal. ^
expressions
equal to
SX - SK
XK
"*''
SA-SN'
'hich
is
AN'
therefore
known
'"^^
1
XXXVin
In
THE
like iniiiiner,
EAULli:i{
if
HISTORY OF CONICS.
when K'
shall
we
liave
^.
and
is
in the
same way we
XK'
A
.,
is
known and
equal to '
..'
- has
known
value.
And, since
yj-.
^-, =
e',
from above,
This
is
ellipse or
a hyperbola according as
is less
or greater than
for in
and on the same side of case on opposite sides of X, while in the former case
lie
on
A A' and
in the latter
1
will lie
on
A A'
produced.
The
here.
case where e =
is
easy,
We
lemma
for the
lemma
must therefore have been assumed by him as evident or quoted as well known. It may therefore well be that it was taken from some known work*, not impossibly that of Aristaeus on solid loci. That Euclid should have been acquainted with the property of
It
conies referred to the focus and directrix cannot but excite surprise
It
is
Pappus
where he
arc.
discussing the various methods of trisecting an angle or circular (p. 284) a method which " some " had used and which involves the construction of a hyperbola whose eccentricity is 2.
is
He
gives
Suppose
it is
))arts.
Suppose
the angle
it
done, and
is
let .ST
Then
RSP
SRP.
ARISTAEUS
KICLID.
all
XXXIX
in Aimllonius,
and
ex-
The
planation
may
known
to
Book
is
focal property
may
work on conies
We
is
focal
up
to
much
The contrary
indeed more probable, and this supposition is supported by a remarkable coincidence between Apollonius' method of determining the foci of a central conic and the theorem contained in Pappus'
31st
lemma
to Euclid's Porisnis.
is
This theorem
circle,
as follows
Let
'
and from A', A let two straight lines be drawn at right angles Let any straight line HH' meet the two perpendiculai-s to A'A. Further let YS be in R, R' respectively and the semicircle in Y. drawn perpendicular to RR', meeting A'A produced in S.
It
is
to be proved that
AS.SA' = AR.A'R',
i.e.
that
Now,
A'S.
is
equal to
Let
SE
RSP, meeting
RP
in
PN perpen-
dicular to
RS.
Then
the angle
ERS
is
ESR,
RE = ES;
Also
But
whence
or
RX=XS, and X is given. RS SP=RE EP = RX XN RS -.RX^SP -.NX. = 2i?A'; .ST = 2.VA', SP"- = iNX-,
..
:
..
J?.S'
..
PN- + SN"-=iNX":
" Since then the two points
.S',
A'
are given,
is
and
PX
is
perpendicular to SX,
NX-
to
PN- + SN^
given,
lies
on a hyperbola."
to the
This
is
lemma
(,
and the
ratio
's
stated in the
same form
in both cases.
AYS.
-A'YR' = ^AYS;
..
^A'SE'=^
A'S =
-ARS;
iSA
:
A'R
or
AR,
AS.SA' = AR.A'R'.
It follows of course
if is
the rectangle
AR
A'R'
is
constant,
AS .SA
is
is
also constant
and
-S'
a fixed point.
in.
complete circle
the axis
used,
AR, A'R'
and RR'
iii.
A A'
of a conic,
Ho
has already
.
proved,
42
that
case
AR
he
now
AS
He
then proves that
.V,
at each
of the
points
to
Thus Apollonius' procedure is exactly lemma to Euclid's Porisiiis, except that the
the (..
the origin and aim of Euclid's Porisms, namely, that tliey were
jiartly a sort of })artly
and
CHAPTER
III.
ARCHIMEDES.
No
Archimedes.
upon no more (without any
There
is
The
mention of the name of the author) in the passages quoted above, which ha'e by some been assumed to refer to a treatise by Archi-
medes
himself.
is
when
passage*
of
\vhere
words
"
iv
rfj
the
Elements
is
simply
mean
that
it is
found in
positive
so
may
commentary on Apollonius,
medes,
is
to
t
(ed.
Oh lite Sphere luid Cylinder, i. p. 2i. The proposition quoted is Eucl. xii. '2. The name appears in the passage referred to as 'RpaKXeios. Apollonius
Heiberg) Vol.
ii.
p. 168.
((
puts
it,
Archimedes was
not easy to explain.
the
first
to
"invent"
(p.
theorems in conies
is
Bretschneider
156)
down
to the
the contempt
malice with which small minds would probably seek to avenge tiiemsolvos for in which they would be held by an intellectual giant like
xlii
THE
HISTORY OF CONICS.
and Eutocius subjoins the remark that the allegation is in his opinion not true, " for on the one hand Archimedes appears in many passages to have referred to the elements of conies as an older
treatise
),
it is
own
and on the other hand Apollonius does not Thus Eutocius regarded discoveries."
the elementary
i.e.
the
refei-ence
otherwise,
if
he had
8(8.
far
Archimedes,
first
tions to
show how
tiie
jiroducing
Archimedes was aware of the possibility of three conic sections from cones other than right cones
to a generator
We
observe,
iirst,
/^
was
On Conoids and
Spheroids
all
we
find the
an acute-
angled cone"
[i.e.
an
ellipse].
The way
in
which
is
this proposition
same
line
treatise,
where
it is is
.shown that
it is
on a straight
plane of the
not perpendicular to
to determining the
Apollonius.
cleides,
is
unfair to Hera-
many
of the propositions of
making the charge of plagiarism by finding Apollonius already quoted by Archimedes as known.
meant that the olementary theory of conic sections as formulated by Apollonius was due to Archimedes otherwise Eutocius" contradiction would have taken a different form and he would not
;
the
AKCIIIMKDK:
circular sections
to do.
(1)
:liii
is
Conceive an
ellipse
with
/>'/>'
as
its
minor
axis
:
and and
suppose
line
CO drawn
ellipse,
let
in the in
Produce OB, OC, OB', and same plane with them draw BED meeting OC, OB' produced E, D respectively, and in such a direction that
be the apex of the required cone.
BE.ED-.EO^^CA^.CO(where
of the ellipse).
BE ED .EO'>BC.
[Both the construction and this
last
CB' CO-.
:
known.]
Now
BD
through this
circle
and having
We
cone, or,
is
ellipse,
that
lies
on the surface
to
of the cone.
Draw
meet
PN
in
perpendicular to BB'.
let
it
BD
M, and
MQ
circle
on
BD as
of the circle in Q.
respectively
xliv
Now
//.]f
MD
EG)
^'...
.
.
QAP
PiV-
=
^
whence, since
PN,
is
is
a straight
line.
But Q
therefore
cone.
is
circle
on
BD
as diameter
lies
OQ
and therefore
on the
all
Let
OC
the given ellipse, and let the plane of the paper be that containing
-LI'
and 0(\
Ijet
phme.
BB'
l)e
are unequal.
Produce OA'
to
so that
OA
OD.
AD, and
(h-aw
FG
through
parallel to
it.
ARCHIMEDES.
Conceive a plane tluOUjih
paper, and in
either
(it),
xlv
tin
AD
perpendiculai to
plan
.f
tlie
it
describe
if
or axis
(b), if
not,
an
ellipse
on
AD as
he the other
d'.An'=CJr-:FC.CG.
Take a cone with apex
ellipse just
ellipse,
circle or
is
drawn.
This
is
when
the curve
an
AD
is
perpen-
dicular to the plane of the ellipse, and the construction follows that
in the preceding case (1).
'
meet
Let
that
ellipse,
to
Draw
PX
in
perpendicular to A A'.
to
AD
M.
Through
and
draw
HK parallel to A' A.
draw
MQ
dicular to both
AD)
meeting the
ellipse or circle
about
AD
Then
QM' HM. MK={QM' DM. MA) {DM. MA HM MK) = {d' (FC CG A'C CA) = (CB' FC.CG).{FC.CG A'C. CA) = CB':A'C.CA = PN^:A'N.NA. NA QM' PN^ = HM MK = 03P 0N\
: : . : .
..
Hence
cone,
it
OPQ
is
on the surface of the cone. The proof that the three conies can be produced by means of sections of any circular cone, whether right or oblique, which are made by planes perpendicular to the plane of symmetry, but not
follows that
also
is
of course
same as the proof for the ellipse. It is therefore to be inferred that Archimedes was equally aware of the fact that the parabola and the hyperbola could be found otherwise than by the The continued use of the old names of the curves is of old method. no importance in this connexion because the ellipse was still called the "section of an acute-angled cone" after it was discovered that
xlvi
THE
])e
KARLIF.H HISTORY
OF CONICS.
pnjducwl by means of a plane cutting all the generating any cone, whatever its vertical angle. Heiberg concludes that Archimedes only obtained the parabola in the old way
it
could
lines of
is
only
no more an objection to the continued use of the term as a well-known description of the parameter than it is an objection to the continued use by Archimedes of the term "section of an acute-angled
but this
is
ellipse
manner.
Zeuthen points
we
and
Spheroids, 11)
[a paraboloid of revolution] be
cut by a plane through the axis or parallel to the axis, the section
will
(
common
is
).
And
its
and
of that
which
an obtuse-angled cone
()
:
if
of the
And
common section of the plane which cuts the and of that drawn through the axis at right angles to the
" If
cutting plane.
(3)
any one
by a plane
be a section of
an acute-angled cone if through the axis, the actual section which comprehends the figure if paralle%o the axis, similar to it." Archiniodes adds that the proofs of all these propositions are
:
ob\ious.
it is therefore tolerably certain that they were based on the same essential principles as his earlier proofs relating to the
These depend, as
will
if
two chords
ARCHIMKDES.
drawn
point.
in fixed directions intersect in
is
!i
in fixed directions
meet
;
in
if straight lines Fd, IIK between two lines forming an angle, any point M, the ratio MG .MK
FM
HM
constant
of the
Tlie following
any plane which meets all the generators of the enveloping cone, and is not perpendicular to the axis, is an ellipse whose major axis is the part intercepted within the hyperboloid of the line of intersection of the cutting plane and the plane through
the axis perpendicular to
it.
line
Suppose the plane of the paper to be this latter piano, and the EC to be its intersection with the plane of section which is Let Q be any point on perpendicular to the plane of the paper.
the section
ff
QM perpendicular to
liC.
xlviii
Lt't
^^-^'
made by
AD
its axis.
Through J/
in this plane
in E, F.
Then the
point on
it.
EF
a
perpendicular to
AD
is
circle,
QM
.
lies in
its plane,
and
(? is
Therefore
QM' =
meet the axis
EM
in
MF.
is
Now
to
let
PT be
let it
parallel in 0.
BC\ and
Draw /''
Then
perpendicular to
AD.
. : .
QM-
BM MC = EM MF
.
MC
= OA'
which
is
OP';
the section through BC.
constant for
all
positions of
it is
Q on
Also
OA < OP,
because
AT<AN,
whence a fortiori
Therefore
and therefore
OT<OP,
is
OA <0P.
on an
ellipse
lies
BC.
Archimedes,
(rt)
it
that in a hyperbola
AN>AT.
[p.
The
first of
as liaving been
known
xxxv]
assumption
is
also interesting.
how
means
PP' be a diameter
of a hyperbola
QV
rP
:
tangent
QT
TP' =
PV
P'
V,
so that
we may
is
the axis.
PV
was familiar
to
him
for he
ARCHIMEDES.
As
otlici
it
may
Apollonius.
The term
diameter,
Sta/xcrpo? and the minor axis a. is For the hyperbola, by the diameter is only understood that part of it which is within the (.single-branch) hj^erbola. Tiiis
8(.
infer
tlie
major axis
is
ellipse
'
'
'
diameter
'
of a hyperbola
its
is
identified
described by
diameter, while the axis of the hyperboloid does not extend outside
it,
as
it
meets
{)
*
(), and
;
'
the
distance between the vertex and the apex of the enveloping cone
[the centre of the revolving hyperbola]
axis
'
(d
7' ).
is
the
diameter
but in a
segment of a parabola that one which bisects the base of the segment
is
In the ellipse called the diameter of the segment diameters otlier than the axes have no special name, but are simply
( ).
drawn through the centre.' The term axis is only used with reference revolution. For the complete figure it is the axis
'
lines
to
the solids of
;
of revolution
for
tlie
a segment cut
oflf
by a plane
it is
segment of the line, (1) in the paraboloid, dravn through the vertex of the segment parallel to the axis of revolution, (2) in the hyperboloid, joining the vertex of the segment and the apex of the
enveloping cone, (3) in the spheroid, joining the vertices of the two segments into Avhich the figure is divided, the vertex of any segment being the point of contact of the tangent plane parallel to the base.
In a spheroid the
the straight line
'
diameter
'
Thus we are
told
that "those spheroidal figures are called similar whose axes have
the same ratio to the diameters*."
(axes)
of an
ellipse
are
called
conjugate
{<;).
The asymptotes
lilies
of a hyperbola are in
section
Archimedes
the straight
(at
nearest
to
the
*
On Conoidn mid
H. C.
THE EARLIER HISTORY OF CONICS.
eieHaL
centre of a liyperbola
nearest
lines
is
meet
(to
,'
for
*
/?),
while what
the jwint
we
call
the
Archimedes
centre
in which the
/,).
:
'
of a hyperbola
indeed the
two branches
of a hyperbola
Apollonius.
When the asymptotes of a hyperbola revolve with the curve round the axis they generate the cone enveloping or comprehenditig
the liyperboloid,
To/i.a5
(. ). 5 ?
It will be convenient
first
them into
classes,
taking
by
assumed
as known.
I.
1.
They
fall
General.
The proposition about the rectangles under the segments of xxxv and xlviii). 2. Similar conies. The criteria of similarity in the case of central conies and of segments of conies are practically the same as tliose given by Apollonius. The proposition that all parabolas are similar was evidently familiar to Archimedes, and is in fact involved in his statement that
intersecting chords has been already mentioned (p.
all
paraboloids of
revolution
vti).
are
sim'ilar
ovv
o/ixotci
3.
'diameter'
(axis)
are
perpendicular to
II.
1.
TuE
The
Ellipse.
^
of
)7.
relations
:
AiV.
A word
acknowledgement
is
tlie
valuable
summary
of " Die
contained
in the ZeilHchri/l
IfiHO, j<p. 41
fur Mathematik xtnd Physik {Hintorisch-Iiterarische Abthcihnig) This article ie a complete guide to the relevant passages in
I
Arcliimedes, though
ARCHIMEDES.
are constantly used as expressing the fundamental property and the
criterion
2.
is
an
ellipse.
QV
also occurs.
3.
-.FV.rV^Q'V"
..'
major axis as diameter, and
an ordinate
circle in p,
PN to
then
pN
4.
P^==
(const.).
The
The
straight line
to the point of
and,
if
the centre parallel to either tangent and meeting the ellipse in two
points, the parallels
If a cone be cut
is
section
either a circle or
if
Also,
by a plane meeting all the generators, the an ellipse. a cylinder be cut by tAvo parallel planes each meeting all
to one another.
and similar
III.
1.
The Hyperbola.
We
find, as
AN' A'N\
.
QV:
:
Q'V" = PV.P'V:PV'.P'r;
but Archimedes does not give any expression for the constant ratios PN' AN. A' and QV^ PV P'V, from which we may infer that
:
.
is
;
arrived at by
producing
procedure
equal to
CA
are then in each case double of the one case of the whole surface, and in the This term for AA', is tlie 'vertex']. other of a segment of which PP' was, no doubt, only used in order to avoid mention of the cone of
through
A A', PP'
(12
Hi
is a section, as the introduction of this cone might have complicated matters (seeing that the enveloping cone also appears); for it is obvious that A A' appeared first as the distance along the principal diameter of the hyperbola intercepted between
it
in
If from a point on a hyperbola two straight lines are drawn 2. any directions to meet the asymptotes, and from another point
lines are similarly
drawn
parallel respectively to
line
through the point of concourse of the asymptotes and all chords parallel to the
the principal ordinate from P, and
respectively, then
If
PX,
at P,
AN>AT.
5.
If a line
is
IV.
1.
and
QV':Q'V" = PV.PV'
]'
We
forms
'^2^^
QV'=2y.Pr
j)
5 /,
and
is
^.
to the
tSs
is
QV- and
of
width equal to
2.
is
applied.
;
in his note
t Tliis
its converse appear in a fragment given by Eutocius on the 4th proposition of Book ii. On the Sphere and Cylinder. is also used in the fragment quoted by Eutocius.
ARCHIMEDES.
aiul a line parallel tu the axis bisects
If
QD
PV bisecting
the chord
Q VQ', and \i
QQ' while
,
y^
p:p,, =
QV'-:QD\
xxxv, xxxvi).
It is
mentioned above
easily derived
(p.
tion
I.
tangent at
that
in E,
and PT, A
intersect
OP
and
..
'.PE = p: 2P1\
=^.
= p.AN.
QD- = PT'
=^
:
OP =
},PT
Thus
Q'
PN-, by similar
:
triangles,
p.
AN
Pa.
AN
V
in
= P 'Pa-]
t.
If the
tangent at
Q meet
the diameter
and
QV he
PV=PT.
.
By
the aid of the preceding, tangents can be
drawn
to a
parabola () from a point on it, () parallel to a given chord. 6. In the treatise On floatimj bodies tQv
( ),
is
ii.
5,
we have
principal
this proposition
If
KF be
the
measured along the axis away from the vertex and equal to half the
parameter,
while
KII
is
dravn perpendicular
to
FH
perpendicular to the
tangent at P.
It
is
subnormal
parameter.
liv
If
QAQ'
that
QQ'
is
diameter through
in
and
if
li
be
any other point on the curve the ordinate from which RlIK meets PV in // and
the axis in /, then (J/ being the middle point of QQ')
PV
II.
PlI
^^^MK
A,
bodies,
{On floating
[There
is
by
vhom
as follows
We
PV - MK
is
jwsittve or zero.
Let Qq meet
AM in
_ ~
0.
Now
PV_
PV.AK
.
PH ~
.MK
{A
-AK)
AllCHlMKDKS.
Iv
b(i
which
is
negative
'TV
MK\
whence the proposition follows.] If any three similar and similarly situated 8.
ments have one extremity
bases
paraljolic seg-
()
of their
bases
common and
line,
their
if
BQ
and
EO
tangent at
he dravn parallel to the axis of any of the segments meeting the to one of them in E, the common base in 0, and each
of the three
^
[This proposition
is
bq^-q^q:
it is
assumed
10).
But
it
may
well be that
was too
well
known
himself.
Quadrature of a jiarabola which the reader could work out for The latter proposition is given below (No. 1 of the next
if
BB
be the tangent at
to the
segment BB^(J^
ER^
To deduce from this first that, if V ^, V^, V^ be
R/J = BO
OQ^.
we observe
Ivi
THE EARLIER
(parallel)
IIIS'IORV
OF CONICS.
V^, F^
meet the
respective segments in
^,
:
/n\
It follows that
B]\
Ji]\ - I\V,
7^3
PJ\
1\V.,.
/>,
1\, P^,
But, since
BE
if
is
the
tangent at
segment BR^Q^,
TJ\ = PJ^
(where
,,
meets
Therelforo,
,,,
]\P^
BE in meet BE in
=
\).
7;,
7',,
V. = ^.''and
^.,
three segments.
:
and
/>/i' is
therefore a tangent to
all
Next, since
ER^
Similarly
and
(?(?,,
BQ^
BQ,
7?^^.
From the
tirst
I'
EO
\BQ^
BqJ
^BO.Q.Q, bq.-bq:
.-similarly Similarly
R& ^BOJQ^^
^^
BQ^.BQ^
it
From
follows that
R^r BQ.'QM'
9.
If
BQ
BQ_, be
BRJi, be any
AllC'llIMKDES.
straight line tlirough
IvU
A',, A', re.siectively,
J>
then
BQ^
BQ,,- nn^
/?,
Bli^.
in E, the
common
EB^
and
..
EO = BO
BQ^,
BQ.^
:
ER.EO^BO:
ER
-.ER^^BQ^
is
BQ.,.
is
But, since A,
the
diameter through A,
we have
in like
manner
:
ER
Hence
10.
ER^ - ^A,
BQ, = BR^
BR^.
BR.^.]
BQ^
Archimedes assumes the solution of the problem of placing, between two parabolic segments, similar and similarly situated as in the last case, a sti'aight line of a given length and in a direction
parallel to the diameters of either parabola.
I,
solved, A7i,
being equal to
l.
Using the
last figure,
we have
BO
'""^
ER
bcCeo
Subtracting,
we
obtain
. '
whence
which
is
/?(9.
0^ -
^^^"^^S
known.
the ratio
And
BO OE
:
is
given.
Tiierefore
B0\
and therefore
0.
determines
A A,.]
either It remains to describe the investigations in which it is expressed or implied that they represent new developments of the of theory of conies due to Archimedes himself. With the exception
Iviii
THE
HISTOKV OF COXICS.
The preface
Archimedes, not
is
)
:
to the treatise
6<;
^,
interesting.
and a segment of a
circle,
he
proceeds
"
And
line,
was recognised by most people that the problem was not solved. But I am not aware that any one of my predecessors has attempted to square the .segment bounded by a straight line and a section of a right-angled For cone, of which problem I have now discovered the solution. it is here shown that every segment bounded by a straight line and a section of a right-angled cone is four-thirds of the triangle which has the same base and an equal altitude with the segment, and for the demonstration of this fact the following lemma is assumed f that the excess by which the greater of (two) unequal areas exceeds the less can, by being added to itself, be made to exceed any given finite area. The earlier geometers have also used this lemma for it is by the use of this same lemma that they have shown that circles are to one another in the duplicate ratio of their diameters, and that
assuming lemmas not easily conceded, so that
:
;
spheres are to one another in the triplicate ratio of their diameters, and further that every pyramid is one third part of the prism having the same base with the pyramid and equal altitude also, that every
:
cone
one third part of the cylinder having the same base as the cone and equal altitude they proved by assuming a certain
is
lemma
to
it.
The
,
it,
is
ras
and
'
it is
section of
i.e.
right through
a diameter.
,
^
an
might perhaps mean a section cutting and the straight line might be an axis or
' ' '
in view of
addition of
expression always
signifies
on the ground that the former the whole of an ellipse, never a segment of it
ii'^iiay,
t Tiie
lemma
is
used in
tlie
and not
mechanical proof only (Prop. 16 of the treatise) which depends on Eucl. x. 1 (see p. Ixi, Ixiii).
:
The Greek
of this passage is
AUCHIMKDKS.
without
satisfied
tlie
lix
lemma.
As
it
therefore
my work now
it
pulilishi'd
to,
has
the same
test as
the propositions
referred
have
by means
of meclianics
The
proof.
first
(
and
0,
and send
and next
also as
^ ;^
demonstrated by geometry.
^ ^).
The remainder,
and
appear only as
that
no doubt accounts for what might at first seem strange, they do not appear in the Conies of Apollonius.
1.
viz.
1/ Qq
be the base
the
any
other point
on
Qq in
QP
in F,
and
the tangent at
in E, then
(1)
QV.VO = OF:FR,
QO
.Oq
(2)
= FP:POf.
(
Here
is
avev
must be wrong and that the Passive would seem that should have been used. * According to Archimedes' definition the height (%) of the segment is " the greatest perpendicular from the curve upon the base," and the vertex "the point (on the curve) from which the greatest perpendicular
it
^
is
4>'
()
below
drawn."
The
vertex
therefore
P,
the
extremity
of
the
diameter
bisecting Qq.
t These results are used in the mcchanicnl investigation of the area of The mechanical proof is here omitted both because it is more lengthy and because for the present purpose the geometrical proof given
a parabolic segment.
is
more germane.
Ix
THE
To prove
(1),
to I'V,
meeting
QP
in K.
Now
therefore,
J'V
DV^QV
:
JiW;
by
jjaralleLs,
PQ PK=PQ' :PF\
In other words, PQ, PF,
..
:
PK are
= =
in continued proportion
'.
PQ PF-^ PF
PF + PQ QF:KF;
:
PF
tlierefore,
by parallels,
QV VO^OF
:
FR.
To piOve
(2),
we obtain from
QV
Also, since
qO =
OF
OR.
TP = PV, EF=^
OF.
Qq:qO^OE:
or
It is clear that the equation (1) of axes of coordinates
OR,
QO .Oq^ER: RO.
above
is
equivalent to a change
(as axis of
.'.;,
say)
For,
if
QV=a, PV
d'
nd
,-e
if
QO =
(1)
X,
RO = y,
_ _ _ Xa
OF OF - y
OF
y ~
(2fi
'
^'
"
whence
Zcutlieu points out
2-~
j/y
y
x).
'
(p.
and
(1)
(2)
ARCHIMEDES.
and either
of
Ixi
Thus the
first
represents the equality of the rectangles formed, two and two, from
'
the fixed lines Qq, PV, PQ and Gq (where Gq is the diameter while the second represents the same property with through q)
;
respect to the lines Qq, (the diameter through Q), he a dianiPter bisectiny in J/, and 2. If
QD
QT ami
Gq.
be the
RM
QV
RW
ordinate
to
For
RW;
and
3.
PV=iPW, PV=^RM.
PQq
is
The triangle
greater than
PQq
is
equal to half
contained by Qq, the tangent at P, and the diameters through Q, greater than half the segment.
parallelogram
q.
It
is
therefore
Cor.
It follows that
a j^olygon can
he inscribed
Jess
in the segment
For,
if
we
continually take
away an area
we can
4.
clearly,
x. 1).
With
the
same
V,
PQq
is
equal
to eight
RPQ.
it
RM bisects Q
Now
and therefore
bisects
PQ
(in Y, say).
is
parallel to
r,
1^
PQq =
5.
Prq, similarly.
1/ there
tJie
be
sei'ies
and
of areas A, B, C, D... each of which is four if the largest, A, is equal to the triatigle A, B, C, D... will be
less
PQq, then
snm of
than the
A PQq
:^
A PQR =
Pqr,
j
PQq =
A,
A PQR
In
like
APqr =
B.
triangles similarly in-
and so on.
Therefore
is
A +B+C+
J)
Given the
seft'ies,
series
A, B, C, D...just described, if
be
tlie
last
of the
then
A + B + C + ...+z+\z=yA.
Ixiv
Now,
exceeds
l)y
A ^
-C -
...
by an area
it
less
than
follows that
yl+j5 +
is
C+...+X
is
which
is
impossible,
by No. 4 above.
/i",
it
the segment
8.
thr equilibrium
{)
=
A'
=^
PQq.
of the second
Book
of the treatise
On
manner indicated
" If in
in Nos. 2, 4
it,
polygon in a parabolic segment after the and 5 above, and enunciates certain
in the following passage
:
a segment bounded by a straight line and a section of a light-angled cone a triangle be inscribed having the same base as
the segment and equal altitude,
if
remaining segments having the same bases as those segments and equal altitude, and if in the remaining segments triangles be
be said
inscribed in the recognised
manner
{the
in the segment.
)
to
Atul
(1)
it is
plain
of the segment, and the next pairs of
that the lines joining the two angles of the figwe so inscribed
to the vertex
by
diameter of the
segment,
(3)
and
that they will cut the diameter in the proportiojis of (he
successive
odd numbers,
the
the
reference
[the
length adjacent
And
places (ev
])y
Archimedes by
;
means
of the
and
But
tiie
been carried out, or at least Ave know of no lost work of Eutocius Archimedes in whicli they could have been included.
proves them by means of Apollonius' Conies, as he does not appear
to have seen the
lirst
two are
easily derived
work on the area of a parabolic segment from No. 2 above (p. l.\i).
but the
ARCHIMEDES.
The
since
third
may be proved
as folloAvs.
figure
If QiQjQoQ^PQ^Qofl/ly
<?,<?,,
a-
Qj/.,
are
:
all parallel
PI',
PV^
PV.^
- -^, .
and bisected by /'K,
...
:
Ixv
we
lia%e,
16
whence
it
follows tliat
PF,
VV,^
,^,
V.J\...
....
=
9
-(/'
1:3:5:7
QQ'
be
by the diameter
and if PV is of constant length, then the areas of tL triangle PQQ' and of the segvtent PQQ' are both constant tvhatever be the direction of QQ'
V,
II.
C.
Ixvi
If
CONICS.
BB'
is
the particular segment whose vertex is A, so that bisected perpendicularly by the axis at the point If where
if
A.y^PV, and
No. 3 on
p. liii)
Also, since
:
AN = PV,
QV BN-=p
..
:pa\
Hence
and
Therefore the triangle
is
BN=QD. BN.AN=QD.PV,
AABB'
PQQ' is
= APQQ'.
of constant area provided that
FV
of given length.
is
equal to ^.
/\PQQ'
7, p. Ixiii].
[No.
is
also constant
ellipse is to that
of a
circle
whose diameter
is to the
equal
minor axis
[This
11.
circle
is
On
The area of an
is d,
h is to that
of a
whose diameter
as ah to d^.
5.]
The areas of
their axes
;
ellipses are
to
under
ellipses are to
and Cor.]
of the applications of
e.g. for
and
others,
known
as vcuacts*.
The word
(p.
vtvci^,
commonly
is
satisfactorily.
Its
meaning
best gathered
says
it
C70)
"
being produced,
As particular cases
of the general
Two
"A
semicircle
ARCIIIMEDKS.
is
IxvU
Sp/it're aiifl Ci/Rii'ler,
whore
the problem
to cut
segments
may bear
The book On
e.g.
Props. 8 and
9,
in
:
If be any chord of a circle a any point on the circumference, to draw through in D and the circle again in straight line OBP meeting and such that DP is equal to a given length. Though Archimedes
and
we may
infei
that he obtained
it
by
means
of conic sections*.
full
(Greeks
position, or
Thus a line has to be laid across two given lines or curves so that it passes through a given point and the portion intercepted between the Unes or curves is
equal to a given length.
veOais
but this
fails to
must pass through a given point, just as the Latin iuclhiatio (and for that matter the Greek term itself) does not explicitly express the other requirement
that the intercepted portion of the line shall be of given length.
*
Cf.
PART
.
I.
CHAPTER
THE AUTHOR AND HIS
ACCOUNT OF THE
COXICS.
Ptolemy Euergetes (247-222 B.C.), that he flourished under Ptolemy Philopator, and that he went when quite young to Alexandria, where
he studied under the successors of Euclid.
to
We
Pergamum, where he made the acquaintance of Eudemus, to whom he dedicated the first three of the eight Books of the Conies. According to the testimony of Geminus, quoted by Eutocius, he was greatly held in honour by his contemporaries, who, in admiration of his n)arvellous treatise on conies, called him the "great geometer*." Seven Books only out of the eight have survived, four in the original Greek, and three in an Arabic translation. They vere edited by Halley in 1710, the first four Books being given in Greek with a Latin translation, and the remaining three in a Latin translation from the Arabic, to which Halley added a conjectural
restoration of the eighth Book.
J. L.
Books have recently appeared in a new edition by Heiberg (Teubner, Leipzig, 1891 and 1893), wliich contains, in addition to the Greek text and a Latin translation, the fragments
TJie first four
of the other
the commentaries
lOiitocius.
works of ApoUonius wliich are still extant in Greek, and lemmas of Pai)pus, and the commentaries of
The quotation
(ed.
is
See ApoUonius
Heibein) Vol.
p. 170,
Coy/cs.
Ixix
Books V.
to
VII. since the monumental edition of Halley, except as first few propositions of Book V., of
which L. M. LudAvig Nix published a German translation in 1889*. For fuller details relating to the MSS. and editions of the
made to the Prolegomena to the second volume of Heiberg's edition. The following is a literal translation of the dedicatory letters in which Apollonius introduces the various Books of his Conies to Eudemus and Attalus respectively.
Conies reference should be
1.
Book
I.
General preface.
"
" If
in other
I too
am
tolerably well.
When
t(j
you in Pergamum,
my
Avork in conies
book which
my
satisfection.
you that I undertook the investigation of this subject at the request of Naucrates the geometer at the time Avhen he came to Alexandria and stayed with me, and that, after Avorking it out in eight books, I communicated them to him at once, someAvhat too hurriedly,
telling
my
Avithout
sailing),
a thorough
revision
all
(as
he
was on the
Wherefore
it
point
of
intention of returning to
them
I noAv take
gradually corrected.
But, since
some other persons also Avho have been Avith me have got the first and second books before they Avere corrected, do not be surprised if you find them in a different shape.
*
fiinfte
de
wbich however goes no further than the middle of the 7tb Book v. and ends ou p. 32 in the middle of a .sentence with thu words " gleich dem Quadrat von " The fragment is nevertheless valuable in that it gives a new translation of the important preface to Book v., part of which
proposition of
!
Ixx
"
INTRODUCTION TO APOLLONIUS.
Now
first
introduction
)
will
the
first
will
solid
most and
here.
In
explanation of
[]
perhaps be best to quote Eutocius who speaks of " that which does not admit that the proposition is general, but says when
it it
and how and in how many ways it is possible to make the required construction, like that which occurs in the twenty-second proposition of Euclid's Elements, From three stniinht lines, irJiich are equal to three {licen straight lines, to
conntruct
a
it
is
than
178].
[p. 30], in
says
who propounds
it is
['
and,
if
]
;
determine at
is
it
possible."
Instances of the
common
enough.
Euclid
vi. '27,
which gives the criterion for the possibility of a real solution of the proposithere expresses the fact that, for a real tion immediately following the
solution of the equation
.r(((
.v}
= b-,
it is
Again, we find in Archimedes, On the Sphere and Cylinder [p. 214], the remark that a certain problem " stated tiius absolutely requires a but, if
certain conditions here existing are added,
it
Many
that, as
, ."
-\
but
it
is
to be observed
as in the cases just quoted, but, closely connected therewith, the determination
of the number of solutions.
word
in tlie preface to
Book
considered.
number
That Book deals with the two conies it follows that, when
;
for determining
latter, the
by their intersections
of solutions
number
to
comes
admit of a solution.
t'OXICS.
Ixxi
when I had discovered had not worked out the synthesis of three and four lines, but only a chance
for it
my
additional
The
the
of a
sections of cones
circle
:
it contains other matters in ad<Jition, none of which has been discussed by earlier writers, concerning the number of points
in
circle
meets
(7)
"The
TrXeov) with
limits
conic problems.
< rat's
but I cannot help thinking that Halley gives the truer reading, for the following
reasons.
(1)
. .
^- ,
;
^,),
is
and the
last
with determinate
Heiberg's
The contents
Book
iv.
show that
the
sense
is
not really
and
it
is
scarcely
conceivable
that
AiJoUonius,
in
less
new
(2)
complicated question.
the plural
There
is
" a section of words are a cone with the circumference of a circle and opposite branches with opposite branches." Thus to combine the reading of our text and that of Pai)pus would give a satisfactory sense as follows "in how many points a section of a cone
:
,
may
where the
[resiiectively]
on the corresponding
literally it
+
implies
iv. given below. has baen translated " more advanced," but
extensions
of
the
subject
beyond
translates
"ad abundautiorem
.scientiam
'ulterius progrediuntur."
Ixxii
ixriiODUCTiox
apollonius.
it
"
When
all
will of course
cis
be open
to those
judge them
they individually
please.
2.
Preface to Book
II.
"
good health,
it is
well; I too
am
moderately
have sent
my
book of municate
studies.
to
my
it
collected
to
if
Peruse
it
carefully
and comin
those
who
are
such
And
whom
introduced
you
in
visit the
neighbourhood
of
to him.
Take care
of
your health.
3.
"
"
Some time
;
ago, I
to
Eudemus
of
Pergannim the
books
first
three books of
my
my
Avorks.
book.
how many
meet
in
points at most
on the sup-
and further
how many
points at most a section of a cone and the circumference of a circle meet the opposite branches [of a hyperbola] *
^.
words
^
is
first ijreface
and
seems
to
me
next sentence.
(1)
(2)
of conic sections with one another or with a circle, of sections of a cone with the double-branch hyperbola, of circles with the double-branch hyperbola.
to specify separately, as essentially distinct questions,
(3)
But
Heiberg "s
(2)
and
AUTHOR AND
HIS
(Ut.vjrs.
Ixxiii
others of a similar
Now
the first-named
({iiestion
Conon expounded
to
foul of him.
which cause Nicoteles of Cyrene with some reason The second matter has merely been mentioned
in
by Nicoteles,
capable of demonstration
it
so
de-
to
it
And
all
Avhich I
many and
various
one.
The
is
of great service
<; 8<;).
can be
re
On
made
:
of the discoveries of
Conon
for determinations
for,
of limits
in
which opinion he
is
mistaken,
all,
even
if it is
possible, Avithout
using them at
events afford a
he mentions a
I
{
TO
(1)
method
of Apollonius.
When
it
is
i\\(i\j/ii
and
it is
impossible,
think, to imagine
intersection
.
the
him drawing
(2)
and
(3)
or
think
it
is
of
and
the
make
We
should
circle,
(2)
branch hyperbola,
(3)
and
cases
e.f).
may
where the curves
toiicli
at
Ixxiv
INTRODUCTION TO Al'Ol.LONlUS.
solutions are possible or that they are so many in number, and again that no solution is possible and such previous knowledge secures a satisfactory basis for investigations, while
;
).
4.
of determinations of limits
( \
from
such
sake of
the
laid
Be
8io-
Moreover, apart
for
usefulness,
they are
worthy of acceptance
themselves, in the same
the
as
demonstrations
in
way
for
mathematics
for this
and
no other reason."
"
"
book
have
down
propositions relating
to
straight lines.
and have
only proved what straight lines touch the sections and, convc'rsel}^
tangents.
first
what properties they have in virtue of which they are For my part, I have proved these properties in the book (without however making any use, in the proofs, of
them
which
I treated of
The
have
of
seem worthy
*
own
I
sake.
Farewell."
closelj'
In
the
Geiiiiiiu translation of
Ixix,
note].
The
prefaces to Books
vi.
and
('OXICS.
Ixxv
ApoUonius to Attains, greeting. " I send you the sixth book of the conies, which embraces propositions about conic sections and segments of conies et{ual and unequal, similar and dissimilar, besides some other matters In particular, you left out by those who have preceded me.
will find in this
"
is
to
to be
And
fully
these matters
in
truth
have treated
before
somewhat more
and
who wrote
Farewell."
ApoUonius to Attalus, greeting. " I send to you with this letter the seventh book on conic sections. In it are contained very many new propositions concerning diameters of sections and the figures described upon
"
them and all these have their use in many kinds of problems, and especially in the determination of the conditions of their possibility. Several examples of these occur in the determinate conic problems solved and demonstrated by me in the eighth book, which is by way of an appendix, and which I will take
;
Farewell."
The
of his
first
own work
explicit distinction
it.
(as we may say) more we are careful not to undei-stand tlie relative terms "elementary" and "advanced" in the sense which we should attach to them in speaking of a modern mathematical work. Thus it would be wrong to regard the investigations of the fifth Book as more advanced than the earliei- Books on the ground
;),(^.),
now
IxXVi
INTRODUCTION TO APOLLOXIUS.
;
differential calculus
of
normals to a
difference
many
other
The only
is
that, while in the case of the parabola the investigation is not very
difficult,
make
grasp.
exceptionally large
The
real distinction
demands on a geometer's acuteness and between the first four Books and the
theory of conic sections as
of
fifth consists
and
the subject in
is
true of
Book
what were considered the and their scope was that prescribed by tradition for treatises intended to form an accepted groundwork for such special applications as were found e.g. in the kindred theory of It would follow that the subjectsolid loci developed by Aristaeus.
Tlius the
first
essential principles;
matter would be for the most part the .same as that of earlier
treatises,
though
it
of Apollonius to
scientific,
One
effect of the
would be possible
than where new ground was being broken. In the latter case the advance would be more gradual, precautions would have to be taken with a view to securing the absolute impregof Apollonius greater conciseness in a treatise
Books
and one
minute
detail.
We
a
find
this
if
contrast in
Apollonius' Conies;
in fact,
treatment
properties
of
of
small
comparison
with Books
v. VII.
between the two divisions of the work
The
is
distinction, therefore,
may
\.,
THE
AUTIIOli
AND
UTS
Ixxvii
special
pendium
of conic sections
series
of
monographs on
For the
third
first
four Books
be seen
tliat
number
the
in the fourth
Book about
inter-
secting
is
for
the rest he
works on Pappus that in his first four Books Apollonius incorporated and completed the four Books of Euclid on the same subject. Eutocius, however, at the beginning of his commentary claims more for Apollonius than he claims for himself. After quoting Geminus' account of the old method of producing the three conies from right cones Avith difierent vertical angles by means of plane
conies.
more
This statement
('^)
(still
purporting
to
quote Geminus),
"
of
tlie
Again he
says,
made
plane."
to
Eutocius,
than right cones, had the same properties as the curves produced in
the old way.
we
a cone
and Archimedes
meet
all
may
be oblique) are
an acute-angled cone." And it cannot be supposed that Archimedes, or whoever discovered this proposition, could have discovered it otherwise than by a method which would
either circles or "sections of
duced
in the same general manner as elliptic sections, which Archimedes singles out for mention because he makes special use of them. Nor (2) can any different conclusion be drawn from the continued use of the old names of the curves even after the more
general
method
of producing
is
nothing
Ixxviii
iNTRonrt iroN
first,
apollonius.
hesitation might well be felt in
and secondly,
it is
not
thought strange,
e.g. in
and
represented by
Hence we must
is
in
might lead to the supposition that every hyperlx)la could be produced as a section of any cone) rests on a misapprehension, though perhaps a natural one considering that to him,
living so
much
later, conies
nius only, so that he might easily lose sight of the extent of the
knowledge possessed by
earlier writers*.
At
it
making an
entirely
new
departure.
we
find
having used
synunetry
in other words,
'
the relation
AN.
A'N^.P'N'-
AN'
A'N',
and
QV
was proved.
*
PV.P'V = (const)
exact basis of
to have been some contusion in Eutocius' mind about the names panihohi, I'lUpse and hyperbola, though, as we .shall see, Apollonius makes this clear enough by connecting them immediately with the method of application of areax. Thus Eutocius speaks of the hyperbola
tlic
obtuse-angled right cone and the right angle at which the section, old way, is inclined to a generator) together exceed
angles, or because the iilane of the section passes
made
in the
{'(\\)
;
of the cone
beyond the apex and meets the half of the double cone beyond the apex and he gives similar explanations of the other two names. But on this intei-pretation the nomenclature would have no significance for in each case we could choose
;
((\\
two right
COXIf'S.
Ixxix
the most general section of an oblique cone, and proves directly from the cone that the conic has the latter general property with
reference to a particular diameter arising out of his construction,
which however
Then, in truly
original
is
scientific fashion,
diameter
is
equally
true
with
all
reference
any other
indica-
new (and
arbitrary) diametei.
Another
(.-)
of
is,
tliiiik,
that
minimum
straight lines
(i.e.
Book V. ApoUonius seems there to imply normals) had only been discussed
by previous Avriters in connexion with the properties of tangents, whereas his own order of exposition necessitated an early introduction of the tangent properties, independently of
normals, for the purpose of eftecting the transition from the original
easily under-
remembered that the ordinary properties of normals are expressed with reference to the axes, and ApoUonius was not in a position to use the axes until they could be brought in as particular cases of the new and ai'bitrary diameter of reference. Hence he had to adopt a different order from that of earlier works and to postpone the investigation of normals for separate and later
it is
when
treatment.
by the names
jjarabola, ellipse
and hyperbola
but
it
fundamental properties were stated by him, and which suggested the discovery or may have been known
whom
may
i.
sentative.
ApoUonius
that expressed by
y^-px, where the axes of coordinates are any diameter (as the axis of x) and the tangent at its extremity (as Let it be assumed in like manner for the ellipse and the axis of y). hyperbola that y is the ordinate drawn from any point to the
original diameter of the conic,
'
IXXX
rXTRODUCTION TO APOLLOXIUS.
.r,
is tlie
Apollonius' procedure
and
x.x,=p
(I
(1),
where d
if
is
from which
]),
then y-
is
equal to a
rectangle of breadth
p,
and d
in other words,
or
7/''=;+^.'
Thus
for the ellipse or hypei'lx)la
(2).
an equation
it
is
obtained Avhich
differs
and
line
y* is less or greater
is
of being equal to
The
is
is
equal
it,
applied to
and neither
it,
falls
short of
it
nor overlaps
is
tlie ellipse
t(j ]>
in
applied
but
falls
or overlaps
respectively.
In Archimedes, on the other hand, while the parameter duly appears with reference to the parabola, no such line is anywhere
ellipse
or
is
=-2^
.
X
it
a;,
a;,'
ellipse, either of
is
is
and
two
areas,
COXICS.
Ixxxi
is
his predecessors
were acquainted with the equation of the central conic in the form
in which
ApoUonius gives
in other words,
whether
tlie special
use
for one side and equal in area to was new in ApoUonius or not. On this question Zeuthen makes the following observations.
(1)
The equation
form
had the advantage that the constant could be expressed in any shape which might be useful in a particular case, e.g. it might be expressed
either as the ratio of one area to another or as the ratio of one
if
the consequent in
d,
be the parameter
(2)
p.
not,
as
rule,
connect his
well-knoAvn method of application of areas, yet the practical use of that method stood in the same close relation to the formula of
Archimedes as
it
(say)
X.
is
C,
ax + x
and, in one place {On Conoids
^=^. =
and
Spheroids, 25,
p.
(3), '
^
420) where
all
(
by ax +
(3)
x^.
x.x^
form in Avhich
it
appears in
(3),
speaking of
it
as
an
^),
in
The equation
(const.)
represents
as a
mean
pro-
portional between
H.
C.
of x^,
which
Ixxxii
last
INTRODUCTION TO APOLLONIUS.
(i e.
a;,
is
Whether
appeared as an auxiliary
there
is
no
sign), or the
is
differently
(4)
made,
immaterial.
The
differences
may
regard Apollonius
of conies
Greek theory
and had
which
Thus, where Archimedes chooses to use projwrtions in investigations for vhich Apollonius prefers the
areas which
to
method of application of more akin to our algebra, Zeuthen is most inclined think that it is Archimedes who is showing individual peculiis
arities rather
predecessors
than Apollonius, who kept closer to his Alexandrine a view which (he thinks) is supported by the
Book
I
II.
is
portions.
cannot but think that the argument just stated leaves out of
step in the
own fundamental
equation.
be regarded as
Archimedean form can hardly a personal variant from the normal statement of
Further,
to
^ = X.Xi
(const.), '
^
and to speak of this as having the advantage that the constant may
l)e
who never
and uses
is
in question,
it
in the case
form
-=
COXICS.
Ixxxiii
Now
the equation
ax
or
_/_ = !'
x^
a-
'
y = ~ .X a
we take some
line equal
to
and erect
it
at that extremity of
it Avhich is on the curve. Therefore, for the purpose of arriving at an expression for y* corresponding to those obtained by means of
of the con-
Archimedes.
Again,
it
is
Archimedean form
i.
of the
fundamental property
ellipse,
he retraces
and proves it again as a deduction from a procedure which suggests a somewhat forced those definitions adherence to the latter at the cost of some repetition. This slight awkwardness is easily accounted for if it is assumed that Apollonius was deliberately supplanting an old form of the fundamental
21 [Prop.
:
8],
property by a
new one
The idea that the form of the equation as given by Apollonius was new is not inconsistent with the fact that the principle of of areas was older than the
on any other assumption.
Euclidean theory of proportions
for surprise
if
;
]
:
difiicult to
explain
geometer as Apollonius intentionally harked back and sought to connect his new system of conies with
so orthodox a
Apollonius, says
by a square
cone exceeding by a square, and in that of a right-angled cone There is evidently some confusion neither deficient nor exceeding."
{ ),
(p.
new
definitions of
674)
"
For a certain rectangle applied to a an acute-angled cone becomes deficient in the section of an obtuse-angled
/2
IXXXIV
INTRODUCTION TO APOLLONIUS.
no question
is
equal to y* exceeds or
falls
The
description
"deficient,
would appear that Pappus somehow confused tlie two forms which the two writers give the fundamental property.
It will be observed that the " oppo.sites,"
.r,
appearing in the
;
ciSet
so that it
in
being
).
4],
They are
it is
first intro-
duced
curve.
in the proposition
I.
14 [Prop.
but
in
i.
16 [Prop. 6]
first
of
their
was reserved
Apollonius.
(1)
This
view
is
The
the double-branch
hyperbola as
curves
and a comparison between the works of Apollonius and Archimedes (who does not mention the two branches of a hyperbola) would lead us to expect that the greater generality claimed by
;
itself,
if
about the
in
(2)
of the
is
two
attested
by the fact
the single branch, he yet continues throughout to speak of them as two independent curves and to prove each proposition vith regard to them separately and subsequently to the demonstration of it for
the single curves, the result being a certain diflTuseness which might
if
the
first
<'(>\/cs.
Ixxxv
and
tlie
same time
if
had then taken as their basis the generalised property. As it is, the difluseness marking the separate treatment of the double
hyperbola contrasts strongly with the remarkable ingenuity shown
by ApoUonius
property
three
in compressing into
common
curves together
to
handed down by
;
tradition,
whereas,
ApoUonius had to add new matter with reference to the double hyperbola, it would naturally take the form of propositions supplementary to those affecting the three single-branch curves.
It
may
I.
38
{d')
[Prop. 15]
makes use
by
the relation
d^
d'
_P
" d'
the parameter of the
where d
is
ordinates to
The actual
i.
16 and
17.
The
idea
may
common
The reference
special
to
Book
and
does not
call for
any
The
(p.
new
sense,
and
Archimedes
The
loci,"
"many
tained in
Book
III.
"the synthesis
of solid
and goes on to refer more particularly to the "locus with It is strange that in the Book respect to three and four lines." itself we do not find any theorem stating in terms that a particular
geometrical locus
is
a conic
section,
Ixxxvi
theoi'ems
INTRODUCTION TO APOLLONIUS.
stating conversely
that
all
points
certain property.
The explanation
of this is probably to be
when
it
was a
and the ground for this may have been that the subject was extensive enough to require a separate book. This conjecture is supported by the analogy of the treatises of Euclid and Aristieus on conies and solid loci respectively, where, so far as we
on
conies,
of such loci
and the
minimum
if
not
Book VI.
is
about similar conies for the most part, and Book VII.
values.
Book VII. and VIII. in the preface to the former suggests that the Book VIII. contained a number of problems having for their
functions
of
lengths
have
given
values.
These
and
it is
possible in the
CHAPTER
II.
GENERAL CnARACTERISTICS.
1.
language.
has
made mathematicians
ponent parts or stages, cannot be better described than in the words of Proclus. He says*: "Every problem and every theorem which
is
complete with
all its
definition^
conclusion
is
{..).
is
(),
enunciation
construction
{^),
setting-out
(),
all
setting-out
proof
(),
what
is
{<;),
Now
that which
The
marks
off"
what
and adapts
is
it
The
the
and makes
clear
thing
which
sought.
The
construction adds
datum
what
is
scientifically
knowledged
facts.
The
These are all the parts of problems and theorems, but the most essential and those which are
confirming what has been demonstrated.
found in
all
For
it is
tliis
equally
necessary to
know beforehand
Avhat
is
should
be demonstrated by means of the intermediate steps and the demonstrated fact should be inferred ; it is impossible to dispense
Proclus (ed. Friedlein), p. 203.
what
by means of a concrete figure, of what the enunciation states in general terms as the property to be proved or the problem to be solved.
follows,
really
is
As
will
appear from
means a
closer description,
Ixxxviii
INTRODUCTION TO APOLLONIUS.
The remaining parts are often Thus
brought
there
is
in,
left
When
is,
then
do
there
ciation
is
wanting?
;
The answer
for,
when
nothing
(jiven in
the eyiunciation
what
is
sought,
this is not
states only
what
is
sought,
what must be knoAvn or found, as in the case of the problem just mentioned. That problem does not, in fact, state beforehand with vhat datum Ave are to construct the isosceles
i.e.
(simply)
that
we
are
then,
is
the enunciation
contains both
(Avhat
is
and
so
datum but
is
also, as, in
In
fact,
in defining the
required to find an
is
But that
is
what the
is
entmciation stated.
is
no
definition
is left
The constituent parts of an Euclidean proposition be readily identified by means of the above description without further details. It will be observed that the word /Aos has here a different .signification from that described in the note to p. Ixx above. Here it means a closer definition or description of the object aimed at, by means of the concrete lines or figures set out in the instead
of the general terms used in the enunciation
rivet the attention better, as indicated
TLva
6
^ .
of the
(';
and
its
word
which the possible solutions of a problem are subject is also described by Proclus, who speaks of determining " whether what is
GENERAL CHAHACTEUISTICS.
sought
is
the
in Euclid as in
/
cviii).
is
Ixxxix
practicable and in
far
it is
In ApoUo-
sometimes inserted in the body of a problem as in the instance ii. 50 [Prop. 50] given below in another case it forms the
;
subject of
li.
52 [Prop.
ii.
51], the
the same
way
as the Stopta/xo's in
vi,
Eucl.
vi.
enunciation of
28
(see p.
is
and
Proclus
thus
it
as clearly as possible.
In order to
to reproduce a
they appear in Apollonius, and accordingly the following propositions are given in full as typical specimens, the translation
on the
letters
as
III.
54 [Prop. 75
Trepi-
, (, (, ( ('(' / ,
)/
((
8e
section of a cune
Tois
Koi
npos TO avTo
fxjOi'iai.
drawn
to
TO
vov
( 8(.(
e';(et
by the
inter-
tK Te
ov
((
of contivct
[1] of that
the ratio
compounded
tlie
inner SOg-
* Proclus, p. 202.
xc
TO fVTos
INTRODUCTION TO APOLLONIUS.
, ( ((( ((' (.
npos
Tas
8(,
ment
and
ttJs
ment, and
[2]
bears to the
fourth
line
part
of
the the
(
,
;
, ' , . ', , (
\
, \,
,
yap
( -(( ({ /
,
\ 8 (( \ ((
,
\ \ (( ,
(-
square
on
the
joining
points of contact.
eVt
(\
and QT, Q'T tangents, and let QQ' be joined and bisected at V, and let TPV be joined, and let there be drawn, from Q, Qr parallel to Q'T and, from Q', Q'r' parallel to QT, and let any point R be taken on the curve, and let QR, (^R be joined
or the circumference of a circle
and produced to
/,
r.
say that
e;^e(
.
(
toG
)
has to the square on Q(/ the ratio compounded of that which the
(,((( '
AV
8,
(\
'
) )
,,
, ,
*
MB AM
( ,,
MB
KG
. ,
^
VP has to the square on and that which the rectangle under QTQ'*h!ifi to the fourth part of the square on QQ', i.e. the rectangle under Q VQ'. For let there be dra\vn, from R, KRWR'K', and, from P, LPL' parallel to QQ' it is then clear
square on
PT
that
LL'
is
is
a tangent.
VQ',
Now,
since
also
QV
equal to
LP
is
OS
'
MB,
,\
S\.
bi
equal to
PL' and to WK' and R]V to WR' and KR to R'K'. Since therefore LP, LQ are tanand
KW
is
MB,
gents,
to
KRK'
drawn
parallel
is
, ,
TO
.,
), ,
"the
rect.
LP,
as the square on
QL
to
LP, that is, the rectangle under LPL', so is the square on QK to the rectangle under R'KR, that is, the reotiingle under K'RK.
the square on
. (( / , ,
TO Se
GENERAL CHARACTERISTICS.
LQ
is
to the square
on LQ, so
is
the
((
KQ
to the
aerjuo
is
i.s
,
,
npos
of c^fi
, .
", ,
8f
square on
i\s
KQ
therefore
c.v
, , ,
KQ to the
But the
has to
ratio
KQ
K'RK the
, (('( ^, , AM ? ( ) ,, AM \) ?
/
f\fi.
compounded of that of Q'K' to A'7?, that is, oirQ to QQ', and of that of
'
QK to
A7i, that
is,
of
r' '
to Q'Q,
, AM
/
.
. ', . , ,
be
which is the same as the ratio which the rectangle under r'Q', rQ has to the square on Q'Q; hence, as the rectangle mider L'Q', LQ is to the rectangle under L'PL, so is the rectangle under r'Q', rQ to the square on Q'Q. But the rectangle under Q'L', LQ has to the rectangle under L'PL (if the rectangle under
L'TL
compounded
BE
QL
has to the
L'TL and the rectangle imder L'TL to the rectangle imder L'PL; hence the rectangle
under
r'Q',
Q'Q the
ratio
QL
to
L'TL
to the
But, as the
QL
is to
the the
rectangle
under IJTL,
so
is
square on
VP to
L'TL
.so is
is
under
7*,
the
rectangle under
Q'TQ
to the rect-
on Q(^ the ratio compounded of that of the square on PV \x> the square
on
PT
Q'TQ
to the rectangle
INTRODUCTION TO APOLLONIUS.
50 [Prop. 50 (Problem)].
II.
,
/ ^'/;
8<)(!
iVi
((,
yap
(. " ,\
\ (((
be
section of a cone
, ( (
\
6
draw a tangent to a given which shall make with the axis towards the same parts with the section an angle
equal to a given acute angle.
C of
be
;
PC
P^V be perpendicular
8(
fOTi
'.
contained by
,.
\
CNT to
the square on
is
given, for
it is
the same as
doOe'iaa
And
PN to
each
given.
^'
( 8. 8(
8(
'
eV
the square on
NT is given,
TNP
for
is
. ^
8 8
.
CNT to
;
the square on
NT
is
given
is is
CN to NT
angle at
And
the
given
therefore also
the angle at
C is
given.
Thus with
in posi-
CN [given]
PC
;
has been
therefore
drawn
PC
fore
is
[given] in position.
is is
Also the
;
section
[given] in position
given.
there-
And
;
the tangent
.(( ( (. ( (
(
therefore
PT
pro-
[given] in position.
((
8((
\
.
LC
of the
bo drawn
will
then
the
;
PT
duced
Let
it
meet
it
asymptote.
tivai
meet
will
in
L
it
((
LT^^ LCT.
for
Therefore
given
- (< ( ,
8(
]
( (\
)
bodflaa
GENERAL CHARAiTERISTICS.
xcm
( ,
]
. ,
iVel
, ,\
.\
, ,
trrt
asymptotes.
Thus
lem
will
proceed as follows
let
the
angle
FED, and
let
the angle
be equal to the angle ACZ, and let AZhe drawn from A at right angles
,,
(\
to
J.l',
and
let
any point
let
be
(\
8(
.,
a perpendicu-
(
.
(
Trpof
\ \oyov (( \ \
\oyov
^8 (
,, .'
, ^ ^
'\ ,
'
?(( . ( ,
\oyov
;
I)F be drawn from it upon EF. Then, since the angle ZCA is equal to the angle ffEF, and also the
angles
a,t
A,
F are right, as
to FIT.
CA
is
to
AZ, so
to
is
EF
But
EF has
it hiis
FIT a
;
to
FD
therefore also
CA
has to
to
AZ a
FD.
EF has
Hence
CA
has to
\.
the square on
a greater ratio
e\fi
than the square on EF has to the square on FD. And, as the square on C.i is to the square on AZ, so is
the transverse to the erect
;
.
(
,
therefore
EF
, ,
\, (
.
EF.
Let
it
KFE;
and
let Z) A'
((( ( \
,
)
,\
.
"
( . (
\(
KF
is
greater
than the rectangle under KFE, the luis to the square on square on FD a greater ratio than the rectangle under KFE has to the square on
KF
is,
FD,
that
the square on
CA
to
And
if
we make,
,\
KF to
the .siiuare
CA
to
INTRODUCTION TO APOLLONIUS.
( ,
apa
eoTt
Kcu
fWi
apa,
TO
HMK.
irpos
another
.smaller
are;i,
,
(
, , . KM , ' , )\ \ \, ? ' . ,
8e
,, .\
*
jJ
\, ,
;
. .
AZ; and
make the
and
the angle
angle
made equal
therefore
the
angle
DKF;
let
CP
it
will
Let
it
cut
at P,
and from
and
the
7*iV perpendicular
therefore
triangle
PCN
a.s
is is
similar
to
DKF.
on
Therefore,
the square
is
CN
to the square
on NP, so
the square on
KF to
is
the square on
FD.
under
the erect, so
on NP and the rectangle under KFE to the square on FD. And conversely, as the square on PN is to the rectangle under CNT, so is the square on DF to the rectangle under
CNT
to the square
KFE;
thereft)re ex
aequo, as the
square on
CN
is is
to the rectangle
under CXT, so
to the rectangle
fore, as
the square on
KF
to
under KFE.
to
There-
CN
to
is
NT,
so
is
is
KF
NC,
FE. was
as
But
is
also, as
;
PN
so
is
to
so
DF to FK
therefore ex aequo,
NT,
And
the angles at
N',
equal to
the angle
DEF.
it
may
not be
grammatical forms
is,
from
extreme import-
ance.
u.se
we should have to say " let a perpendicular be drawn " or, more peremptorily, "draw a perpendicular," the Greek expression is
been
e.g.
GENERAL CIIAUACTERISTICS.
the former Avord expressing in
or "suppose
itself
drawn"
like.
and the
used very effectively along with the usual expression for a propor-
KE, EM, which can ; hardly be translated in English by anything shorter than " Let
tion, e.g.
,
tJie
it
Neatest of
,, '
drawn," and similarly
the word
it
XCV
it //are
;^,
all is
done."
HK
be so taken that
is
to
as
to
KE."
possible to abbreviate
(),
is
line
;;
or
is the aiujle
understood
contained by
or
-),)
AB
and
the angle
or
is
is
AB
(^),
by AB,
AB
that
square on AB.
is
The
result
is
much
The
tradition
Euclidean
and the corresponding propositions in Euclid's third Book about circles. The following are some obvious examples.
Eucl.
III.
1.
Ap.
II.
45,
iVf/j-
(.
ToO
KfVTpov tvpuv.
" (.
( .
Eucl. in.
fVi
2.
/,
4,
7repi0epiiaf
fVi
evuela
(,
'
I.
10.
Eucl.
' . (, ( ( . . ,
.
II.
( ( (
pev fVi
(vdeui
( ((.
26.
tj
,
ntpi-
iv (Kkti^ft
XCVl
Eucl.
INTRODUCTION TO APOLLONIUS.
III.
7.
.
of an ellipse
V.
4 and
6.
If a point be taken
(
(yyiov
''
Se
(, 8
('
(.
As an may
Thus
ae\
equal to
if
Tivts,
and
from
drawn
8e
from the given point will be that which is equal to half the latus rectum, the greatest the remaining part of the axis, and of the rest those which are nearer to the least
will
re-
mote
instance of Apollonius' adherence to the conceptions of
first
Book
of the Conies
be mentioned which
in
first
line be
I. 17 we have the proposition that, if in a conic a straight drawn through the extremity of the diameter parallel to the
fall
without
This
the conic
is
drawn that
it is
a tangent.
argument recalls the Euclidean definition of a tangent to a circle as " any straight line which meets the circle and being produced does
not cut the
circle."
We
8( ( " \, ( (
( (((
((
/.
^ (( (((
), \
32.
.(
an
els
tvuda
(-
Another instance
fact that,
is
found in the
when enunciating
In this he follows
"a
GENERAL CHARACTERISTICS,
plane figure bounded by one line."
that the word circle alone
circle, e.g. in
is
XCVll
It
is
Euclid
iv.
16 and Apollonius
37.
2.
Apollonius, like
dovn
as
This plane
property
theory,
indeed when, by
way
necessary to prove that a cone can be found Avhich will contain any
given conic.
As
(p. xxi), it is
Menaechmus
- ay,
y^
- bx,
xy = ah,
and, in like manner, the Greek geometers in general seem to have con-
it
was
in their
known
loci of
Hence finding
was understood as being synonymous with localising it in a cone, and we actually meet with this idea in Apollonius i. 52 58 [Props. 24, 25, 27], where the problem of " finding" a parabola, an ellipse, and a hyperbola satisfying certain
a particular conic
Menaechmus and
his
contemporaries would
defini-
regard the
loci
really curves.
When
by
of
38
a sort
There
is
no reason to suppose that the method adopted in the was diflferent. We know from Pappus that
])y
their original
names
whereas,
.
if
(as
H.C.
^''^^^^""
V5
.___>.lll
{UKIVERSITT.
XCviii
INTRODUCTION TO APOLLONIUS.
title might be thought to imply) he had used in his book the methods of solid geometry, he would hardly have failed to discover a more general method of producing the curves than that implied by We may also assume that the other predecessors their old names.
the
method
for
among
time
his
the properties of conies which were well-known before there are many, e.g. the asymptote-properties of the
way
(2)
it
would
ellipse
Thus
it
and then
by projection
Definite order
liave
and aim.
Some Avriters
hap-hazard
author's mind.
way without any definite plan having taken shape in the This idea may have been partly due to the words
down everything
is
having put
as
it
occurred to him
but
it is
clear
Books Avhich
had been communicated to various persons before they took their Again, to a superficial observer the order adopted in the final form. first Book might seem strange, and so tend to produce the same
impression
;
then
it
Book
to the connexion of
and returns at the end of the particular conies with the cone, which is
But,
if
the Book
is
examined more
object
is
aimed
made use
wards
but for the time being they are simply links in a chain of
GENERAL CHARACTKRISTIC'S.
xcix
kind of circular cone are identical with those which are produced from sections of cones of revolution.
The order
is
as
118.
First,
of the conic
which
is
the
referred
to the
particular
diameter which emerges from the process of cutting the cone, and
the tangent at
its
Next, we are
and the original diameter. Then follow properties of tangents extremity of the original diameter and (2) at any other point of the curve which is not on the diameter. After these come a series of propositions leading up to the conclusion that any new diameter, the tangent at its extremity, and the chords parallel to
it
tween
(1) at the
new diameter)
have to one another the same relation as that subsisting between the original diameter, the tangent at its extremity, and the ordinates
and hence that the equation of the conic when referred to new diameter and the tangent at its extremity is of the same form as the equation referred to the original diameter and tangent*. Apollonius is now in a position to pass to the proof of the
to
it,
the
He
proceeds to propose
tlie
ellipse, or
hyperbola,
when
and the corresponding parameter are given, or, in other words, when the curve is given by its equation referred to given "Finding" the curve is, as stated above, regarded as axes.
ordinates,
it
he
first
ordinates are at right angles to the diameter and solves the problem
for this particular case, going
then,
secondly, he reduces the case where the ordinates are not perpen*
The
up
to this point
is
attested by a formal
recapitulation
introduced by Apollonius himself at the end of i. 51 and concluding with the statemt-nt that " all the properties which have been shown
to be true with regard to the sections
will equally result
by reference
are taken."
9^
INTRODUCTION TO APOLLONIUS.
tlie diaiiieter
dicular to
by
is
his procedure
that
it is
at right angles
is
by
it.
Thus what
is
proved here
Book.
not the
mere converse
been
all
If that
had
that
hoAvever, that
is
succeed
in
proving,
concurrently with
of
the
other curves than can be derived from right cones, and that
conies have axes.
all
The contents
the
treatise
from being a
intention
Avith
a definite
throughout.
In like manner
it will
as,
however,
it is
Books
shall
CHAPTER
III.
As
may be
Any
he remembers Avhat In
use.
thought,
it is
employed by the Greek geometers were much more extensively used than they are in modern geometry, and were consequently handled
by Apollonius and
to
his
contemporary readers
witli
much
greater
a modern mathematician.
Hence
it
in a piece of algebraical
it
The principal machinery used by Apollonius as well as by geometers comes under the head of what has been not inappropriately called a geometrical Algebra; and it will be
tlie
earlier
convenient to exhibit the part which this plays in the Conies under
the following important subdivisions.
(1)
This theory in
in the fifth
Cll
INTKonUCTIOX TO APOl.LONIUS.
;
ApoUoiiius
it is
suffices to
show how
far
Thus
obvious that
supplies a ready
method
of effecting the
operations of
multiplication
and
division.
Again,
suppose,
for
example, that
we have a
cti,
...
,,
so that
We
have th
\aj
of the
a^
method
of proportions enables
(cf.
an
the
sum
The application of
areas.
355 B.C.) or not, it is due to Eudoxus of Cnidus (408 no doubt that the method of application of areas, to which AVe allusion has already been made, was used much earlier still. have the authority of the pupils of Eudemus (quoted by Proclus on
presents
there
is
Euclid
I.
(17
tc
eX\enj/i<;),
Avhere
we
The problem
in Euclid
i.
44
is
" to
apply to a given straight line a parallelogram which shall be equal to a given triangle and have one of its angles equal to a given
rectilineal angle."
The
means
of
addimj together or subtracting any triangles, parallelograms, or other figures which can be decomposed into triangles. Next, the second Book of Euclid (with an extension vhich is
found in
vi.
27
29)
su^jplies
means
problems of
modern algebra
second degree, and provided, so far as the solution of quadratic equations is concerned, that negative and imaginary solutions are
excluded
in
THK MKTlloDS OF
it is
<
)!,!,(
)XIUS.
ClU
figures,
to
where one solution by algebra would cover both cases. Book ii. of the Elements makes it possible multiply two factors with any number of linear terms in each
It is readily seen that
by
That theorem itself supplies a method of dividing the product of any two linear factors by a third. The remaining operations for Avhich the second Book affords the means are, however, the most important of all, namely,
the aid of the a]rplication-i\\QorQn\.
(a)
is
equal to that of a
given rectangle
14],
which ])roblem
is
the
geometrical solution
ii.
of
5, 6.
In the
E, making
as centre
first
case {a)
to
we produce
the side
bisect
BE equal
BC
then
circle
meeting
CB
produced in G.
Then
Also
FG'^FB'+BG\
FG' = FE'^AB.BE^FB ',
common FB",
BG- = AB.BE.
=(
if
(1).
and
BG
or
is
found.
(6)
we
have,
is
divided cijually at
/J.
Dli +
CD' - Cn-.
[Eucl.
II.
').J
Now
suppose
All -a,
UB-x.
CIV
INTRODUCTION TO AJ'OLLONIUS.
Tlien
ax a* =
the area of the
rect.
A
h^,
if
gnomon
is
given (=
say),
and
if
is
given
ax
is,
-b'
(a) to
in the
{b')
and
CV
possibility of a
clear that
it is
a necessary condition of
(
tlic
?:
and that
tlu
geometrical solution derived from Euclid does not differ from our
practice of soh'ing a quadratic
side
a;'
and
to this
method and
to the old
way
Let
an
ellipse.
.He
says, "
[i.e.
equal to
CB-] be applied
opposite
square
"
of
the
ellijjse
two geometrically must not however be understood from The contrary may be this that he was unaware that there are two solutions. stating the inferred from the proposition vi. 27, in which he gives the
* It will be observed that, while in this case there are real solutions,
It
b-^l-\
for,
ment,
in
the text translated by Simson, of the two cases where the base of the
is
applied parallelogram
greater
and
less
the dis-
and we must therefore assume that, in the case given above in the text, Euclid was aware that x = AD satisfies the equation as The reason why he omitted to specify the former solution is no well as x BD. doubt that the rectangle so found would simply be an equal rectangle but on BD as base instead of AD, and therefore there is no real object in distinguishing
two solutions.
This
is easily
tlie
equation as a
()
statement of the problem of finding two quantities whose sum equations (//-) are given, i.e. as equivalent to the simultaneous
and product
x + y = a,
x)j
= b\
really only
apparent solutions are simply the result of interchanging the values of .r and ij. This form of the problem was known to Euclid, as appears from Prop. 86 of the
Data (as translated by Simson) " If two straight lines contain a parallelogram given in magnitude, in a given angle if both of them together be given, they
:
shall each of
them be given."
x^i^ax = b'^
From
have
cvi
INTKODUCrioN To ATOI-LOXIUS.
Again, from the proposition in Euclid
ii.
6,
ha\e,
if
is
bisected at
and produced
to
JJ,
AD.Dn + CB'^CD\
C/
"'
figure,
AB - a, BD = x.
Then
and,
if
AD.DB = ax + x\
ax +
1/
is
equivalent to finding a
gnomon equal
in area to 6*
and having as
line
CB
or -
Thus ve know
- j and
/>',
and
we have to find, by the Pythagorean proposition, a square the sum of two given squares. To do this Simson draws BO at right angles to vl^ and
0,
equal to
equal to
circle
CO, and describes with centre C and radius CO a produced in D. Thus BD, or x, is found. meeting A
joins
Now
AD. DB
+ CB--^
CD-
= C0'
= CB' +
whence
or
B0\
A1).DB = B0\
a;
+ ,*
6*.
This solution corresponds exactly to Apollonius' determination of the foci of the hyperhola.
THK MKIHODS
The equation
can be dealt
If
witli in
AI'OLI.OML'S.
6"
CVil
x'
ax =
a similar manner.
if
AB^a,
then
and
solved,
so
that
AD - X,
,t
ax =
AM = the gntnnon
we have
its
CMF,
(0
by
w hich
Thus we can
find
(and therefore
AD,
innnediately preceding.
Hence Euclid has no need to treat this case separately, l)ecause is equal to AD the same as the preceding except that here instead of BD, and one solution can be derived frou) the other. So far Euclid has not put his propositions in the form of an
it is
In vi. 28, 29 has in ii. 5, 6 supplied the means of solving them. however he has not only made the problem more general by substituting for the sqttare by Avhich the required rectangle is to
exceed or
fall
the
first
case
of
(the deficiency
possibility
[vi.
condition
/09
27]
the obvious
ax
-=
h'.
Of the problems
problems, to the
in vi. 28, 29
first of which the 27th prop, is necessary, are the most general and useful of all in the elements, and are most frequently made use of by the ancient geometers in the solution of other problems and therefore are very ignorantly left out by Tacquet
;
and Dechales
who pretend
that they
propositions
strange that, notwithstanding this observation of Sinisun's, the three vi. 27, 28, 29 are omitted from Todhunter's Euchd, which contains
:
"
We
sixtli
Book
solution which
Euchd
now to be never required and have been condemned as useless by modern commentators see Austin, Walker, and Lardner."
;
various
once restored to the I would suggest that all three propositions should be at text-books of Euclid with a note explaining their mathematical significance.
CVlll
INTRODUCTION TO AI'OLLONIUS.
of these propositions are as follows*
:
The enunciations
VI. 27.
"
Of
all the
parallelograms ajrplied
to the
same straight
applied
line
and
and
similarly situated to
that
which
upon
the
half of the
to the
half and
"
similar to
VI. 28.
To a given
a parallelogram equal
and deficient by a jmralMogram similar to a given jiarallelogram : But the given rectilineal figure must not he greater than tlie parallelogram applied to half of the given line and
to
a given
rectilineal figure
similar
to
tloe
defect.
VI. 29.
"
To a given
to
a given
rectiliiieal figure
and exceeding by a
parallelogra7n similar
to
a given one."
Corresponding propositions are found among the Data of Euclid.
Thus Prop. 83
states that,
^'
If a parallelogram equal
to
a given
states
given in species, the sides of the defect are given," and Prop. the same fact in the case of an excess.
It
is
8-4
propositions,
worth while to give shortly Euclid's proof of one of these and vi. 28 is accordingly selected.
The
1883-8).
CIX
the
is
to
be similar.
Bisect
similar
AB
at
JE,
and on
describe a parallelogram
OEBF
it.
D
;
[by
vi.
18].
Then, by the
follows that
AG
to
or greater than
it
problem
solved
if
the latter,
the parallelogram
EF is
greater than C.
Now
construct a parallelogram
LKNM
EF
if
and similarly situated to D [vi. 25]. Therefore LKNM is similar and similarly situated to EF, while, GE, LK, and GF, LM, are homologous sides respectively,
over
C and
similar
GE>LK,
and
GF>LM.
GF) equal respectively to Make GX (along GE) and GO LK^ LM, and complete the parallelogram XGOP. Then GPB must be the diagonal of the parallelogram GB [vi. 26]. Complete the figure, and we have
(along
EF = C + KM,
and
by construction,
XO = KM.
gnomon EliO,
is is
equal to C.
is
Hence
equal to C.
AB -a, SP = x,
and that
c is the ratio of
;
KN,
then have,
we
ax,
, = m.- -,
so that
ax
b
c
C . =
in
tln^ ('(juation
ax +
- X' c
ex
If
INTRODUCTION TO APOLLONIUS.
we compare
these equations witli those by which Apollonius
viz.
it is
is
that
instead of
particular
parallelogram
;;
is
taken
whose
is
sides
are
/>,
d.
at right angles to d.
Subject to these
to
the
similar
that of
"
Euclid
(5 )
"applied to
figure
"a
"having as
its
width
similar
to
that contained
by the
be seen from what has been said, and from the book
that Apollonius
is
nothing
if
in the
Avhich
is
made
of
we may conclude
areas are stated by Apollonius without proof, though they are not
is
that his readers as well as himself Avere so imbued with the methods
of
work out any necessary intermediate step for themselves. And, with regard to the manner of establishing the results assumed by Apollonius, we may safely infer, with Zeuthen, that it was the practice to prove them directly by using the procedtire of the second Book of the Elements rather than by such combinations and transformations of the results obtained in that Book as we find in The the lemmas of Pappus to the propositions of Apollonius. kind of result most frequently assumed by Apollonius is some
able to
relation
line divided
by points on
it
into a
method
and
his
number of parts, and Pappus' amounts practically to the proniore likely that Apollonius
it is
contemporaries would,
after
the
manner
of
yeonietrical
algebia,
draw a
in
is
and squares,
e.g.
and thence,
many
cases
one rectangle
equal to the
that
THE METHODS OF
AI'Ol.LOXirS.
clear.
CXI
in.
An
instance
it is
will
make
that,
this
if
In
Apollonius
2G
[Prop. 60]
assumed
E,
B, C,
he points on a line
in the order
named, and
if
we
set
oft'
EB, and
A' along
EB'
lemma
equal to
29 [Prop. 61]
is
more
likely to
Book
(3)
iliary lines.
The Greek geometers were fruitful in devices for the compression sum or difference of the ai-eas of any rectilineal figures into a single area and in fact the Elements of Euclid furnish the means The Conies of Apollonius of effecting such compression generally. contain some instances of similar procedure which deserve mention
of the
;
There
is,
first,
the abscissa
*
x.
AVhile the
is,
in
On
the other
it,
and
to proceed semi-algebraically.
hand Pappus' method is simply to draw a line with points on Thus in tliis case [Lemma 4 to Book ni.,
first bisecting BC CE.EB + BZ-^ = EZ\
p. 947]
he proceeds as follows,
in Z.
and
while
It follows that
whence
(and
CE.EB = DE
A + CA.A B,
CA = BD).
CXll
INTRODUCTION TO APOLLONIUS.
neatness
is
line.
The Cartesian equation of a central conic referred to any diameter of length d and the tangent at its extremity is (if (/' be the length
of the conjugate diameter)
,
d"
-d"
and the problem is to express the right hand side of the equation in the form of a single rectangle xY, in other words, to find a simple construction fur }' where
^
Apollonius' device
is
d"
_d"
such that
to take a length
_ (d~'d''
(so that
is
length
d).
If
PP' be
x,
and
of
PL
perpendicular to
PP' and
2, 3),
Then,
if
PV = x,
_
Vli
and
if
VB drawn
parallel
PL
meets P'L
in
we have
by
similar triangles,
p_VB
so that
d~ P'V~d + x'
VP ^]) +
-
=
and the construction
Again, in
for expressing
v.
y"^
Y,
for
is
therefore effected.
81],
1-3 [Prop.
is
used
as base
drawn perpendicular to
'
is
an ordinate to the
of length equal to
axis.
AM
is
and
(where p
is
is
A A'), and
CM
joined.
7W meets CM in
/
//, it is
2 (quadrilateral
MA Xll).
Cxiii
area equal to the difference between the squai-e on a normal PG and the square on P'G, where P' is any other point on the curve
than
P'.
The method
measure
XG
is as foUoAvs. If is the ordinate of P, along the axis away from the nearer vertex so that
PN
:
in //, as
GH
is
now
second auxiliary
line.
NG - Nil,
NGH,
N'GH'.
NV = 2
P'G'
Hence, by the aid of the expression for y^ above, the areas PG'
figures,
and
it is
proved that
-PG' = 2A HKH',
effected
so that
difference
of the
by means
two
CM, GH.
(4)
Book
VII.
The seventh Book investigates the values of certain quadratic functions of the lengths of any two conjugate diameters PP', DD'
in central conies of different excentricities,
to the
values of
The
whole procedure of Apollonius depends upon the reduction of the ratio CP' CJ)'^ to a ratio between straight lines MH' and Mil, where //, //' are fixed points on the transverse axis of the hyperbola
or on either axis of the ellipse, and is a variable point on the same axis determined in a certain manner with reference to the
position of the point P.
:
The
proposition that
:
Book
is
The bearing
of the proposition
may
INTRODUCTION TO APOLLONIUS.
coordinates are the principal axes of the conic, and
if a,
h are the
we
have,
e.g.,
cp.,ci>'
^<"^-^-) -{(!)"
CP*-GD*
where
.'>,
-(!)}
of the curve,
we
obtain
CP'-CD'
Apollonius' procedure
is
whose
coordinates are {x
0),
whence the
equal to
and we have
h
_a'-b'
(i)>
and
"'=4.
2
(2).
From
(1)
we
derive at once
whence
AH
A'll
or
^p^iAA'.
CXV
:
we have
only to divide
'
in the ratio p
AA'.
AA',
This
is
what
1' is similarly
ratio
2\i
'.
and
clearly
AH = A'H',
(2),
A'H=AH'.
f
Again, from
we have
,
a\
a'
In other vords,
or
A A' \A'M=CT:
CN A'M:AM=CN:TN
draw
(3).
AQ
is
parallel to
the tangent at
AQ
bisected
is
by
CP;
to
and, since
AA'
is
bisected at C,
it
follows that
A'Q
parallel
CP. Hence,
if
QM'
be the ordinate of
Q,
CPN
AQM',
TPN
coincides with
A'M':AM'=CN:TN.
(3),
it
appears that
M'
or, in
Q by
the construction
Since
that
now
//,
//',
x',
h vere so determined
CP' + CD'
x'
GP'-CD'~
it
A'
follows that
CP""
:
CD'' =
x'
+ h:x'-h,
:
or
The construction is similar for the ellipse except that in ^^' is divided externally at H, H' in the ratio described.
2.
The use
of coordinates.
of the most characteristic features of the Greek treatment of conic sections. The use of coordinates is not peculiar to Apollonius, but it will have been observed that the same point of view appears also in the earlier Avorks on the subject. Thus Menaechmus used the characteristic property of the paraljola which
We
y'
px
He
equation xy =
c*,
of
CXVl
INTRODUCTION TO APOLLONIUS.
of equation for the parabola,
mode
of
central conic
~ = (const.)
can easily be put into the form of the Cartesian equation.
its
extremity as axes
this relation,
general
and proceeds
to deduce
from
when
found, the
His method does not essentially differ from that of modern analytical geometry except that in Apollonius
geometrical operations take the place of algebraical calculations.
We
in the
have seen that the graphic representation of the area of yform of a rectangle on
to PP',
as base, Avhere
(;r,
y)
is
any point on
P'Z
PL
as rectangular axes
is
That an equation
because
of this
x,
repre-
of his separate
Book work on plane loci the following proposition " If straight lines be drawn from a point meeting at given angles two straight lines given in position, and if the former lines are in a given ratio, or if the sum of one of them and of such a line as bears
find in Pappus' account of the contents of the first
:
we
is
=h
are positive.
is
a, b
The
whose base
is
equal to y^
CXvii
.auxilitary line is
determined with
is itself
referred,
has only to be borne in mind that, in order to show the area y' as
it
a rectangle,
and
}'
should
be right.
But, as soon as the line P'L was once drawn, the object
vas gained, and the subsidiary axes of coordinates vere forthwith dropped, so that there was no danger of confusion in the further
development of the theory.
Another neat example of the use of an auxiliary line regarded from the point of view of coordinate geometry occurs in i. 32 [Prop. 11], where it is proved that, if a straight line be drawn from
the end of a diameter parallel to
its
the
Apollonius
first
from
ordinate
KQV
is
drawn.
their
Then,
if y',
y be the ordinates of
A',
respectively,
and
common
and tangent as
23, 24)
axes,
we have
?/''>?/*
or xY,
where
PL before
Let
along
y'^
(with
PP PL
,
be equal to
xY\
so that Y'
>
7,
and
let Y'
be measured
Y').
is
(so that, in
to,
VR -
Y,
and YS =
Then the
the straight line P'L, and the locus of the extremity of Y' for
different points
on
Y',
PK
is
Pti.
It follows, since
PS intersect,
that there
where
Q',
F=
and therefore
PK
respectively,
that
PK
curve between
cannot lie and A". Hence and the curve in the manner supposed. Here then we have two auxiliary lines used,
viz.
Y^P+'^x,
d
and
= mx,
CXVIU
INTRODUCTION TO APOLLONIUS.
where
is
some constant
determined
;
of intersection of
PK and
the conic
is
b}'
auxiliary lines
found by a
it is
necessary to bear in
original equation
illustrated
by the
where
several
xy, y^, X,
up into parts Avhich are seen to be, constant multiples of and y, where x, y are the coordinates of any point on
;
and we have
there-
many
and the
modern method is that the Greeks did not making the fixed lines of the figure as few as
form as
possible.
but rather to
cc^,
xy,
. .
number
of terms.
is
compression
efiected
by means
of one or
auxiliary lines.
In
and it appears that the properties between areas) can be equally well expressed relatively to the two auxiliary lines and to the two original axes of reference, we have clearly Avhat amounts to a
to the original axes of coordinates, of the conic (in the
form
of equations
transformation of coordinates.
3.
Transformation of coordinates.
case
is
A simple
ellipse,
first
found as early as
i.
15 [Prop.
5],
may
with
by a simple transference
THE METHODS OF
Al'OLLONIUS.
Cxix
We find in fact,
of
as
i), the end of the conjugate diameter. an intermediate step in the proof, the statement
its
(0
where
x,
the rectangle
RT.TE
diameter and
for diameter
y are the coordinates of the point Q vith reference to the its conjugate as axes and the centre as origin and
;
is
and
where
_d
'
The equation
as axes
is
its
conjugate
I.
41
CP,
it
CV QV
QV
is
to the
compounded of the ratio of CP to the other side of the parallelogram on CP and of the ratio;? d, then the parallelogram on QV is equal to the diiierence between the parallelograms on CP, CV. Suppose now that the parallelograms on CP, CV are squares, and therefore that the
in the ratio
:
parallelogram on
(^
is
a rectangle
it
follows that
,
fdy
=
S.y
(1).
Apollonius
is
now
the tangent at
its
extremity.
as
use the
line.
new diameter
CQ meets
the tangent at
in E, as in the figure of
47 [Prop. 21];
CXX
then,
to
if
INTRODUCTION TO APOLLONIUS.
from any point
it is
on the curve
tiie
ordinate
if it
7? IF is
draAvn
PP\
PE,
and,
meets
CQ
in
F, the triangles
is
CPE,
CWF
CW,
-we
are similar,
in
each
two similar
i.
parallelograms on CP,
41
we take a
triangle on
RW
R WP
is
WU lying
along
WP is
RW
WU~
Apollonius
satisfies this
'
CP
d'
the tangent at Q.
The proof
as follows.
i.
From
37 [Prop. 14],
QV'
CXXl
Cr. It then follows that the two triangles 7? 11', CFW same relation to the original axes, and to the diameter have to the new axes, consisting QQ', as the triangles RFM, of QQ' and the tangent at Q, and to the diameter PP', respectively.
have
tlie
CUM
Therefore, in order to prove that a like relation to that in (2) similarly determined with
Q and
the diameter
',
it
has to
be shown that
CUM^ AEMF.
CPE,
i.
see note on
in.
[Prop. 53].
We
(2) above,
or the quadrilateral
acqt-acfw^apuw, QTWF=ARUW,
from
to
new axes
of coordinates consisting of
Now
is
PF by the
relation
so that
'
CXXIl
INTRODUCTION TO APOLLONIUS.
-alue for q should accordingly
be given by
i.
50 [Prop. 23].
essentially the
No
;
it
may be
of effecting the
same transformation
(p.
liii)
There
of
all,
areas.
is another result, and that perhaps the most interesting which can be derived from the foregoing equations between We have seen that
7?/=
so that
i.e.
,RF
aCFJV,
aCPjE,
AEUW+ aCFW=
CFRU^ ACPE.
if
the quadrilateral
Now,
a variable point on
in fixed
the curve,
we
observe that
RU,
are
drawn always
CPE is
constant.
and
if from
any point
the area
on
dravm
to
PF, QQ'
meeting QQ',
PP in F, U respectively,
of the quadrilateral
then
Conversely, if in a quadrilateral
moveable jjoint
in given directions
ami meeting
if the quadrilateral has a constant area, then the locus of the j)oint
is
an
ellipse or
a hyperbola.
is
a conic.
But, as he says in his preface that his work contains " remarkable
loci,"
we must
CFRU
is
constant,
converse
way
to him.
CXXlll
GFRU
is
constant
is
axes
is
+7/ + / = ,
is
where
a,
y,
are constants.
the equiva-
and tangent
to another diameter
in other
from
pair of
2)<^i'>'
conjugate diameters
conjugate diameters
cu7've referred to
to
hij
the equation
second
of
the
means of
the
e.g.
by
constancy of the area of the quadrilateral CF'R'U' for any point R'
RU
as axes.
The axes
of reference
may
CFRU^ CF'R'U',
or the other relations derived immediately from
it, viz.
F'lRF^ lUU'R',
or
FJR'F'^JU'UR,
iii.
3 [Prop. 55].
The coordinates of R' would in this case be R'l, R'J. Similarly an equation can be found corresponding to the property
in
III.
HFQ = quadrilateral
IITUR.
Again, in. 54, 56 [Prop. 75] lead at once to the "locus \nth respect to three lines," and from this we obtain the well-known equation to a conic with reference to two tangents as axes, where
the lengths of the tangents are h,
k, viz.
'HD'
CXXIV
INTRODUCTION TO APOLLONIUS.
latter equation can also be derived directly
The
from
in.
41
same proportion.
Thus,
axes,
if X,
if
and
y be the coordinates of Q with reference to qR, qP as qp = x^, rq =?/, (cf. the figure of Prop. 65), we have, by
the proposition,
CXXV
Method
of finding
remembered that Menaechinus' solution of the problem of the two mean proportionals was eifected by finding the points of intersection between any two of the curves
It will be
.r*
ay, y^ =^bx, xy
ah.
It
lie
is
two curves
on the
x^ + y'
bx
ay =
0,
and therefore that the two mean proportionals can be determined by means of the intersection of this circle with any one of the three curves.
Now,
attributed to ApoUonius,
we
two parabolas lay on the circle. on the authority of loannes Philoponus* (who quotes one Parmenio) that ApoUonius solved the problem thus. Let the two given unequal straight lines be placed at right angles, as 0, OB.
section of the
We
have
it
Complete the parallelogram and draw the diagonal OC On OC OBC, produce OA, OB, and
through
C draw DCFE
in
(meeting
OA
and
OB
E)
so that
DC ^
FE.
postulate unjn'oved."
Now DC=FE,
*
and therefore
DF= CE.
is
On
II.
The passage
Vol.
CXXVl
INTRODUCTION TO APOLLONIUS.
circle
on
OC
OD.DA=FD.DC
= =
..
CE.EF
OE.EB;
(1).
OD:OE = BE:AD
OD:OE=CB:BE
= OA:BE
Also, by similar triangles,
OJ):OE = OA: AC
=
It follows
:
mean
is
(2).
(3).
from
(1), (2)
and
(3) that
AD
proportionals.
we may conclude
in the
above figure was determined by draAving a rectangular hyperbola with OA, OB as asymptotes and passing through C. And this is
the actual procedure of the Arabian scholiast in expounding this
solution.
Hence
it is sufficiently
Avas obtained
on
OC
as
i.e.
by the
+ y^ bx ay xy
"
solution attributed to Apollonius is given by In this solution M, the middle point of OC, is taken, and with 3i as centre a circle has to be described cutting OA, OB such that the line DE passes through C produced in points D,
The mechanical
Eutociust.
and
this,
i-^der
about
as
it
of D,
OA OB) from
in. p. 56.
are equal.
ycip
Pappus
6o\oyo0nes CTepebv
opyaviKuii
7G 78.
.\\
elvai
11(,
CXXvii
other,
is
essentially the
same as the
because,
DC
be
made equal
to
FE
from J/ perpendicular to
This coincidence
as that
is
DE
nmst
bisect
and therefore
MD = ME.
is
the same
by loannes Philoponus to Apollonius except that Philo obtains the required position for DE by mov-ing the ruler Eutocius adds that this about C until DC, FE become equal.
attributed
solution
identical
Philo's
Tcpov),
is
method
because
tinuous parts,
It may be mentioned here that, when Apollonius uses the problem of the two mean proportionals in the Conies, it is for the purpose of connecting the coordinates of a point on a central conic
(
5.
more con'enient in practice dividing the ruler into equal and conis, by possible to watch the equality of the lines DC, FE
is
(?
(-
it
^)
trial
whether
MD,
ME are equal*.
i.e.
of
51,
52 [Prop. 99].
Method
a given point.
made
Without entering into details, for vhich reference should be to v. 58-63 [Props. 102, 103], it may be stated generally that Apollonius' method of finding the feet of the various normals passing through a given point is by the construction of a certain rectangular hyperbola vhich determines, by its intersections with the conic, the
required points.
The
Suppose
is
as follows.
FGO
Let
FN
Then,
OM
the
perpendicular from
if
on the
axis.
we take
Archimedes, Vol.
iii.
p. 70.
'
CXXVlll
INTRODUCTION TO APOLLONIUS.
if (x,
y), (,,
y^ be
the coordinates of P,
respectiA'ely,
we have
^
c,
NG
X-
y,
NG
Pa
-"'
-,';=o
(1);
xy
{^.-f)y-y.-^j
xy
{l+^)-x,y-..y,x = 0.
of
b'
The
intersections
these
feet
rectangular
of
respective
conies
give
the
the
various
normals passing
the ;;?'-
through 0.
Now
bola, as
Pappus
being unorthodox.
He
is
speaking
(p.
between the three classes of "plane" (tTriVcSa), "solid" the still more complicated " linear" problems
and
says,
means of by means
Such procedure seems a serious error on the part of is discovered by conies or higher curves, and generally when it is solved
yeVovs), as, for
"
(- ),
(),
;
and
example, the
problem in the
in the
Book
the parabola, and the solid vcwts with reference to a circle assumed
spiral
by Archimedes
first
for
it
is
possible
pounded by the
latter...."
The
allusion
must
clearly be to the
when
tions
affect
On
it is
this
doubt some
force,
because
CXXIX
l)y
c<jmbining
,,
by
Multiply
(1)
and we have
if,
--(x,-'|).^-f
=0,
But there
is
which could not have been expressed in the geometrical language which the Greeks used. Moreover we have seen that in Ajiollonius'
solution of the problem of the
and a
circle is found.
We
that Apollonius could have reduced the problem of the normals to a parabola in the same way, but that he purposely refrained from
doing
(1)
he
regarded
problem, and
making
it
a " plane
"
problem.
H. C.
CHAPTER
IV.
41.
If
of which
totiches
j)arabola,
extremities of
If in a central conic parallel tangents he drawn at the a fixed diameter, and if hoth tangents be met hy any variable tangent, the rectangle under the intercepts on the parallel tangents is constant, being equal to the square on half the parallel dia42.
meter,
43.
totes
i.e.
the
diameter conjugate
tangent
to
to that
Any
from
the
asymp-
whose product
is
is constant.
There
liave
tive propositions,
an obvious family likeness between these three consecuand their arrangement in this manner can hardly
It is true that in. 42 [Prop.
tlie foci
66]
is
of a
might be supposed that it had its place in the book for this reason only; but, if this were the case, we should have expected that the propositions about the foci would follow directly
central conic, and
after it instead of being separated from
68].
it
by
iii.
have also a strong positive reason for supposing that the arrangement was due to set purpose rather than to chance, namely the fact that all three propositions can be used for describing a conic by
We
means
the
of tangents.
Thus,
if
two tangents
first of
cxxxl
<,'ive tlie
the same means, those cases, namely, in vhich the fixed tangents
in a special
manner.
As
means
That Apollonius was in a position to solve this problem is proved by the contents of two of his smaller " (two Books On cuttiwj treatises. One of these, of a proportion), we possess in a translation by Halley from the Arabic under the title De sectione rationis the other, now lost, ' (two Books On cutting off a space, which means was
<;
cutting off from two fixed lines lengths, measured from fixed points
on the
lines
respectively, such
constant area).
Now
any number of tangents to a parabola reduces precisely to that which is discussed with great fulness in the former of the two treatises, while the construction of any number of tangents to the ellipse and hyperbola in accordance with the conditions of
III.
of the
if
points of contact P,
are given,
we have
point
(in
r,
a straight
J}
line
which
will
intersect
given
tangents
respectively) in such a
/*?
:
way that
rq
= qp pP,
:
or that
Pr
is,
Pq^qp :qR;
:
we must have Pr
qj)
= Pq
qR
(a constant ratio).
In
fact,
we have
to
draw a
line
tangent measured from the point of contact is to the intercept on the other tangent measured from the intersection of the tangents in
a given
first
ratio.
How
Book
<;.
to
do this
is
shown
If,
qP
in
r,,
r^
and the
fixed
tangent
qR
in
;;,, p,^,
we have
to
draw a straight
CXXXU
INTRODUCTION TO APOLLONIUS.
qR
=
r,,
jO,
respectively
which are
in a given proportion,
i\r
:
such
tliat
J)
?*,?.,
and
this
problem
is
Book
off"
<;.
is,
in that treatise
to
draw from
given proportion,
measured from two fixed points A, B, which are in a is to be e.g., in the accompanying figure,
OKM
drawn
so that
AM BN
:
is
a given ratio.
is reduced to a more special one in which the fixed point occupies a position B' on the first line so that one of the intercepts is measured from the intersection of the two lines. Tlie reduction is made by joining OB and drawing
B'N' parallel to from the point B' in which OB, MA intersect. Then clearly B'N' BN is a given ratio, and therefore the ratio B'N' AM is given. We have now to draw a straight line ON' cutting MAB', B'N' in points J/", N' such that
: :
B'N'
=- a,
given
ratio,
suppose.
This problem
is
solved in the
first
is
substantially as follows.
Draw OC
parallel to
MA
produced in C.
Now
suppose a point
found on
AM such
that
cxxxill
is
niado
to
we have
AM
CXXXIV
intercept B'N'
lines.
is
INTRODUCTION TO APOLLONIUS.
measured from
B', the intersection of the
two fixed
the
or
we
manner
in the case
where
is
GM.MD
,
CB'.AD
has
its
maximum
value.
Apollonius
, have We
,
B'C
CM
B'M
MD=AD=MA^
MD CM,
B'C _ C}1
whence, since
B'M
WIr'M~A'"WA'
and therefore
B' M"" = B'C.B'A.
Thus
is
also.
According, therefore, as
value of
solutions.
OC
At
the end
we
JB'A
B'C.
Thus,
if
we
B'A, B'N' as axes, and if Ave denote the coordinates and the length B'A by li, we have
of
by
[x,
y)
^J)
If
h + x-2>Jhx'
to
we suppose Apollonius
is
described
that in which
;
is
is
tangent at
for, as
above,
B'M _
B'A ~
B^
B'M
-^,
by
parallels,
CXXXV
same
N'M
if
respectively in
tlie
proportion.
Further,
we put
for
we
/i
k be those lengths,
k^
which
is
y h + x-2s/hx'
easily be reduced to the
as coordinate axes,
sym-
metrical form
'^(*;
7/Ni
'
II. In the case of the ellipse and hyperbola the problem is to draw through a given point a straight line cutting two straight lines in such a way that the intei'cepts upon them measured from fixed points contain a rectangle of constant area, and for the ellipse the straight lines are parallel, while for the hyperbola they meet in a point and the intercepts on each are measured from the point of
their intersection.
These are particular cases of the general problem which, according to Pappus, was discussed in the treatise entitled
and, as
we
cases
in question
were included.
,
()
line
work corresponded
particular
We
how
is
The reduction
is
case where one of the points from which the intercepts are measured
effected in the
same
manner
we should take
the point
represented by
OC AD =^
.
We
ON'M
so that
B'N'
.AM^OC .AD,
AD
B'N'
UcT-AJr
But, since B'N',
OC
are parallel,
B'N' _ B'AI
~0C ~
CM'
rx,.
Therefore
-4
J/
AD
DM
CXXXVi
INTRODUCTION TO APOLLONIUS.
B'M
is
MD = AD
B'C, which
is
given.
Hence, as
problem
/,
is
It
we
find in Apollonius
to the general theorem that, if a straight line cuts off from tivo fixed
straight lines intercepts,
respectively,
the first
lines.
measured from given points on the lines which contain a rectangle of given area, the envelope of straight line is a conic section touching the two fixed straight
Thus, suppose
conic and E,
BCD
F to
tangent
CD.
If a fifth
Therefore
CONSTRUCTION OF A CONIC
Conversely,
if
15Y
MEANS OF TAN(;ENTS.
cxxxvil
DC
A,
AP
We
EF
of the chords
find the
bisected
by
it,
AD
thus
we can
ordinate to
EF
of contact of
AD, and
hence
we can
diameter
EF and
first
Cf.
Lemma
XXV. of the
Book
investigations.
CHAPTER
V.
(
first
it,
seen, specially
mentioned in the
preface of Apollonius as a
is,
as
have
He says
adding
that he found that Euclid had not worked out the synthesis of the
locus,
successfully,
new and
p.
Apollonius,
confirm
We
must
to be found
it
is
propositions iu that
Book
way
in which he
went
to work.
it is
really a
But we have,
in fact, in
any point of a
to
conic there he
drawn
in fixed directions
conic
and
their chord
square on the third line is constant. where the two tangents are parallel is o])tained from iii. 53 [Prop. 74j, and the remaining three propositions, iiL 54-56 [Props. 75, 76], give the proof where the tangents
to the
drawn
The proof
CXXxix
find the
theorem of the
at
all,
any
is
It will
be seen from the note following Props. 75, 76 that this theorem
easily obtained
from that of the three-line locus as presented by Apollonius in those propositions but there is nowhere in the Book
;
to the former.
is,
may
that the theoretical converse was regarded as a mere appendage to it. But, from the nature of the case, that converse must presumably have appeared as an intermediate step in the inestigation of the locus, and it could hardly have
been unknown
Aristaeus,
even to earlier
geometers,
such
as
Euclitl
and
who had
and, in doing
so,
we have
to bear
mind that the problem must have been capable of partial solution before the time of Apollonius, and that it could be completely solved by means of the propositions in his third Book. We observe, in the first place, that iii. 54-56 [Props. 75, 76], which lead to the property of the three-line locus, are proved by means of the proposition that the ratio of the rectangles under the segments of any intersecting chords drawn in fixed directions is
constant.
is
a particular
having t\vo of
and it will be seen that the the rectangles under the segments of in-
that a conic
is
two
sides parallel.
For,
if
is
direction cutting
77^ =
If
^. (^""^')
to
//,
is
a chord
Cxl
LNTRODUCTION TO APOLLONIUS.
meeting the diameter which bisects chords parallel to
it is
DC
lines
met by
Rr
lie
in
on
drawn through A,
JJ
Then,
if
x, y, z,
u be
the distances of
shall
quadrilateral
ABCD, we
have
(const.).
^
yu
And,
since
ve
point
relation,
as
to
RI.RK
~
'
we
shall
have
R I .Ir
We must then try to find a conic where is a given constant. whose points R satisfy the given relation, but we must take care to
determine
condition
locus
is it
in such a
manner
as to
time that the points of the conic so found do really satisfy the given
;
for, of cour.se,
know
that the
a conic.
It seems clear, as shown by Zeuthen, that the defective state of knowledge which prevented the predecessors of Apollonius from
The
transition
was
in fact, in
Cxli
possible
cf)nipetence
of of
but the
ditiiculty in the
way
the earlier step was apparently due to the fact that the conception
of
sinjj;le
occurred
cordingly,
any one
the
case
before
\vhere
Apollonius.
the
four-line
His
ac-
locus
complete
of
through
locus,
all
and
of the
These extensions to
Ixxxiv
seqq.),
His predecesinit
and
will be seen
that
the converse,
the construction
of
The difficulty would come in where the conic was a hyperbola with two branches. Assuming, then, that the property of the four-line locus was established with respect to an inscribed trapezium by means of the
particular case present no difficulty to them.
what was wanted to complete the theory \vas extension to the case where the tangents are tangents
tangents,
ratio
the
op-
to
between the rectangles referred to in tliose cases where no tangent can be drawn parallel to either of the chords, or where a
tangent can be drawn parallel to one of them only.
(1) that
Now we
find
Apollonius proves the propo-sition for the case where the tangents touch opposite branches in in. 19 [Prop. 59, Case i.]. Also (2) the proposition in. 23 [Prop. 59, Case iv.] proves that,
cxlii
INTRODUCTION TO APOLLONIUS.
is
where there
no tangent
and
ii.]
and proves
that,
if
meets the
it is
not parallel in
parallel
to the
constant ratio
then tQ^:tq'.
(p. 140) that the method adopted for determining the complete conic described about a given trapezium ABCD,
Zeuthen suggests
which
is
, may
have
form of certain rectilineal figures connected Avith such a similar This procedure is exemplified in Apollonius, ii. 50-53 conic.
[Props. 50-52], Avhere a certain figure
is
determined by means of a
;
and the
suggestion that the same procedure was employed in this case has
it
We
()
two
ABCD
the value
lines in
Hi
Ir
of
segments of
different
of
I.
(which
if
the conic
is
the term,
excluded).
Let the points of the auxiliary figure be denoted by accented letters corresponding to those in the figure on p. cxl.
We
know
the ratio
and,
if
we choose any
we know
LlOCUS.
P\
(3)
bisected by the diameter, (4) a point Q' with the tangent at that
point.
Then the intersection of the tangent at Q' with the diameter and the foot of the ordinate to it from Q' determine, with P\ three points out of four which are harmonically related, so that the
remaining one, the other extremity
{})')
Hence the
II.
is
determined.
Suppose that the conic has no tangent in either direction. In this case we know the ratio between the tangents to the
manner
just
described
then, by
is
means
hyperbola
III.
determined.
in the direction of
In this
bi.secting
case, if the
tangent Pt parallel to
AD
meet in
t,
the half
parallel to
AB.
We
have then
tq
tif
If
we now choose
t'P' aibitrarily,
we
it,
(3)
two points q,
'
on the curve.
cxli
INTRODUCTION TO Al'OLLONIUS.
''^''6
If y^i Vi
.-Tj,
*^e ordinates of q^
.r/, a/
we have
is
determined.
The point
its
it
distances from
lie.
line
on which
must
IV.
tlie
direction of
AD.
;
Let the tangent at P, parallel to AB, meet the diameter bisecting BC, AD vat, and let tq parallel to AD meet the conic in q we then have
t'q
t'F'
If
we choose
(1)
either
t'q
or t'P' arbitrarily,
t'T',
we have
the diameter
(2)
in
t',
T' respectively,
Since
the tangent at
C't'
P',
.C'T' = \.a",
the length of the diameter.
where
is
a'
cxlv
[Prop. 42.] Therefore, by symmetry, T'q is the tangent at q. Hence we can find the centre C by joining ', the middle point of Pq, to 0\ the point of intersection of the tangents, since Y'O' must be a diameter and therefore meets t'T' in C The Thus the auxiliary conic can be readily determined. relation between the diameter a and the diameter h' conjugate to it
is
given by
tig*
.
tr ~
a* ~
;^ _
a'
Thus
it is
seen that, in
all
nius supply
means
is
for determining
that which
sought.
;
The
e.g.
made
in various Avays
AB,
is
given;
and we can effect the transition by means of the ratio between CA and CA'. There are, hovever, indications that the auxiliary figures would not in practice be used beyond the point at which the ratio of the
diameter
(a) bisecting
is
its
conjugate ()
determined, inasmuch as
we
find in
Apollonius
values of a and b
when the
ratio j-(=
-,-,
is
given.
The problem
to
be solved
is,
and
while
given,
its
well as
conjugate.
is
figure,
a straight line
drawn through
parallel to the
known
H. C.
cxlvi
INTRODUCTION TO APOLLONIUS.
and meeting
DA
produced in 0.
Also
let
OB
ellipse)
again in E.
OB.OE
whence
a?
OA.OD'b"
bisecting
The
line
BC
will
AVe have now, for the case of the ellipse, a proposition given by Apollonius which determines the value of a* directly. By
III.
27 [Prop. 61
(1)]
we know
that
OB' + OE' +
whence
a' is at
^;
{0' +
OD') = a\
once found.
ill.
24-26,
The construction in the case of the hyperbola is also facilitated by means of the asymptote properties. In this case, if the letters have the same significations as in the figure for the ellipse, we find the centre by means of the chord BE or by using the auxiliary similar figure. The asymptotes are then
(2)]).
60 and 61
or
If the required
AK.KD=lh\
curve
is
AD
in
K, Z,
,
of the first of the four
manner
show that P', the end of the diameter, is at the middle point of the intercept between the intersection of the diameter with the tangent at Q' and with the ordinate from Q' i-espectively. The curve can then be determined by the simple use of the
ordinary equation of the parabola.
been
means by which the determination of the locus with reference to a quadrilateral of any form whatever might have been
reduced to the problem of finding the locus with reference to a
trapezium.
As
Apollonius' third
cxlvii
was not
affected
by Apollonius'
the key in the
completion of the theory of the locus, but that the key must be
Zeuthen (Chapter
8) finds
Euclid*.
He
notes
first
tliat
Archimedes' proposi-
on
is
p. lix, Ix
which one
other sides.
(Thus
QT, while
at
infinity
PF
is
on the parabola in which they meet the two is turned about Q and takes the position turned about its intersection with tlie parabola
PQ
and takes the position of the diameter through Q.) This suggests the inquiry whether the same means >vliich are
used
also
to
effect
the
in
transition
in
this
be employed
the
more general
now under
consi-
deration.
As
lost, it is
necessary to
;
Pappus gives of their contents and the Porisms which is there preserved in its original
:
form
is
as follows t
tivo
If from
drmvn
straight
Hues which
on a
and
if one
of the straight lines so dra^vn cuts off from a straight line given in position a certain length measured from a given point on it, then the
other straight line also tvill cut off a portion
line hearing
from
another straight
a given ratio
[to the
is
former intercept^
true also
when a
four-line
locus
is
substituted for the first-mentioned given straight line and the two
any two fixed points on the locus. Suppose that we A and C, being two opposite the quadrilateral A BCD to which the locus is referred,
and suppose the lines from which the intercepts are cut off to be drawn respectively parallel to the sides A, EC of the CE, A
quadrilateral.
Let
*
let
AD, J J/ meet
is
That the Porisim of Euclid were a very important contribution to geometry them in Pappus (p. G48) as a collection most inRenionsly adapted for the solution of the more weighty problems
indicated by the description of
(is
t Pappus,
p.
. \>
).
(-
cxlviii
INTRODUCTION TO APOLLONIUS.
CE
in D',
M'
respectively, while
CD,
CM
meet
in
D",
M"
respectively.
distances of
distances from
BC,
AD
, CD .
parallel to
its
Then the
from CD,
BC
respectively
be equal to
^^ Li hd
and the
J/ from AB,
DA
will
be equal to ^
-fttt?,
AE D 31
Therefore the fact that the ratio of the rectangles under the
distances of
A BCD
is
constant
may be
D"3r
where
If
-mf^^AE =
CE
''^
say
.,.
(1),
the
/i is a new constant independent of the position of M. now be determined by means of the position of a point locus, we have
F of
D"M"
where F\
_ D"F" _
F"M "
^*'^'
F"
AF,
CE
and
of
CF,
AE
respectively.
And,
ABCF.
cxlix
reduce
it
was only
or
AD
AF
AF.
to any quadrilateral
in the
parallel
BC)
are denoted by
x, z,
from BC, A D
locus
is
(reckoned parallel to
A) are
Then the
= \.yii
(1),
where
If
/*,
is
a constant, and
x,
Avith reference to
BC,
as axes.
A D,
respectively,
u= =
Since (
PM PQ-MQ
from
(2).
MQ)
is
the distance of
measured
parallel to
A,
let it
be denoted by w,
figure,
Therefore, from
(1),
we
derive
""
'=-cjr'y^
to take a point
on
AE such
'
that
D'E _D"G AE ~ CE
(The point
is
cl
INTRODUCTION TO APOLLONIUS.
Heuce
D"G
'
CE'J
GM" ~ GE -y
where
, is
from the
line
CG
measured
parallel to
BC.
locus
is
accordingly transformed
2/w,
is
ABCG.
The method here given contains nothing which would be beyond the means at the disposal of the Greek geometers except the mere notation and the single use of the negati\^e sign in (- 3iQ), which however is not an essential difference, but only means that, whereas by the use of the negative sign we can combine several cases into
one, the Greeks
Lastly,
it
should
with
- yy'
4-
dx +
e7j
0,
X (ax +
fiy
d)
-y {yy
e).
conic from the general equation of the second degree wanting only
CHAPTER
VI.
POINTS.
clear that he
five points.
is
and given a
() between
distances of any point on the locus from each pair of opposite sides
of the quadrilateral
measured in any fixed directions is also given. of the conic through the five points is reduced to the construction of the four-line locus where the constant
given.
is,
of the construction of a conic through five points however, not found in the work of Apollonius any more than the
The problem
The omission
of the
by the fact
own
necessary for the solution, no doubt regarding the actual construction as outside the scope of his treatise.
would have
been natural to give in a treatise on conies the construction of a The explanation of the omission may be conic through five points.
that
it
to
This
the
may
first
when
remembered
that,
in
a Greek
geometer
would regard the problem as being in reality three problems and involving a separate construction for each of the three He would conies, the parabola, the ellipse, and the liyperbola.
clii
INTRODUCTION TO APOLLONIUS.
discover
then
for
that
the
construction
possible
would be a completely
ellipse or a
problem not solved along with the Further, if the curve were an
find a
would be necessary to
must be satisfied by the particular If it points in order that the conic might be the one or the other. were an ellipse, it might have been considered necessary to provide
expressing the conditions
against
its
^
find
degeneration into a
it
circle.
Again, at
all
time of ApoUonius,
a
/xos
Thus it would follow that the complete treatment of the problem by the methods then in use must have involved a discussion of considerable length which Avould have been disproportionate in such a
work as that
It
is
of
ApoUonius.
note
interesting to
how
far
what we actually
in
of Book IV. on the number of points in vhich two conies may intersect are instructive in this connexion. These methods depend (1) on the harmonic polar property and (2) on the relation between the rectangles under the segments of intersecting chords drawn in fixed directions. The former property gives a method, vhen five points are given, of determining a sixth and by repeating the process over and over again we may obtain as many separate points on the curve as we please. The latter proposition has the additional advantage that it alloAvs us to choose more freely the particular points to be determined and by this method can find conjugate diameters and thence the axes. This is the method employed by Pappus in determining an ellipse passing through five points respecting vhich it is known beforehand that an ellipse can It is to be noted that Pappus' solution be drawn through them*
The methods
not given as an independent problem in conic sections, but it is an intermediate step in another problem, that of finding the dimensions of a cylinder of which only a broken fragment is given such that no portion of the circumference of either of its bases is left whole. Further, the solution is nmde to depend on what is to be
is
*
Pappus
(ed. Hultsch), p.
107G seqq.
I'lVK I'oINTS.
rliii
found in ApoUonius, and no claim is advanced that it contains anything more than any capable geometer could readily deduce for
himself from the materials available in the Conies.
Pappus' construction
is
substantially as follows.
points are A, B, C, D, E,
and are
sucli
connecting the different pairs are parallel, we can reduce the problem to the construction of a conic through A, B, />, E, F, where EF is
parallel to
AB.
For,
if
EF
and
be drawn through
parallel to
AB, and
if
CD
meet
AB
in
EF
in 0',
we
intersecting chords,
CO.OD AO. OB =
:
CO' O'D
.
EC
O'F,
whence O'F
where
We EF is
if
known, and therefore F is detoriiiined. have therefore to construct an ellipse tli rough J,
is
/>',
/),
E, F,
parallel to
AB.
AB,
And,
V,
EF
respectively, the
line joining
V and
is
a diameter.
Suppose BB to be the chord through JJ parallel to the diameter, and let it meet AB, EF in G, U respectively. Then R is determined by means of the relation
RG.CD
-.BG.GA -RlI.llD
FH .UK
(1).
cliv
INTRODUCTION TO APOLLOXIUS.
let I) J},
In order to detenuine R,
respectively.
RA
be joined meeting
EF in
A",
Then
RG GD BG GA = {RH
.
:
IIL) {DII
. :
UK),
l^y
similar triangles,
RH.IID
II. IIL.
we have
FU.HE^KII.HL,
whence IIL
is
is
determined.
And
the
intersection of
AL,
DH
determines R.
D,
RF meeting
/'//.
the diameter in
M,
respectively.
And, by the
same procedure
as before,
ve
obtain
HE
RII. IID =
FW WE
. .
P'W WP,
.
by the property
x\lso
of the ellipse.
. : :
FH HE RH .HIJ = F W WE
triangles.
iV
WM,
by similar
Hence
P'
W WP =
.
W WM
.
and similarly we can find the value of P'V. VP. Pappus' method of determining P, P' by means of the given WP amounts to an elimination of one A'alues of P' V VP and
.
FW
of the
unknown
Q, Q'
(),
It follows
W, Q, Q' are thus known, while P, P' remain to be found. from (a) that
FV VW=QV:
:
VP,
V.
whence
FW -.VW^PQ:
we
obtain,
From
this
by means
of
(/3),
so that
or
PQ.PQ'^QV.Q'W.
can be found, and siinihuly
/''.
Thus
THE CONSTUKCTION OF
It
is
I'OINTS.
civ
noteworthy that Pappus' method of determining the exPP' (which is the principal oVyect of his
by
five points
and there
is
But there
a simpler
expedient which
we know from
acquainted with, and Avhich can be employed for the same purpose
four-line locus
is
a conic.
8.
On
The problem
is
as follows
on a straight
line, to
det(irmine
AP.CP-.BP.
has a given value.
DP
straight line
The determination of the points of intersection of the given and a four-line locus can be immediately transformetl
has reference.
Hence
it
is
important to examine
all
we
This
is
contained
Book
of Pappus,
who
is
number
of
It
much
had the investigation not been intended as a means of solving other important problems. The conclusion is and therefore irresistible that, like the Books above mentioned, that On determinate section also was
would have been
likely
<;
meant to be used for solving problems in conic by means of the equation To determine
sections.
AP.CP^X.BP.DP,
where A, B, C, D, are given, is now an easy matter because the problem can at once be put into the form of a quadratic equation, and the Greeks also would have no difficulty in reducing it to the But, if it was intended for application usual application of areas.
644.
clvi
INTRODUCTION TO APOLLONIUS.
the
in
further investigations,
complete
discussion
of
it
would
naturally include, not only the finding of a solution, but also the
C and
the
B, D, for the cases where the points in either pair coincide, where
one
is
infinitely distant,
and
so forth
so that
we should expect
And
it
we
find in
Pappus,
vho
further makes
clear that,
what though we do
mined by the equation for different values of , yet the treatise contained what amounts to a complete theoi-y of Involution. Thus Pappus says that the separate cases were dealt with in which the given ratio was that of either (1) the square of one abscissa measured from the required point or (2) the rectangle contained by two such abscissae to any one of the following (1) the square of one abscissa, (2) the rectangle contained by one abscissa and another
:
separate line of given length independent of the position of the required point, (3) the rectangle contained by two abscissae.
also learn that
We
the
lemmas too
(1)
maxima and minima wei-e investigated. we may draw other conclusions, e.g.
in the case Avhere
From
that,
=1,
and therefore
has to be
AP.CF = BP.DP,
Apollonius used the relation*
BP :DP = AB.BG: AD DC
.
(2)
of the
C and B, D
:
by means of
the relation t
results of the
work On determinate
determined by
the
field
lemma
(p.
704)
and
is
(p. 732).
APPENDIX TO INTRODUCTION.
NOTES ON THE TERMINOLOGY OF C4REEK GEOMETRY.
The
propositions from the Conies of Apollonius which are given
at length in Chapter II. above will have served to convey some idea
of the phraseology of the
following notes
is
to supplement
Greek geometers and the object of the what may be learnt from those
;
propositions by setting out in detail the principal technical terms and expressions, with special reference to those which are found in
Apollonius.
headings.
It will be convenient to
fuller expression
point
point
cvTos
points
finite
*
, () , (, , ' , ( , , ?, , , 8, , ^1.
A point
the point
to
or
simply
a
to
'
earlier geometers
letter)
was
A, "the point
(is
put the
A*."
Any
tlie
is
yci'o'/xcvov
(so)
taken
a point not
tv
the surface
^5
79
in two
and
so on.
The following
centre
are
names
apex or vertex
of bisection
alone, a
point of division
])oint
extremity
iripas.
line is
a straight line
(
;
or
is
straight
line
eWeia
a curved line
{.) '
rj
the letters)
AB.
The same
same variation
of
ca.'ie
after txl,
are found frequently in Aristotle, particularly in the logical trefttises and the
Physics.
,
civiii
APPENDIX TO IXTRODrCTIOX.
but
thus
alone
, ,
;
conic
as a curve)
Of
.
77
<;
on
is
the curve.
?A
)
is
e.g.
a circle or a
is
that extremity of
a peiyendicular
to
line jyroduced is
eV
cv^cta?
^.
or
<
it
we
ctti
For a
line passing
likewise
Of a
line meeting
another line
are used
;
a;(pis
of concourse
joining H,
the points
],,
to
point of meeting
cVi
of contact,
is
, ^ '
Ave
r^^ii^
(with ace).
, ,
from
passes through
;
(with ace),
Iws ov
line
;
until
meets
;
or
tlie
the jjoint
ctti
to
The
line
bisected in
middle jioints
is
,
ivill
it bi/
\,
eis 8e
Straight lines cut off ov intercejyted are the part cut off ivithout (the curve)
cut off an eqrial length
point on a line
it
:
agreeing with
from
the line
/. '? (
is
,, . ;
;?,
evOeLa
its
the line
is
joining their
/'/
cktos
or
on
aV
by an adjective
thus
the extremity
of the
from axis, eV
extremity,
drawn
to the
^, //,^,
drawn so
as to meet at the
section.
2.
Angles.
angle
;
An
right
is
,
;
an acute angle
o^tia
,
a'ul
obtuse
at right angles to
(with dative) or
at right angles to
A (drawn) from
to cut at right
r;
angles
' ...
.
Vei'ticalliJ
opposite (angles)
to the
cone,
.
;
angle
is
,
of the
;
of
/cci/xcrut
f/tf
),
also used
triangles (e.g. in
of a double
and
two halves
The For
we
more usually
?(3 (together)
or
equal
to
a right angle
ai viro
'.., ,
<;
, ,"
.
but
The angles
The adjacent angle, or the sjipplement of an angle, is <; To subtend (an angle) is either Avith a simple accusative, or with and ace. (extend under) as in at
-.
', '
the angles
stibtend.
which
the
homologous sides
3.
A phne is
of a
a figure
or
diagram
circle)
,,,, , ,, , \
(A
which
is
not in
.
tlie
^.
same
The
line
is
their
/;.
A rectilineal fig^ire
circle
is
(Euclid),
quadrilateral
a five-sidedfigure
etc.,
its
circximference
a segment of a circle
/7/
or
less,
than a semicircle
circle
of a
', '^.
Of
to tlie
quadrilaterals, a parallelogram
is
a rectangle
Diagonal
figure
is
/.
<:
or frequently
,
being a
side.
;
a semicircle a segment
a segment greater, or
angle
is
a square
with ur without
upon a given
is
<
radius
is
. ZW
the square
clx
is
APPENDIX TO INTRODUCTION.
^,
TO
cffi?/.
(),
circle,
the figures
is
on
,
is
is
But
or a segment of a
on a given straight
inl
.7/
to
.
is
^75
on
,' .
;
rectangle applied
its
breadth
ZE
is
TO
, ZE
?.
.( on
is
The
to
rectangle contained by
or
{)
the
another straight
line)
a rectangle equal
taov
the use of the word SiW/Ats and the various parts of the verb
expresses a square (literally a potver)
;
/
sum
In
of
With
is
StVa/iiai.
thus in Diophantus
it
is
unknown
language
an algebraical equation,
i.e.
for
af.
In geometrical
in
" the ratio which " (as one might say) " the inner segment has to the remaining segment j)ote7itially," meaning the ratio
8,
areas,
oV
/7/
of the square of the inner segment to that of tlie other. Archimedes speaks of the radius of a circle as being
(Similarly
to the
is eq^ial etc.)
two
like
manner, when
equal to another,
lines) the
/
is
used of a straight
line, it
means
producing an area
contained by
contained by
rectangle Zs.
;
;
MN
on
it
the square
^is
in Apollonius (straight
;
to the rectangle
the square
on
it
is
equal
to the rectangle
is
the
square on
MN
equal
7;
on.
,
is
to the
To
5^
of its being possible
^,
kci/xcvos
T17V
and so
is
in Euclid L
construct a triangle
from
{equal
to)
EA.
/xcvov
Equiangular
sitiiated
triangles
/ ,
/?
is
. ',
<;
and
lines
to
similar
;
,?,
similar
KEO
and
similarly
clxi
?, 6<; ?,
A
a circle
is
a right cone
(,
the apex
the straiyht
^,
is
by
its
motion about the circumference of the fixed jyoint through which the
/, .
double cone
is
6 ^,
is
8vo
<;,
ellipse,
aans
A circular
expression
branches of a hyperbola
<;
is
branches are
In the middle of a proposition, where we should generally use the word curve to denote the conic, Apollonius generally uses
sectimi,
^^ ^.
or
;
is
, .
8vo
the
also at
evavTiov
<;
in the
(<;
conjugate opposites.
Of
(.<;
sometimes
.
is
5.
Diameter
{secondary).
/?,
and
,
is
conjttgate diameters
the other
{erect)
,
or
^
of
or Sorrcpa
(i.e.
that
8,
(in
the plural)
bisecting diameter is
conic
is
simply
^
it
) 8.
first
<;
radi us of a central
article).
}.
ordinate-wise,
Ordinates.
is is
the adverb
<;
u.scd
and the
that
can be
Cbdi
APPENDIX TO INTRODUCTION.
This verb
is
signifying to draw.
either
is
or
drawn doion to the diameter from a point on the curve, and the latter when it is drawn uptvards from a point on a diameter. Thus inl means suppose an ordinate drawn to the diameter, which
,
or
the
<;
//
diameter
is
/'?
ordinate
is
alone or
similarly
and
and sometimes alone, the other word being understood; are used alone for is an ordinate or
is
> -,
or
.
it
An
/?
7.
Parallel to
an ordinate
is
in one word.
Abscissa.
abscissa of
The
following).
8.
Parameter.
full
The
phrase
are applied the rectangles which in each conic are equal to the
,8 -,
an ordinate
is
rfj
by
from
Similarly
or
we
the
the
which
is
is
'
which
to
The "figure"
figiire
of a central conic.
is
The
(to
cTSos)
the technical
is
transverse side
() ( )
of the figure
tyj
is
called the
ei-ect
is
the
side
(8).
We
expressions, to
the figure
to (the
on
(the
diameter)
AB
applied
diameter)
AB
and
(the
parameter)
and
TO
rrj
diameter which
? )
?
on
the the
fillEEK
tt7S
OEOMETUY.
ayo/ACVTj
clxiU
;
is
the figure
on
(of
[the
is)
parallel tangents).
onefourth of the figure is, with reference to TO a diameter PP', one-fourth of the square of the conjugate diameter
two
DD\
i.e.
CD-.
10.
Tangents
touch
is
etc.
To
lines
most conniiouly
being of course
the tangent at
is
Pappus,
tvill lie
p.
664, 28,
on a straight
The word
/^' ^
iwnj/aveiv is also
is
/
let
is
used of points
, . -^/
whether used
of straight
Ii/ing
on a
lav
locus.
Cf.
p.
664,
2,
if
it lies
position).
eV
commonly used
of touching, e.g.
'
.
avTov
11.
totes is
innj/avovaa toucliing any one of the sections at random. Point of contact is chord of contact The point of intersection of two tangents is
-^. :
The following
eav
it
the curve)
aV
d/rawKi) from
,
He
End.
i.
found in Apollonius
"
from
and
(outside
if {there he
the other
Though the
wider meaning and was used of any lines which do not meet, in
,
in
(i.e.
Asjnuptotes.
technical term used by Apollonius for the asympit is
whatever direction they are produced. Geminus, distinguishes between (a) plane and (b) those which are not.
one
which are
distance,
one plane " some are always at the same distance from
parallel),
one another
as a
line
and the
Comment,
in
p. 177.
Clxiv
APPENDIX TO INTRODUCTION.
The same use
of
(ii.
in its
.^ lyyiov
( ,,
ii.
? ,
The
original enunciation of
/;
the
el<;
^ ^/?
and
the section,
ctti
^ tJie
lines
14)
(the
the lines
which do not
14 [Prop.
"36] is interesting:
eyyioV
fts
asymptotes
if produced to infinity,
distance less than
similarly
we
/^
the expression
/aiS?
'/
/;,
For
is
on one of
the
asymptotes containing
the
section.
the asymptotes
n7S
.
find
and
the curve is
12.
Given
hoOiU or
in magnitude
be,
(^ /';
alone, as in
/^
AB
given in positio7i
(of straight lines).
given
given in position
/', or
even
we frequently
^'
^'
and
or will
without
AE.
A more remarkable
ellipse is that
,
of
commonly found
to
2)Ci')'cdlel
'
-
in such expressions as
(given) in 2)Ositio7i,
^'
of the
^'
ttj
AB, used
(given) in positio7i.
all the
circle,
is
by hypothesis equal,
they meet
by hypothesis.
means which
13.
is
contrary
to the hypothesis.
In a theorem
by TO Thus
is
required
to be
proved
is
sometimes denoted
is
^ ',
in a
problem
to
'^'.
then
;
would have been done.
ij
an
axis, that
To draiv
NOTES ON
171
TERMINOLOGY OF ORKEK
is
the
manner required
ayayitv
In the
is generally introduced by Apollonius by the words on and in one case ApoUonius abbreviates the re-statement by saying simply otl
what
:
be
is
said
. (. .
When
when a
;
clxv
the solution
it is
required to prove
as well as Euclid
',
I say
true
;
it is to
proved
SeiKTov, let it be
The
synthesis of a
hrj
^7;
li.
).
required
to
renuiins to he proved
dra^v hiov
dyayav.
Constructions.
These are nearly always expressed by the use of the perfect imperative passive (with which may be classified such perfect
imperatives as
from
'^,
from
.'
,
we
/?
;
^
its
:
from
in
The instances
if tve
ApoUonius
,
,
tic
it,
for in
the
same manner as
yap
/ say
that...,
having joined
AB
nJvai, to
, ^, ' ^, , ,
having drawn
the taiigent
; ,
shall prove
we have a somewhat
this touches.
common
to
dratv
produce
to cut off
/,', ',
and
to
,
the angle
vhile in
^
to take
to bisect
rrj
make (one
'
'
avrrj
line
before, after
epou/xev
most
to
join
iirtCevy
or supply
to construct to
to describe
compounds,
to erect
divide Staipetv,
^'.
;
angle
^
let
AB
erected
(?
out,
'
;
7;/
and
to
it,
as centre
at
a.
distance
071
' (
let
ypatvo<;
a7igle) be
made equal
let
clxvi
APPENDIX TO INTRODUCTION.
detailed enumeration of the various perfect imperatives
;
No
is
necessary
elegance.
but
it
Let
it
he conceived is
thus
let
meaning literally to break off and word is generally used of two straight lines meeting and forming an angle, e.g. of two straight lines drawn from the foci of a central conic to one and the same point on the curve, ,
curious
its
is the jyoiiit
- ,,
(,.
'
Similarly, in a propo-
sition of
/icvos ApoUonius quoted by Eutocius from the drawn from the given jwints to meet on the circum-
ZoBkvTiMv
ctti
'
in
to it the
let
15.
etc.).
:
thus
8
for
or
is bisected
Of a magnitude having another added is used in the same way as participle of rj KP having something subtracted. Thus means KP minus or phis BO is equal to
added.
BO
is
(with gen.)
is
AEB
equivalent to
AE EB
.
+ ZE^
is
A curious expi'ession
meaning
thus Kotvov
t/ie
sum of ^,
or
or ofTZ,
,, .
or
Kotvo's is
used
common [magnitude)
be
the dative,
on Pi as base similar
;
, 8 ^, ,
case.
To
exceed
is
or
to
exceeds
by
,
;
is
expressed by
from
by the triangle
'
' 2.
differs
clxvii
the
half of AB,
of the figure, TO
<; AB
;
AE
proportional between
, . ?
IG.
AEA.
Proportions.
is
Ratio
?,
tvill be
The
of
;
ratio of
? ?
TO
I'he ratio
?
to
;
,, 7
A
to
is 6
. ,
,
A
ov
is
;
'
or
the
<</
a mean
dvaXoyov,
oV
to
, or
made
than
has
to
rot
HB,
has
to
7?
(or
,
oV
to the
TO
?
is
)
i.e.
same as
the
7^7?
greater (or
less) ratio
the ratio
irpo%
8.
? ?, ? /7/
to
, to , or
The following
on
A2
is to the rectangle
cv
' 2, ?
;
under B2,
dents are
^,,
?.
so is
,?
:
as the square
aVo
A2
In a proportion the
antece-
? ? . ? ^, ?
yovevv
v
all the antecedents (taken together) to all the consequents (taken together)
,?
is to
is
cj?
enunciation of Euclid
19
eav y
? , ?? .
5?
If as away,
whole.
whole
the
is to
a whole
remainder as
e.g. eVei
,?
To
be
Similarly in Apollonius
,?
AE
?,?
is to the
we have
? ?,
?
^
is
away
away
the (parallelogram)
to whole.
compounded of
(or
...),
the ratio
) , ^, ?
remainder
is also to
is
the rectangle
under
to (the
part) taken
remainder as whole
the ratio
compounded of
from
((,
sides
tlie
vy(yo<;
clxviii
a compounded
of
.^
ratio,
to
APPENDIX TO INTRODUCTION.
is
a ratio compounded
latter ratios
two
ratio.
Another way
is
to use
compounded
of
two others
addition.
the ratio
Similarly
ratio of
KOLvov
Taking
AH
)
:
the rectangle
So that
sponding
, .
the
:
)
d.
?2
ratio of k'% to
/ 2
<;
is
contained by
taking
, EZ
AH
ttj
common
ratio.
is
as a middle term
as a
common
altitude
/,/'/,.
eivat
so that
teryns
-)
=c
v. Thus alternately (usually called permutando or alternando) means transforming the proportion into a c-b d. a==d c. reversely (usually invertendo), b b is composition of a ratio, by which the ratio a becomes a + b b. The corresponding Greek term to componendo is Avhich means no doubt, literally, " to one who has compounded," or " if we compound," the ratios. Thus is used of the inference that a + b :b = c + d: d.
: :
:
^
:
means
Similarly
divisio7i
or subtraction in the
^
d.
(sc.
:
dirimsndo
of subtracting
thus a - b
=cd
conver-
la-b and
to the
inference that a
St
,
tlie tlie
ab=c
d.
distantia), is applied
to
a
that All
d= A
D.
, ,
etc.
(',
in Euclid.
,
17.
it
,
Inferences.
on
ovv
line
;
clxix
In one place we
Are in
recijrrocal pi-ojiortion is
..
is
'.
for therefore
e.g. iv rrj
Vi0avcta
the surface,
is
a straight
and
in conjunction with
starting point of
inference, so
since, then,
what
oTt,
that is
or by
.
on
/cpo'v,
some other word, in order to mark the an ai'gument rather than to express a formal that can usually translate it by then, e.g. Vei oty
\ ...(
A
corollary
it is
it is,
is
some-
up an argument.
is
that is
often introduced
the
by
, ,
/
proved at
It is at once clear
, , . ^. - ,
wherefore
for
or
TOts
, (..,
before
rijs
'^, from
same
the
time.
this reason
,
way
this it is clear
for
same reason
in the same
as above or before
will be
similarly
it
shown
;!(^77,
the
same
the
same proofs
apply
Conversely
converse
elp
By
w/tat
was
cttI
8(.8(.
CTTi
By
the similarity
by parallels
parabola, hyjyerbola
The properties
. , ,., ,
;
proved in the case of the hyperbola 8ia. for the same (facts) have been
t/ieir
doubles
{])ropei-ty
proved true of
when
the
8.
Much more much sooner does
18.
it
Cf.
of
t/ie
the) section,
sections
Conclusions.
it
Which
which
is
to
prove
this
is
18
absurd
and
impossible,
,^
so
that
in
the
H. C.
clxx
.
19.
APPENDIX TO INTRODUCTION.
similarly inferred
Distinctions of cases.
\
-^
,
*
or
(^^^
20.
.,
for
...,
section
)' ,^ ,
In
towards the direction of the point E, inl the same side of the centre as AB, hrl
same parts as
the
iv
c<mv
2)arts
of tJie section, and used and passing inside. The concave parts
) ,
its
the convexities
not having
its
curve
7ri
side iav
.
Infinite, Infinity.
-, ,
to
of the
if
it
touc/oes
with
its
its
concave
will touch on
concave side
Having
^.
(
is
21.
Unlimited or infinite
nitely
^^.
infinite
we
shall
(.
find an
number of diameters
'
way
'
i
TFTHF
THE CONE.
If a straight line indefinite in length, and passing always
to
the moving straight line will trace out the surface of a double
cone, or two similar cones lying in opposite directions and meeting in the fixed point, which is the apex of each cone.
The
circle
is
called
the base of the cone lying between the said circle and the
fixed point,
The cone
base.
so described
In this latter
is a scalene or oblique cone except where the axis is perpendicular to the ca,se the cone is a right cone.
If a cone be cut
resulting section
is
lying on
base.
is
is
the
is
BC, and
let
that
.4
plane parallel to
H. c.
Suppose now that the cone is cut by any the plane of the base BC, as DE, and let
I
DE in DE and
o.
Let
be any point on
circle
Join
Ap
BC
in P.
lines
0,
lines
DE
in the straight
OP, op
respectively,
..
OP, op are
parallel.
op:
OP = Ao:AO.
remains constant
for all positions
And,
BPG being
circle,
OP
of ^j on the curve
DpE, and
is
the ratio
all
Ao: A
is
also constant.
Therefore op
surface
constant for
circle.
Hence
all
are parallel
to
the
4.]
Next,
axis
let
and perpendicular
BG, and
triangle
let
the
ABG.
drawn at light angles to the plane and cutting off from it the triangle
similar to the triangle
i.e.
the
ABG
is
(
ABG
AHK
such that
AHK
but
is
lies
ABG.
).
HK
is
called a
The
references in this form, here and throughout the book, arc to the
THE CONE.
Let
with the
circle
surfiice,
be any point on the intersection of the plane II and F any point on the circumference of the
BG.
the triangle
tively in
Draw PM, FL each perpendicular to the plane of ABC, meeting the straight lines HK, BG respecM, L. Then PM, FL are parallel.
the straight line
Draw through
it
BE
parallel to
BG, and
DME,
PM
and
is
BG
of the cone.
Thus
circle,
the
section
DPE
is
DM.
ME= PM\
is is
But, since
the angle
DE AD
parallel to
BG,
the
angle
ABG which
is
.
:
equal to
by hypothesis
are
the
vertical
angles at
M.
Hence
..
And
is
any point on the intersection of the plane Therefore the section made by the plane
HK
HK
is
circle.
they-e
Thus
of the
and
Suppose a cone to be cut by any plane through the axis section ABG, so that BG is a diameter
Let
be any point on the circumference
a parallel to
Further, let
on the surface
of the cone but not lying in the plane of the axial triangle.
AQha
if
necessary, to
meet
12
FLF' be
FLF'
;
the chord
follows
Join
AL, AF'.
Then the
straight line
it
through Q parallel to
HK
is
also parallel to
therefore that the parallel through Q will meet both is in the plane of the axial triangle AF'. And
AL
and
AL
ABC.
plane Therefore the parallel through Q will meet both the and the other side of the surface of the of the axial triangle
cone, since
AF'
lies
on the cone.
Then
and
FL = LF'.
[I.
QV=
VQ',
or QQ'
is
bisected
by the plane
C]
passing Again, let the cone be cut by another plane not base in through the apex but intersecting the plane of the perpendicular to BC, the base of any axial line a straight
triangle,
DME
and
let
be
DPE,
the point
the resulting section of the surface of the cone o( lying on either of the sides AB, The plane of the section will then cut the
AG
PiU joining
to
DE.
the curve of section, and through
Now
Q
Then
of the
let
Q be any point on
DE.
surface
THE CONE.
triangle.
But
it lies
DPE\
it
will therefore
PM.
Therefore
parallel
PM
bisects
is
to
DE.
is
Now
diameter.
Hence, if a cone he cut by a plane which intersects the circuku' base in a straight line })erpendicular to the base of any
axial triangle, the intersection of the cutting plane
and
the
plane
of the
found
cone.
[I. 7.] it is
which
it
ABC is
at right
Again,
dicular to
if
PJ/ be
made by a plane
DME
perpen-
BC, and if PJ/ be in such a direction that it does not be either meet AC though produced to infinity, i.e. if parallel to AC, or makes with PB an angle less than the angle AC and therefore meets CA produced beyond the apex of the cone, the section made by the said plane extends to infinity
6
For, if
it
HK
all
to
is
through
base.
HK, QQ'
produced and draw through V on BC. and QQ' parallel to DE, the plane parallel to that through DE, BC, i.e. to the
PM
are
HQKQ' is a circle. And D,E,Q,Q' on the surface of the cone and are also on the cutting
Therefore the section
plane.
circle
HQK,
[I. 8.]
and
on
in like
manner
any point
PM produced, and
VK
and
HV VK
.
from P.
diameter
both
V
;
is
is parallel
AC,
VK
PM
HV,
and in the case where the meets CA produced beyond the apex of the cone, VK increase together as V moves aAvay from P.
indefinitely
HV increases
Thus
QV
increases
as
the
section
extends
to
infinity.]
If on the other
hand
extend to
if its
infinity.
is
be a
But,
circle
if
it
plane
is
the
section
not be a
circle.
9.]
For
in
let
DME,
THE CONE.
circular base.
Take the
BC meeting
the
are
Then P, P\
line.
of section.
Therefore
PP'
is
a straight
PP'
be a
VQ'.
circle.
Q on
it
parallel to
DME.
the axial
V,
QV=
Therefore
PP'
is
the
circle.
Let
HQKQ'
to
the base.
circles,
QV = HV. VK,
QV' = PV.VP'. ..HV.VK = PV.VP', so that HV: VP = P'V: VK. the triangles VPH, VKP' are similar, and /.PHV = ^KP'V; .. ZKP'V = ZABC, and the section PP'
..
is
subcontrary
PQP'
is
not a
circle.
remains to investigate the character of the sections mentioned on the preceding page, viz. (a) those which extend
It
to infinity, (b) those Avhich are finite but are not circles.
D.VE and
that
ABC is
8
Avhose base
bisects
BG is
DME
all
Then,
if
the
plane of the section and the plane of the axial triangle intersect
in
PM,
PM
is
bisecting
parallel to
BE.
is
If
so bisected in
V,QV is
said to be an ordinate, or
a straight line
drawn
PAi
;
ordinate-'wise
and the length cut olf from the diameter by any ordinate Q V will be called the abscissa of Q V.
to the diameter
(' '),
PV
to
let
Proposition 1.
[I.
11.]
First
let
the diameter
one of
to be
AC, and
QV be any ordinate
PL
(supposed
diameter
PM.
draiun p)erpendicidar
PM in the plane
: :
PL PA = BC'^
QV' = PL.PV.
Let
HK be
is
draAvn through
parallel to
BC.
Then, since
QF
H,
is
also parallel to
DE,
it
Q,
parallel
to the base
of the cone
and therefore
THE CONE.
produces a circular section whose diameter
at right angles to
is
UK.
Also
QV
is
HK.
..
HV.VK = QV\
by
parallels,
Now, by
and
..
HV:PV=BC:AC
VK:PA=BC:BA.
HV. VK.PV.PA=BG':BA.AG. QV .PV .PA = PL PA = PL.PV:PV.PA.
:
Hence
..
QV''
= PL.PV.
It follows
diameter
PM
PL
drawn
at right angles to
()
('
PM,
to the fixed
PM with
rectum
PV.
Hence the
called a
Parabola.
()
or
The
PL
is
or the
Karar^opLevaL
'^<;).
will for
Thus
12.]
to
Next
angles
that
to
:
let
not
the
be parallel
AC
but
let
it
meet
CA
produced beyond
in
apex of tJie cone in P'. Draw PL at Hght the plane of the section and of such a length
.
PL
PF = BF
in R,
FG AF\
:
where
AF
is
a straight
line
through
be
parallel to
drawn
parallel to
it is to
PL
be
meet
VR
proved that
QV' = PV.VR.
As
that
before, let
HK
be drawn through
parallel to
BG, so
QV' = HV.VK.
10
HV:PV=BF:AF, VK :P'V=FC:AF.
..
HV.VK :PV.P'V=
BF.F('.AF\
.PP'
Hence
QV :PV .P'V=PL
..
PL
but overlaps
by a length equal
is
to the difference
between
VR
()
and PL*.
equal to a
Hence the
*
Hyperbola.
PR
exceeding by a figure similar and similarly situated to that contained by I'l^ and PL, i.e. exceeding the rectangle VL by the rectangle LR. Thus, if QV=y,
PP':^d,
y-=px + ^.x-,
which
is
extremity.
THE CONE.
11
PL is called the latus rectum or the parameter of the ordinates as before, and PP' is oallcil the transverse -rrXayia TrXayia). The fuller expression transverse diameter is also used; and, even more commonly, Apullunius speaks of the diameter and the corresponding parameter together,
(
/;
/;
/69)
ifkevpa),
rectum
(i.e.
and the former the transverse side { irXayia TrXeupa), rrj on, or applied to, the diameter of the figure i.e. of the rectangle contained by PL, PP' as drawn.
),
[Coil.
(')
The parameter
PL
will in future
be denoted by
jj.
It follows
QV':PV.P'V=PL:PP'
that, for
QV iPV.P'Visa constant
or QF^ varies as
ratio,
PF.P'F.]
Proposition 3.
[I.
13.]
If
meets
AC in
P' and
BG
in
M, draw
A F parallel
to
PM nieetiiuj BG produced in F, and draw PL at right angles to PM in the plane of the section and of such a length that
PL PP' = BF.FC AF\ to PL meeting P'L in R.
:
:
Join
It
VR
parallel
luill
he proved that
QV"'
= PV.VE.
12
Draw
Then, as before,
Now, by
..
Hence
..
Thus the aquarc on the ordinate is equal to a rectangle whose height is equal to the abscissa and Avhose base lies along the fixed straight line PL but falls short of it (iWeiTrei) by a length equal to the difference between VR and PL*. The
section
is
As meter
called
before,
PL
is
the
transverse
or
(with
or
without
of
last
diameter
the
proposition).
PL
will henceforth
be denoted by p.
[Cor.
It follows
QV':PV.PV' = PL:PP'
that, for
any
fixed diameter
PP',
a constant
ratio,
QV^:PV.P'V is
or
QV varies SisPV.PV.]
* Apollonius describes the rectangle
PR
VL
contained by PP"
and PL,
If
i.e.
Thus ApoUouius' enunciation simply expresses the Cartesian equation to a diameter and the tangent at its extremity as (oblique) axes.
referred
THE CONE.
18
Proposition 4.
[I.
U.]
If a plane
through the apex, the sections of the two parts of the cone will both be hyperbolas which will have the same diameter and equal
later-a recta
coiTesponding thereto.
And
OPPOSITE BRANCHES.
Let
BChe
B'C
BC
B'C
in
D'E\ Then
'
DE
must be
parallel to
DE.
Let
BC
DE
B'C
at
BC
A
is
B'C
in
B'C, which
to
will therefore
be a diameter
will cut
DE'
DE.
^*
Let
^.4i?^'
iiA
be drawn through A parallel to MM', the straight Hr^ join.ng the n.ddle points of DE, D'E' and meeting C^
respectively in P, P'.
'
Draw
perpendiculars PL,
P'L
to
MM'
PZ
Since
.PP' = BF.FG:AF\
P'L':P'P=B'F'.F'C':AF'\
now ifP,
DPE
Sl^aT/plh'ofa/^'"
Also since i)'^'
is
^^^-^
^^^
^^-' ^^^
-'^"
when
he axial triangle
triangle meets
DPE' i, also a
And
It
C'A produced beyond the anex A fh! ''^'^"" ^ ^ ^' '^" hyperbola.
of
he
lia
same diameter
MPP'M.
PL = P'L'.
We
Hence
AF'\
-.PL^P^L'.
ITS CONJUGATE.
Proposition 5.
[I.
15.]
C, the middle point of the diameter PP' of (oi a double ordinate BCD' he draiun to PP', BCD' will bisect all chords parallel to PP', and will tJierefore he a diameter the ordinates to which are parallel to PP'.
If through
ellipse,
second diameter,
parallel
to
In other words, if the diameter bisect all chords parallel to a the second diameter will bisect all chords
the first.
to
BOB'
will he a third
Let
QF
Q draw
\
QQ
let
parallel to
in
and the
ellipse in Q'
and
QV
Q to
PP'.
16
PL
3]
is
if
is
joined and
PL
to
meet P'L, we
have [Prop.
QV is parallel to Q'V and QQ' ..PV.VR = PV'.V'R. Hence PV PV'=V'R': VR = P'V P'V. PV: PV''-PV=P'V' P'V- P'V, PV:VV' = FV':VV'. or ..PV=P'V. GP=CP'. Also CV = CV\ By subtraction, and Qv = vQ', o that QQ' is bisected by BD'.
and
QV = QV,
.'.
..
(2)
that
right angles to
:
DD' and
of such a length
vi'
DK.
Join
parallel to
DK to meet D'A^ in
Then, since
..
r.
ES
parallel to PP'.
PC =CP', PS = SL
(P^) =
and
CE=EH;
.
(>Sri/).
Now
and
..
But
..GO'-QV' = RT.TE. = CP' - Ov' = P'y vD. (A). ..D'v.vD = RT.TE PP' PP' = PP' PA, by hypothesis. ..DD' :DK = DD":PP"' = CD' GP' = PG.GE:GP' = RT. TE RT\ DD' P7i^ = D'v vr = D'v .vD vD. vr D'v .vD:Dv.vr = RT.TE: RT\ D'v .vD = RT. TE, from (A) above Dv.vr = RT = CV'=Qv\
CP- - Q F'
. : : : :
,'
^^
..
ITS
CONJUGATE.
17
DK
is
DD'
is
a third
PF.
Cor.
We
have
or
PP'
and
PL
is
Thus the
of PP',
DD' and
the
corresponding
Def.
all
Diameters such as
PF,
Proposition 6.
[I.
16.]
of the diameter of a hyperbola with two branches a line be drawn parallel to the ordinates to that
the
jjoiiit
If from
middle
a diameter conjugate
to
the
former
If
one.
any straight
line
diameter, and meeting the two branches of the hyperbola in Q, Q' respectively, and if from C, the middle point of PP', a straight
line
V,
be drawn parallel to the ordinates to PF meeting QQ' we have to prove that QQ' is bisected in v.
in
be the Let QV, Q'V be ordinates to PF, and let PL, parameters of the ordinates in each bmnch so that [Prop. 4] 2 H. c.
FL
18
Draw VR, V'R parallel to PL, P'L', and let PL, P'L be joined and produced to meet V'R, VR respectively in
PL = FL'.
R',R.
Then we have
.'.
Also
PV. VR = P'V V'R, and V'R :VR = PV:P'V'. PV V'R = PR RL' = RP PL = P'V VR. PV .P'V=V'R' .VR = PV. RV, from above
: :
:
..
and
or
PV '.PV=P'V:P'V', PV + PV PV = RV + RV RV,
...
: :
VV .PV=VV':RV';
But
..
by addition,
or
Hence Gv
is
a diameter conjugate to
PR.
2,
QV:PV.P'V=PL:PP',
Q'V"
and
..
:RV.PV = P'L'
or PF PV= P'V'.]
:
PR,
:
QV=Q'V, PL = P'L':
PV.RV=PV.RV',
The middle point
is
whence, as above,
Def.
hyperbola
the
line
dmwn
mean
).
secondary diameter
Proposition 7.
[I.
20.]
In a parabola
vanes as
This
is
the square on
an ordinate
to
the diamete?'
the abscissa.
1.
ITS CONMIYIATE.
19
Proposition 8.
[I.
-21.]
In a hi/perhohi, an
to the
ellipse,
or
<i
circle,
if
QV be
ani/ nrdindte
diameter PP',
QV'xPV.P'V.
[This property
is
QV':PV.P'V=PL:PP'
obtained in the course of Props. 2 and 3
;
QV^ =
PV .VR
follows.]
Q'V
Then
QV' = PV.VR,
qV^PV. V'R';
..
QV .PV.PV= PV.VR
PV.P'V
= VR :P'V=PL:PP'.
22
20
Similarly
..
QT* PV'.FV = PL
:
PP'.
QV':Q'V"' = PV.P'V:PV'.P'V';
constant ratio,
and QV^
or
PV .P'V is
QV'ocPV.P'V.
Proposition 9.
[I.
29.]
If a
will, if
Let PP' be the given diameter and C the centre. Let meet one branch in Q. Draw the ordinate QV to PP', and
off
CQ
set
GV
CV.
along
PP' on
the
centre
equal
to
Let
V'K
be the ordinate to
is
PP' through V.
We
QGK
a straight
line.
Since
it
follows that
PV= P'V
But
..
[Prop. 8]
QV=KV';
KV are
line.
parallel, while
GV = GV.
Therefore
QGK is
if
a straight
Hence QG,
ITS
CONJUGATE.
21
Proposition lO.
[I.
30.]
In a hyperbola or an
is bisected at the centre.
the centre
Let PP' be the diameter and G the centre and let QQ' be any chord through the centre. Draw the ordinates QV, Q'V
;
Then
QV
Q'V'
= (77^
..
GV'\ by
similar triangles.
:
GV
GP'
GV = GP"
and
GV'\
But
..
GP' = GP";
CV'=GV'',
are parallel
..
GV =
GV'.
GQ=CQ:.
TANGENTS.
Proposition 11.
[I.
17, 32.]
If a straight line he draxmi through the extremity of the diameter of any conic parallel to the ordinates to that diameter,
the
line
straight
line
will
it
touch
the
conic,
and no
othei' straight
and
the conic.
It
first
the
manner described
For, if not, let
within
as
PK, where
PM
is
PM,
But
meet
PM and will
falls
be
KP
produced
without
the conic
therefore
it Avill
not be bisected at P.
PK
must
fall
and
will therefore
touch
it.
It remains to be proved that no straight line can between the straight line drawn as described and the conic.
(1)
fall
let
PF
be parallel
PV.
If possible, let
PK
fall
PF
KV
parallel to the
Then
KV':PV''>QV'
PV
>PL.PV:PV' >PL:PV.
Let
V be taken on
PK
such that
KV:PV' = PL.PV',
and let V'Q'M be drawn Q' and in .1/.
parallel to
in
TANfiKNlS.
Then
KV'.PV'^FL-.PV
= PL.rV': PV" = q'V"\PV'\
and
KV PV = MV"
PV'\ by
parallels.
Therefore
MV" = Q'V'\
is
and
Q',
MV = Q'V.
and therefore does not between
fall fall
Thus
outside
it
PK
:
which
PF
and the
ellipse
or
Let
let
PF be
fall
between PF And the curve. Draw the ordinates, meeting the curve in Q, and draw
PK
KV parallel
VR
to
per-
24
pendicular to
PV.
let it
(produced
if
necessary)
meet
VR
in R.
Then
QV = PV. VR,
PF
so that
Take a point S on VR produced such that KV' = PV.VS. Join PS and let it meet P'R in R'. Draw R'V parallel to PZ
in V, and through meeting QV, meeting the curve in Q' and
PK in i.
Now
..
KV' = PV.VS,
so that
VS:KV=KV:PV, VS:PV=KV':PV\
Hence, by
parallels,
is
mean
proportional between
V, VR',
MV' =
is
Q'V'.
Q',
Thus
outside
it
PK
:
fall
which
Hence no
between
PF and
the curve.
TANGENTS.
Proposition 12.
[. 33, 35.]
be taken on the diameter of a parabola outside If a point and such that TF = PV, where V is the foot of the
the curve
ordinate
from Q
to the
diameter
FV,
the line
TQ
or
will touch
the parabola.
We
line
TQ
TQ
produced
does not
K, a point on
TQ
or
TQ
produced,
parallel
draw
Q'KV
to
in
q.
Then
Q'F'^QF^ >KV'^: QV\ by hypothesis. > TV'"- TV\ .-.PV .PV>TV'"- TV\
: : :
. :
Hence
4>TP
and, since
^TP.PV=TV\
.'.^TP.PV'>TV'\
But, since by hypothesis
TF'
is
not bisected in P,
^TP.PV <TV'\
(which
is
absurd.
Therefore
TQ
fall
and
is
therefore a tangent.
* Though the proofs of this pioposition and tlie uext follow //; form the method of reductio ad absurdtim, it is easily seen that they give in fact the direct demonstration that, if A' is any point on the tangent other than Q, the point of contact, A' lies outside the curve hecause, if KQ'V' be parallel to QV, it is proved that KV" >Q'V'. The figures in both propositions have accordingly been drawn in accordance with the facts instead of representing the incorrect assumption which leads to the iibsurdity in each liise.
2
can fall bettveen
= PV.
Also no straight
line
TQ and
the curve.
this,
of reductio
Proposition 13.
[I.
34, 36.]
In
point
diameter and
TP
TP' =
PV
TQ
cm^e.
We
have
to prove that
no point on
TQ
or
TQ
produced
falls
TANGENTS.
If possible, let a point
27
fall
on
TQ
or T(^ produced
within
an ordinate meeting the curve in Q'. Join P'Q, V'Q, producing them if necessary, parallels to TQ meeting V'Q, VQ in /, and draw through P'
parallel to
,
Q'KV
respectively.
in
Also
let
M.
:
Now, by
..
hypothesis,
parallels,
by
P'H
= P'H:NM.
Therefore
PN = NM.
Hence
or
.:
or
It follows that
..
.>.. '.
:
:>:; ,
.
>
0.1/,
> OP
'PQ'
> >
.OP
'fQ'\
by similar triangles
P'V
or
PV
'
P'V .PV
",
..
P'V.PV:P'V'.PV'>TV':TV"; .'.QV':Q'V">TV':TV" >QV':KV'\ Q'V < KV, which is contrary to the hypothesis.
does not cut the curve, and therefore
it
Thus
TQ
touches
it.
PP'
from
and
QV is
the ordinate
'TP:'TP'
Also no other straight
[This again
reductio
is
= PV:
VP'.
line
TQ and
the curve.
ad absurdum.]
* See the note
on
tlie
previous propo^iition.
28
Proposition 14.
[I.
37, 39.]
to
In a hyperbola, an ellipse, or a circle, if QV be an ordinate the diameter PP', and the tangent at Q meet PP' in T, then
(1)
(2)
CV.CT = CP\
QF-
:
CV. VT = p PP'
:
[or
CD'
CP^].
pI
(1)
Since
QT is
the tangent at Q,
:
..
TP TP + TP' TP ~
:
'
= PV
,
:
[Prop. 13]
TP' =
PV + P'V PV ~ P'V-
2CP:26T=26T:2CP;
and
for the ellipse or circle,
2CT:2GP = 2CP:2GV;
therefore for
all
three curves
CV,CT=CP\
TANGENTS.
(2)
29
Since
Avhence
or
..
CP,
or
and
But
..
Cor.
PV:VT=^CV:P'V, CV.VT=PV.P'V. QV PF. P'F= /) PP' (or CD' CP*). [Prop. 8] QV (7F. Fr = ^j PP' (or CD CP'). VT is equal to the ratio It follows at once that QV
:
:
compounded
of the ratios
PP'
(or
CD'
CP') and
C7
QF.
Proposition 15.
[I.
38, 40.]
If Qv be the ordinate to the diameter conjugate to PP', and QT, the tangent at Q, iiieet that conjugate diameter in t, then
(!)
(2)
(3)
Cv.Ct=CD\ Qv' :Cv.vt = PP' :p [or CP' CD'], tD tD' = vD' vD for the hyperbola,
: :
and
have (1)
tD
tD'
= vD vD' for
:
the ellipse
and
circle.
for the
preceding proposition, we
QV
CV. VT = CD'
CP'.
[Prop.
U]
But
and
..
QV:CV=Cv:CV, QV:VT=Ct:CT;
QV
.
CV.
:
VT= Cv.Ct
:
CV.
CT.
Hence
Cv Ct
CV. CT = CD'
CP'.
[Pn.p. 14]
And
(2)
CV.CT =CP';
..
Cv.Ct
= CD\
:
As
before,
QV
But
(or;)
PF).
30
THE
COyJC.S
OF AIOLLONIUS.
:Qv;
and
QV: VT = vt
Qv'
:
.-.QV'.CV.VT=Cv.vt:Qv'.
Hence
Again,
Cv
vt
= CP' = PP'
''
;).
(3)
and
..
Thus
and
Cor.
tD
iD'
It follows
tD'
:
= vD'
vD vD
iZ)
vD'
c?Vcie.
Gv
compounded
of the ratios
PP'
(or
GP^
CZ)'^)
and
i/i
Qv.
PROPOSITIONS LEADING TO THE REFERENCE OF A CONIC TO ANY NEW DIA:\IETER AND THE TANGENT AT ITS EXTREMITY.
Proposition 16.
[I.
41.]
In a hyperbola, an ellipse, or a circle, if equiatir/alar parallelograms (VK), (PM) be described on QV, GP respectivehj, and
tneir
|^= ^^
[...
%.
{VN)
to
atid if
ated
(PM),
{VN){VK) = {PM),
the lower sign
applying
to the hyjjerbola.
Suppose
to be so taken on
KQ produced
: .
that
so that
Thus
or
QV.QO = PV.P'V (1). Also QV: QK = {CP CM) (p PP') = (CP CiM).{QV: QO), (QV QO) .{QO:QK) = (CP CM) (QV QO)
:
:
But
and
(2).
32
..
CP'
ivom
i\).
Therefore, since
PM,
:
VK are
:
equiangular,
:
Hence GP' +
and the
and hence
so that or
{VN) {PM) = {PM) + {VK) {PM), ( VN) = {PM) + { VK), {VN){VK) = {PM).
:
:
is
order to show his method of dealing with a somewhat complicated problem by purely geometrical means.
is
follows.
We
and
have
CD'
QK GP^'CM'
3r
CP.GM'
:
CD'
^''^''CP.CM
or
GV'~GP
CD'
GP\
QV.QK = GP.GM{^l'-l
..{VK) = {VN)-{PM),
{VN){VK) = {PM).]
TRANSITION TO
NEW DIAMETER.
33
Proposition 17.
[I.
42.]
to
RW
the diameter
RU
parallel to
it
respectively;
and
if through
Q a
in
the diameter he
in
drawn meeting
RW produced
and
the tangent at
E, then
QT is
..
a tangent,
(1).
ZA
and
..
Proposition 18.
[I.
43, 44.]
In a hypei'hola, an ellipse, or a circle, if the tangent at Q and the ordinate from Q meet the diameter in T, V, and if he the ordinate from any point R and RU he parallel to QT ; if and the parallel to it through meet CQ in F, also
RW
RW
respectively, then
H.
C.
THE
.?
OF APOLLONIUS.
We
whence
have
:
QV
parallels,
CV. VT = p PP'
:
.
[or
CD'
OP'],
QV VT = (p
by
therefore,
RW:WU={p: PP')
of the triangles
(CP PE).
:
that proposition.
follows
that
the same
true of the
triangles themselves,
..
CFW ~
CPE =ARUW.
the exact significance of this
proposition, which
is
transformation of coordinates.
this:
The
proposition amounts to
If
GP, GQ are
fixed semidiameters
and
is
a variable
all
GFRU
constant for
positions of
on the
conic.
CQ
If
are
{CP being
the axis of
a).
we
draw RX parallel to CQ to meet GP and RY parallel to CP to meet CQ, the proposition asserts that (subject to the proper
convention as to sign)
But, since
or
or
or
ARXU= yy-.
TRANSITION TO
Heuce,
if x,
NEW DIAMETER.
li,
3i
/
[I.
/=
A,
which
is
Proposition 19.
45.]
If
to it
the tangent at
Q and
respectively,
and
R parallel Qv, Rw
PP', meeting
if also
Rw
meet
CQt.
36
and the triangles QvC, Qvt are the halves of equiangular parallelograms on Cv (or QV) and Qv (or CV) respectively: also
CPK is
and
the triangle on
CP
similar to Qvt.
CQv =
is
RW,
AC/iu= A Ruw ~ A
CPK = A Ruw -
CQt.
Proposition 20.
[I.
46.]
In a parabola
parallel
to the
any point
at the point.
parallel
let it
Q and
meet the diameter PF in U. Let QM drawn parallel to PF meet RR' in 31, and the straight lines drawn ordinate-wise through R,
R',
in F, F',
respectively.
We
and
ARUW=njEW,
AR'UW' = CJEW\
W'R' = Therefore, by subtraction, the figure R away the common part R'W'WFM, and we have
P' W.
Take
RMF= A
And R'F'
is
R'MF'.
parallel to
RF;
..RM=MR'.
TRANSITION TO A
NEW DIAMETER.
.37
Proposition 21.
[I.
47, 48.]
In a hyperbola, an
point
to the
ellipse,
or
joining any
tangent at the
point
38
the figure
is
drawn
for
the hyperbola,
AR'U W = quadrilateral
W'F';
by subtraction, the figure F'W'WF= the figure R'W'WR. Taking away the common part R' WFM, we obtain
AFRM = AF'R'M.
And,
.
FR, FR'
are parallel,
RM=MR'.
(2) as the figure is
drawn
AGPE-ACFW = ARUW,
ACRE - ACFW = AR'UW,
..
,
by subtraction,
ARUW-\-
common
and,
AFRM=AF'R'M,
and, as before,
(3) if
RM = MR'.
is
RR'
and
Q
We
QG
RR'
branch,
GQ
will
bisect
RR'
is
parallel
to
the
tangent at
Q'.
Q is parallel
follows.
We
and
have
14],
and
10];
CT=CT'.
it
as CQT,
CQ'T',
follows that
parallel.
TRANSITION' To A
NEW DIAMETER.
30
Proposition 22.
[I.
49.]
Let the tangent to a parabola at F, the extremity of the ainginal diameter, meet the tangent at any point Q in 0, and the
parallel through
to
the diameter in
and
let
RR
U
he
any
chord parallel
to
;
the tangent at
meeting
PT
in
and
EQ
produced
in
then, if
UQ:QE=p':2QT,
it is to
he proved that
RM' = p'.QM.
In the figure of Prop. 20 draw the ordinate
Q V.
Then we
have, by hypothesis,
Also
Therefore the
Add
..
to
the quadrilateral
Subtract the
CJQU= ARMF,
.
and hence
RM MF = 2QM
: :
QT
(1).
But
or
RM' =
QM.
Proposition 23.
[I.
50.]
If in a hyperhola, an
centre in
;
ellipse,
to the
QL (= p)
:
OQ QE = QL 2TQ
40
and
to
Q' is on
QC if further Q'L be joined {wJiere QC produced and CQ= CQ'), and MK he drawn parallel
;
QL
to
meet Q'L in
(where
to
is the
point of concourse of
CQ and RR,
to he
a chord parallel
the tangent at
Q): then
it is
proved that
RM' = QM.MK.
In the figures of Prop. 21 draw
ing
CHN parallel
ii!
to
QL, meet-
QL
in
and
in F.
meeting
CQ
Then, since
Also
CQ = CQ\
RM:MF=QH:QT
(A).
Now
and
circle,
ARUW=QTWF,
..
,
ARUW = ACQT-AGFW;
..
subtracting
3IUWF,
we
have
and, subtracting
ARMF=QTUM.
have
It will
triangles
is
[III.
But Eutocius gives another proof of Prop. 18 which, he says, appears in some copies, and which begins by proving these two triangles to be equal by exactly the same method as is used in our text of the later proof. If then the alternative proof is genuine, we have an explanation of the assumption here. If not, we should be tempted to suppose that Apollonius quoted the property as an obvious limiting case of Prop. 18 [I. 43, 44] where II coincides with Q but this would be contrary to the usual practice of Greek geometers who, no doubt for tlie purpose of securing greater stringency, preferred to give separate proofs of tlie limiting cases, though the parallelism of the respective proofs suggests that they were not unaware of the connexion between the Compare Prop. 81 [V. 2], where general theorem and its limiting cases. coincides with B, though we have Apollonius proves separately the case where for tlie sake of brevity only mentioned it as a limiting case.
actually proved.
;
TRANSITION TO A
NEW DIAMKTKR.
41
Now
QT
..QH + ^fN
:
..
..
[by
QM{QH + MN) QM{QT+MU) = RM' RM.MF; RM* = QM(QH + MN) (B)] = QM.MK.
opposite branch of the hyperbola.
to
to
PE.
:
Q'E'
=OQ:QE=p' 2QT=p'
:
2Q'r,
follows.
straight lines
drawn
straight lines
drawn through the centre are diameters also that the conies can each be referred indiiferently to any diameter and the
tangent at
its
extremity as axes.
(Problem.)
52, 53.]
and
straight line
is
to
be the
axis.
Let
AB
given length.
Produce
proportional
to
so that
AC
pa-
>
and
let
:
be a mean
:
between
AG
and
(Thus pa
or
AC = S'
AG^,
it
and
AC>lpa,
Avhence AC'^
>
--
2AG > S,
so that
is
possible to describe
to
an
isosceles triangle
S.)
AO = AG, DC = S.
AGOE, and
about
as
diameter, in a plane
AUG,
describe a
circle,
and
let
as
PROBLEMS.
apex and the said
cone because
circle as base.
43
Then
the cone
is
a right
Produce
OE = AG = OA. OE, OA to H, K,
This plane
and draw
parallel to
AE,
and and
let
HK
parallel to the
AB
at right angles in
N.
Now
.-.
Hence
AE = AE: AO, since AE= 00== S,AO = AC; pa:AO = AE':AO' = AE':AO.OE. PAP' is a parabola in which ;; is the parameter
Pa.
of the ordinates to
AB.
[Prop. 1]
Let the
line
which
is
to
be the diameter be
let
PM,
let
parameter, and
MP
be produced to
to
so that
PF = ^p.
FT
Make
the angle
FPT
bisect
perpendicidar to TP.
dicular to
PM, and
PN perpen-
TN;
perpendicular to
FP
Now
with axis
A and draw LAE through A PT in and let NA.AL = PN\ AN and parameter AL describe a parain
TN
meeting
since
PN^ = LA AN.
.
Also
is
PT
be a tangent to
Therefore
it
since
is
AT = AN.
And
PM
parallel
to
AN.
PM
the
a dia-
to
tangent
PT, which
OEP
..OP:PE=FP:PT,
= p:-2PT,
by hypothesis.
Therefore
is
the diameter
PM.
44
(Problem.)
Giveti
a straight
line
AA'
in
is
a given angle.
Let AA', Pa be the given straight lines, and let a circle be drawn through A, A' in a plane pei-pendicular to the given plane and such that, if G be the middle point of A A' and DF
the diameter perpendicular to
'.
Then,
1sr
'.
for
if not,
suppose
DC:GG = AA':pa
Draw GO
parallel to
(GG being
less
than GF).
Join
circle in 0.
AG,
AE
parallel
in B,
to
DO
A A'
PROBLEMS.
45
Then
Z0EA = ZAOD=
..
AnD=zOAE:
OA = OE.
for apex and for base the and whose plane is perpendicular
whose diameter
AE
AOD.
to
The cone
will therefore
be right,
since
OA =
OE.
Produce OE,
OA
//,
and draw
parallel to
AE.
Draw a
circle
plane through
HK
will
AOD.
PP'
This plane
Avill
at right angles to
A'A
produced.
Let
GO
meet
HK in M.
Then, because
meets
HO produced
PAP'
is
a hyperbola.
And
=
But
.0:0'
OM MH] OM
:
= A'B.BA :B0\
A'B BO = BA:BO =
:
.,
^^
'''''^'^' '"'''"^^"'
..
A'B.
BO'=OIiP HM.MK.
: :
Hence
Therefore
pa,
is
HM
MK.
PAP'
cor-
[Prop. 2]
Let
PP',
lines,
OPT
is
C
to
On CP
that, if
let
NH
drawn
parallel to
PT
meet
CP
produced
in
H,
NH':CH.HP=p:PP'*.
* This conetruction
we may
infer that
it
is assumed by Apollonius without any explanation ; but was aiTived at by a method simihir to that adopted for
46
Join
NO
meeting
PT
in T,
and take
it
A
to
on
CN
such that
CA^=CT. CN.
'
so that
^..
AC =
CA', join A'K, and draw
Produce
AC
to
A'
so that
EOAM
-^,
through
parallel to
PN
meeting CP,
, A'K
in
0, Jlf respectively.
With AA'
This
will pass
as axis, and
AM as
through
because
PN^ =
AN .NK.
a similar case in Prop. 52. In fact the solution given by Eutocius represents sufficiently closely Apollonius' probable procedure.
If
Nm
HNN'
Thus we have
to
draw
at a given inclination to
:
PC
and so that
N'H:NH = PP'
Take any
straight line
o/3
p.
and divide
it
at
7 so that
a.y = PP':p.
Bisect 07 in
Then
is
draAV
GR
at right
angles to
in
PT
which
and
R.
Then
RF drawn
parallel to
meets
CP
produced in F.
to S" so that
PT will be Divide FR at
the tangent at R.
:
.S'
GR
produce
FR
RS' = RS.
it
meet
CF
in
H.
Then Nil
is
was required
to
find.
The proof
is
obvious.
PROBLEMS.
Also
47
because
PT
CP
be the tangent at
will
CT.CN=CA\
Therefore
chords parallel to
the given angle.
PT and
Again we have
:
2CP = NH'
:
CH HP, by construction,
.
and
2CP 2PT = GH
:
NH
.\ = '..
=
:
^GH.HP.NH.HP;
= OP
therefore
is
HP
by similar triangles
The opposite branch of the hyperbola with vertex A' can be described in the same way.
Proposition 26.
[I.
(Problem.)
60.]
Criven
any
angle, to
lines bisecting
each other
48
THE OOXTCS
f)F
APOLLONIUS.
that
From PP'
hyperbola
draw PL perpendicular to PP" and of such a length PL = DD"' then, as in Prop. 25, describe a double with diameter PP' and parameter PL and such that
;
the ordinates in
it to
PP'
Again, draw
DM
DD' = PP'^
DM perpendicular to DD' of such a length that and, with DD' as diameter, and DM as the
;
to
DD'
PP'
is
conjugate
Proposition 27.
[I.
(Problem.)
Given a diameter of an
corresponding parameter,
its
and
the angle
:
ordi-
nates
to
find the
let
its
parameter.
Let
'
PROBLEMS.
he the diameter and
it,
49
straight line of length
AL,
Pa perpendicular to
the parameter.
circle
on
AA'
Take
meeting
AD on A A'
A
in
equal to
AL.
circumference in 0.
produced in
T.
HKMN
parallel
parallel to A A' meeting the and produce it to meet A'A on OA produced draw Through any point to OE meeting OA', AA', OF in K, M,
and
Join
OFN EO
respectively.
TO A =
and
OEA +
OAE = AA'O +
=
OE,
'= ',
^ OA'E
HK
is
parallel to
whence
and
OH = OKH, OH=OK.
C.
50
circle draAvn
with diameter
the triangle
OHK,
a cone be described.
OH = OK.
Consider the section of this cone by the plane containing
AA', AL.
This will be an
Pn
ellipse.
And
AA' =
AD
:
AA'
= AF:
= TO
AE
TE
'.TO.
Now
and
that
or Pa
is
'. = T0': TA TO.TA = HN NO, = TO NO, by similar TO' TA.TA' = HN.NK:NO\ j)a:AA' = HN.NK:NO\
;
:'
triangles,
[Prop. 3]
Secondly,
still
meter, let
them be BB',
BM respectively.
Let
C be
',
through
it
draw^^',
AA" = BB'.BM:
and draw AL,
parallel to BB', such that
BM
thus
BB'
= AA' AL
:
Now
with
'
it,
PllOHLEMS.
as diameter
ellipse in
51
and
AL
as the corresponding
parameter describe an
perpendicular to
'
are
as above.
= BC":AC.CA',
(2)
BM
BB' = AC'
BC
= AC':BC.CB',
so that
BM
is
Thirdly,
let
CPT, where G
;
is
given diameter
ing to PP'.
PP'
and
let
PL
Take a point N, on the semicircle which has CP for its drawn parallel to PT satisfies the
NH
:
NH' CH.HP
*
PL
25
is
PP'*.
\vithont explanation.
This construction
assumed
If
semicircle on
CP
42
2
Join
that
CN and
Draw
produce
GT.CN = CA\
it
to
so that
AN'.NK = PN\
CA
in
Join
AM
in
through
perpendicular to
E,
(and
0,
therefore parallel to
PT in
an
and A'
produced
axis
Then with
as in the
required.
first
A A'
AM describe
This
will
ellipse
be the ellipse
For
through
a similar reason,
will
For and
GA' = GA.
(2)
PT will
We
be the tangent at
]3
:
GT GN=GA\
.
.
(3)
have
2CP = NH^
:
and
2GP 2PT =
:
..
ex aequali
NH:HP
:
= OP PE.
Therefore
is
to
drawing
NHN'
PT)
so
N'H:NH = PP':p,
tiie
and
method shown
mutatis mutandis.
ASYMPTOTES.
Proposition 28.
[IL
1,
the corre(1) If PP' he a diameter of a hyperbola and sponding parameter, and if on the tangent at there he set off on each side equal lengths PL, PL', such that
PU = PL" = ip
and are accordingly
(2)
(3)
PP' [= GB'l
any
finite point
defined as
asymptotes.
the
same asymptotes.
(1)
CL meet
the hyperbola in Q.
Draw
the
ordinate
QV, which
will accordingly
be parallel to LU,
:
Now
p. PP'=p. PP'
:
PP"'
54
p:PP' = QV':PV.P'V.
PV.P'V=CV\
i.e.
CV - CP' = C]^,
is
which
is
absurd.
Therefore
point,
GL
any
finite
= P'M" = CD\
is
are asymptotes.
Now MM',
straight line.
PL = P'M,
and PGP'
Therefore
LGM is
a straight
line.
So
also is
tAvo
conjugate
hyperbolas.
at P,
P, D, U.
Then
[Prop.
diagonals of
LM,
Also
Therefore
GD.
which PP'
DD'
its
conjugate.
Similarly
DD'
a transverse
diameter and
Therefore
conjugate,
i.e.
conjugate
hyperbolas
have
their
asymptotes
common.
ASYMPTOTES.
Proposition 29.
[II.
2.]
No
luithin
the
angle
between
the
If possible, let
CK
be an asymptote.
Draw from
the
straight line
through
and
Then, since PL = PL', and RR, LL' are parallel, iiF= R'V, where V is the point of intersection of and GP.
RR
And, since
Therefore
PKRL
..
is
a parallelogram,
PK = LR, PL = KR.
(1).
QR > PL.
RV"-
AhoRQ>PL';
:
Again
and
thus
RQ.QR'>PL.PL', or GV' = PU GP'=p:PP', P PP' = QV'.PV.P'V = QV':GV'-GP': RV GV' = QV':GV'-GP' --^RV'-QV: GP';
PL' .GP' =
whence
which
is
impossible,
Therefore
GK cannot
be an asymptote.
56
Proposition 30.
[11.
3.]
If a
straight
line
touch a hyperbola
;
at P,
it
will
meet
LL'
luill
he bisected at P,
and
Pr = ip.PP'[=GD'].
is
[This proposition
on the tangent
lengths
PK, PK'
are asymptotes
impossible.
K' must
on the asymptotes.
on a hyperbola,
to
find
and
Let GL, GL' be the asymptotes, Produce PG the point. Draw to P' so that GP=GP'. parallel to GL' meeting GL in K, and let GL be made equal to twice GK. Join LP and produce
PK
it
to L'.
Take a length
LL'^
such
with
that
=p.PP', and
/.//.
diameter
that the
ASYMPTOTES.
57
Proposition 32.
[II.
8,
10.]
If Qq
be
any chord,
it will,
the asymptotes in
two points as R,
will
and
(1)
QR, qr
(2)
RQ.Qr = lp.PP'[=CD'l
Tako
CV
meeting
the curve in P.
Then
CF is a
parallel to Qq.
[Prop. 11]
meets
L,
it
asymptotes
(in
L').
also
Then
to LL',
(1),
since
Qq is parallel
it
and
LP = PL',
follows that
RV
th(
58
Proposition 33.
[II.
11, 16.]
asi/7)ip-
If
totes in
Q,
K, K', and if CF
(1) (2)
KQ.QK' = CP\
QK=Q'K'.
meeting the asymptotes in L,
L',
Draw
the tangent at
and
let
the chord Qq parallel to LL' meet the asymptote.s in R, is therefore a double ordinate to CP.
r.
Then we have
:
CP' = (PL
CP) (PL'
.
CP)
= (RQ:KQ).(Qr:Q]r)
= RQ.Qr:KQ.QK'.
But
Pr==RQ.Qr;
.KQ.QK' = CP\
[Prop. 32]
Similarly
(2)
K'Q'.Q'K=CP\
. . ;
whence
it
KQ QK' = CP' = K'Q' Q'K KQ {KQ + KK') = K'QXK'Q' + KK'), follows that KQ = ''^'.
..
.
ASYMFrOTES.
59
Proposition 34.
[IT.
12.]
If
angle,
Q, q he
straight lines
any two points on a hyperbola, and parallel to meet one asymptote at any
to
HQ QK = hq. qk.
.
r.
We
But
and
have
..
liQ .Qr
:
= Rq .qr;
:
[Prop. 82]
RQ Rq = qr
:
Qr.
:
RQ Rq = HQ
qr
..
:
hq,
Qr = qk
QK
QK,
qk.
or
HQ hq = qk HQ QK = hq
:
60
Proposition 35.
[II.
13.]
//' in the
straight line be
drawn
it
will
Let
.^
let
EF
be drawn
Then
only.
EF
produced
shall
let
it
not
meet the curve. Take Q, any point on the curve, and draAv QH, QK each parallel to one asymptote and meeting the other let a point
;
F be
taken on
EF such
that
HQ.QK=CE.EF.
Join OF and produce it to meet the curve in q and draw
;
qh,
Then
and
hq.qk
= HQ. QK,
[Prop. 34]
HQ.QK=CE. EF,
:.hq.qk=GE.EF:
is
by hypothesis,
which
impossible,
hq
Therefore
Again,
For,
if
EF will meet the hyperbola in one point, as R. EF will not meet the hyperbola in any other point. possible, let EF meet it in R' as well as R, and let
be drawn parallel to QK.
RM, R'M'
Then
which
is
ER RM = ER'
.
R'M'
[Prop. 34]
impossible,
Therefore
point R'.
EF
ASYMPTOTES.
61
Proposition 36.
[II.
14]
Let
Draw two
in li, r
and
R',
to
meet
Q'q' in F.
Then
and
r'q'
q'R =
rq qK,
.
q'R > qR
..
q'r'
<
qr,
and hence, as successive chords are taken more and more distant
from the centre, qr becomes smaller and smaller.
rH
less
asymptote
Cr.
if
HM will then meet the curve [Prop. 35] in a point M. MK be drawn parallel to Qq to meet Cr in K,
= rH,
And,
whence
MK < S.
62
Proposition 37.
[II.
19.]
Any
tangent
to
the
conjugate
hyperbola
luill
meet
both
of contact.
(1)
will therefore
This tangent will then meet the asymptotes [Prop. 30], and meet both branches of the original hyperbola.
(2)
and the
original hj^perbola in Q, Q.
Then
[Prop. 30]
DL = DM.
LQ = MQ'
;
whence, by addition,
DQ =
!>(/.
ASYMPTOTES.
63
Proposition 38.
[11.
28.]
If
totes in
a cJiord
q, then
Q'Q.Qq'=2GD\
Let
CD
Then we have
RQ.Qr=CD\
RQ'.qr=CD';
.'.
2CD' =
RQ
Qr
+ RQ'
'
= QQ'.Qq.
20.]
If Q
he
CE
he
drawn from
to
hyperhola in E, then
(1)
the tangent at
and
(2)
CQ,
GE will
he conjugate diameters.
Let FP',
let
DD'
QF
to
PP', and
EW
the ordinate
from
T,
t
to
DD'
in
let
the tangent at
meet EU,
CE
in 0,
Let p, p' be the parameters corresponding to PP', in the two hyperbolas, and we have
(1)
DD'
p'
DD' = PP'']
60
and
PP' .p =
:
CV.VT: QV\
..
OD' = CV.VT.QV"- =
EW GW WU. EW CW WU.
: . : .
[Prop. 14]
VT:QV=EW.GW.
Therefore, by division,
the
triangles
CVQ,
EWU
the angles at V,
are equal.
^QCV= ZUEW.
But
VCE =
CEW,
since
Therefore, by subtraction,
(2)
HE:EO = EU
so that
*S'
S,
is
diameter
Also
or
EE'
[Prop. 23]
Ct.QV= GD\
Ct
(since
QV = Cv),
Now
and
Ce
And
..
(A).
But
S:EU=OE:EH
= CQ
..
:
QT.
S.CE CE.EU=CQ'
:
-.CQ.QT.
Hence, by (A),
..
S.CE=CQ\
2S.EE' = QQ'\
that EE'^ is equal to the QQ' and the corresponding parameter.
are conjugate diameters. [Prop. 26]
')
where 2S
is
And
similarly
may be proved
rectangle contained by
Therefore QQ',
H. c.
EE'
66
Proposition 40.
[II.
87.]
the Jf Q, Q' cij-e any points on opposite branches, and middle point of the chord QC/, then Cv is the 'secondary"
diameter corresponding
to
to the
QQ'.
it
to
in
q.
Join
PP'
Then we have
CQ'
Therefore Qq
is
Cq,
and
Q'v
= Qv.
parallel to Cv.
in V.
Now
QV=Cv=Vq,
= Cq.
to
PP'
are parallel
Qq,
and
[Prop. 6]
he drawn to a conic, and V he the middle point of the chord of contact QQ', then TV is a diameter.
EQ
VE be a diameter, meeting TQ' in E. Join meetiug the curve in R, and draw the chord RR' parallel to QQ' meeting EV, EQ' respectively in K, H.
For, if not, let
Then,
.since
RH
is
parallel
to
QQ',
and
QV=Q'V,
RK = KH.
RR'
is
the diameter
EV,
RK = KR'.
:
Therefore
KR' = KH
which
is
impossible.
Therefore
in
EY
is
it
may be
proved
is
Conversely, the diameter of the conic draiun through T, the luill bisect the chord of
separately
proved by
Apollonius by means of
an easy
rediictio
ad absiirdum.]
Proposition 42.
[II.
40.]
If tQ, tQ' be tangents to opposite branches of a hyperbola, and a chord RR' be drawn through t parallel to QQ', then the lines joining R, R' to v, the middle point of QQ', will be tangents
at
R, R'.
68
Join vt. vt is then the diameter conjugate to the transverse diameter drawn parallel to QQ', i.e. to PP'.
But, since the tangent Qt meets the secondary diameter
in
t,
Cv
a = Ip
PP' [=
CD'].
t is
[Prop. 15]
reciprocal,
and
and the
v.
Proposition 43.
[II.
In a
bisect
conic,
or a
circle,
meet
Let Qq, Rr, two chords not passing through the centre, in 0. Join CO, and draw the diameters Pj>, P'p' re-
Then
let
Qq,
Rr
one another.
For,
if
possible,
each be bisected in 0.
(iU
parallel to
is
a diameter
parallel to GO.
is
can
CO.
the
Therefore
which
is
impossible, since
PP'
is
not a diameter.
Therefore Qq,
Rr
Proposition 44.
[II. 44,
(Problem.)
45.]
To find a diameter of a
conic.
conic,
and
the centre of
a central
Draw two parallel chords and join (1) The joining line will then be a diameter.
(2)
and these
will
meet
in,
and
Proposition 45.
[II.
(Problem.)
4G, 47.]
To find
ic.
the axis
(1)
let
PD
be any diameter.
Draw any chord QQ' perpendicular be its middle point. Then let
thr(jugh
parallel to
PD, and
drawn
axis.
AN
V^
PD
will
be the
and, inasmuch as
it is
it
bisects
the
a.xis.
And
there
is
70
(2)
conic,
on the
C and
radius
CP
describe a circle
Let PP',
PQ
be two
let iV,
common
Then ON,
CM
will
They
are also
Proposition 46.
[II.
48.]
axes.
No
Through P'
to
curve again in R.
71
CL
is
an
axis,
PL = LR\
CR.
it is
therefore also
CP =CP' =
Now
PP'
cannot meet the same branch of the hyperbola in any other points than P, P'. Therefore the assumption is absurd.
In the ellipse draw
axis
RK,
PP'.
PH
which
is
parallel to
it
Then, since
CP =
CR,
CP' = CR\
or
KR \
(1).
.\CK'-CH' = HP'-KR'
Now
and
BK.KB' + CK' = CB \
BH.HB'
..
CH'=CB\
CK' - CH' =
BH HB' - BK
. .
KB'.
KB', from
(1).
Hence
But, since
.'.PH' = BH.HB',
and
..
RK'=- BK.KB'.
P,
BB'
which
is
absurd.
Hence
CL
is
not an axis.
72
Proposition 47.
[II.
(Problem.)
49.]
To draw a tangent
outside the curve.
(1)
to
on the curve.
DraAv
so that
per-
A to
AT = AN.
PT
Then, since
In
the
AT=AN, PT is
case
the tangent at P.
coincides
[Prop. 12]
particular
where
with
is
A, the
the tangent.
the diameter
ecpial
Let the given point be any external point 0. Draw OBV meeting the curve at B, and make BV to OB. Then draw through V the straight line VP
to the tangent at
parallel
[drawn as
in (1)]
meeting the
curve in P.
Join OP.
OP
is
the tangent at B,
an ordinate
to
to
[Prop. 12]
73
Proposition 48.
[II.
(Problem.)
49.]
To draiu a tangent
or outside the curve.
to
cases.
Case
I.
Q ou
the curve.
Draw
take
QN
A A'
is
A A'
-.
produced, and
:
on
such
that
A'T
AT = A'N
A
AN.
Join TQ.
Then TQ
the tangent at Q.
[Prop. 13]
coincides with
is
or
A' the
the tangent.
Case
Join
II.
P, P'.
CO and produce it both ways to meet the hyperbola Take a point V on CP produced such that
in
P'V:PV=OP':
OP,
[drawn
and through V draw VQ parallel to the tangent at Join OQ. as in Case I.] meeting the curve in Q.
Then, since
QF
is
parallel to the
tangent at P,
QV
\s
an
P'V:PV=OP'
Therefore
OP.
[Prop. 13]
OQ
is
the tangent at Q.
74
THE COMCS OF
Al'OLLUNlU.S.
Case
Bisect
III.
at
(J
CO
H, and through
draw
HP parallel
to the other
in P,
Join
OP
and produce
it
to
in L.
OP PL = OH HC,
:
:
whence
OP = PL.
OL
touches the hyperbola at P.
lie
Therefore
angles
Join CO.
its
parallel to
CO, and
let
be
middle point.
is
Draw through V
Then
PP'
produced a point
draAv through
such that
w
li.
the straight
to CO. Now take on OC CO Cw = ^p PP' [= C'Z)*], and line wR pai-allel to PP' meeting
. .
the curve in
Join OR.
it,
Then, since
.
Rw
is
parallel to
is
CP
and
Ciu conjugate to
while
CO Cw = CD^, OR
the tangent
[Prop. 15]
at R.
75
Proposition 49.
[II.
(Problem.)
49.]
To draw a tangent
outside the curve.
to
an
ellipse
curve,
There are here two cases, (1) where the point is on the and (2) where it is outside the curve and the con;
I.
and
II.
When
the point
is
Proposition 50.
[II.
(Problem.)
50.]
angle equal
I.
conic
making with
the auis
an
let
DEF
be the given
acute angle.
Draw
DF
EF
at
H,
and join
DH.
the axis of the parabola, and
the angle
Now
angle
in P.
make the
Draw
that
Then
to prove that
76
Since
or
..zFTN = zDEF.
II.
Then,
DEF
must be
gi'cater
DEF
let
be the given
be so taken
angle and
DF
on
DF
that
HEF=zACZ,
Let
or
half the
angle between
the asymptotes.
asympt(jte in Z.
he the tangent at
meeting an
77
CA^
AZ'
(or
CA'
CfB')
:
= EF' FH\
:
on
FE
KF':FD^>CA':AZ\
DK
angle
angle
ACZ.
Hence,
the angle
DKF
is
less
than the
the hyperbola in
some point P.
has to be taken on
:
EF produced
the
CB'
= KF .FE FD\
and from
this point
constructions are similar for both the central conies, the angle
AGP
DKF in each
case.
Draw now
tangent PT.
PN
Then
78
First,
let
CP
and
bisects
tangent at
= '. ,
Therefore the
is
.PC
be
parallel
to
BA'.
Then, by
parallels,
parallel to
PC
not parallel to
A',
and we
have
ZPCN^
'
:
PN
BAG.
..
whence
PK' CN'
.'.
[Prop. 14]
CN^NT.
Let
FDE be
FDF
ABA', and
let
DG
circle
bisecting
FE at Divide FE in
EiM
:
right angles in /.
MF = GN
NT,
the chord
From
draw (JL
perpendicular
to
HK,
DFI,
and join
EH, HF.
The
triangles
BAG
:
are
FP
Now
..
:
ID' = GA'
:
GB\
since
OD 01 > LH LM,
01)
01 = LM.
:Df<LH:
79
DG:DI<HK -.HM,
whence
But
..
CN.
NT FN' < KM MH
:
:
<KM.MH:MH'
Let
where
is
some point on
It follows that
produced.
Join
ER, RF.
Now
and
(since Ci\r
:
iVT
EM
..
J/i?^).
CN :EM = PN:RM.
CPN,
PTN,
and
ZCPT= ^ERF;
it
whence
follows that
CPT is
is
less
than
is
EHF,
or
^5^'.
not, the
Therefore, whether
CP
parallel to
A' or
CPT
ABA'.
51, 53.]
To draw a tangent
iDitli
to
any given
conic
making a
given angle
of contact.
parallel to the axis,
I.
80
II.
CPT
it
acute
the
Jiyperhula,
and
for
the
ellipse
be
less
proved in Prop.
Suppose
angle
to
first
is
ABA'.
draw
CP
parallel to
A' (or
AB) and
to
draw through
(or A'B).
Next suppose
and
for
to be
e(|ual to
ABA': and .
SI
Imagine a segment of a
circle
(EOF)
equal to the
angle
Then,
a point
D
if
:
on the
DM be
DM^
is
EF, the
ratio
EM .MF
:
CB\
i.e.
to the ratio
GN .NT PN\ we
should have
CPT =
and
ON NT
.
PN' =
= \
EOF,
:
EM MF
.
and
it
PCN,
PTN
are respectively
similar to
DEM, DFM*.
PCN.
DEM
would be
The
Draw CP
PCN
is
DEM,
Also
'
let
Then
GN NT PN' =
.
CA'
GB' =
EM. MF DM\
:
PGN, DEM that the tiiangles PTN, DFM the triangles GPT, EDF*.
and the triangles
..
It
the angle
necessarily
PCN
DEM, CP must
DEM
If
is
is
less
per-
we have
EM. MF DM' = CA' CB' = GA' AZ\ EM' DM' > GA' AZ\
: : :
..
DEM
is less
We
have now shown that the construction reduces itself D on the segment of the circle, such that
EM.MF-.DM'^CA'-.GB'.
for granted
proved.
H. C.
t)
82
This
as follows
in
Take lengths
, /3
a/3
;
yS7
= CA' CB\
:
for
the
hyperbola
and away
from
the ellipse
and
let
be bisected in
.
to
EF\ and
Draw 01 from 0, the centre of the circle, perpendicular such that on 01 or 01 produced take a point
OH: HI =
(the points 0,
By: /3,
relative to one another
H, I occupying positions
, , ).
Draw HD parallel to EF to meet the segment in D. Let DK be the chord through at right angles to EF and meeting
it
in
M.
Draw
Then
OR
bisecting
:
DK at right angles. RD DM OH HI = 8y ^.
=^
:
KD DM = 7
:
7yS
:
so that
KM DM =
:
^.
Thus
:^
= CA' CB\
:
is
found.
CA'= CE^,
or
i.e.
we have
EM. MF = DM\
DM
Note.
between / and
FLI = lz CRT,
..
which
is less
is less
than ^
ABA'
FLI
than
ABC,
83
whence
CA'
OB" >
FP
f/
fiJ
>L'l :IL.
It follows that
:
y>
fL,
:
so that
7^ > L'L
IL,
so that
:^>:.
if
7^ > OL
LI,
Hence,
8 .^ = : IH,
is less
than IL.
62
EXTENSIONS OF PROPOSITIONS
Proposition 53.
[III.
1,
1719.
4,
13.]
P, Q being any two points on a conic, if the tangent at (1) and the diameter through Q meet in E, and the tangent at Q and the diameter through in T, and if the tangents intersect at
0,thm
(2)
AOPT = AOQE.
If
be
any point on a hyperbola and Q any point on and if T, have the same significance CPE = CQT.
be the ordinate from
(1)
Let
QV
to
the
diameter
through P.
Then
for the
parabola we have
TP = PV,
so that
[Prop. 12]
TV=2PV,
CJ
and
EV = AQTV.
EXTENSIONS OF PROPOSITIONS 17
Subtracting the
19.
85
common
area
OPVQ,
AOQE = AOPT.
For the central conic we have
GV.CT=CP\
or
..
CV :GT=GV':CF']
ACQV:ACQT = ACQV:AGPE;
.'.
AGQT = AGPE.
OTGE
and each
AOPT = AOQE.
(2)
GD
parallel to the
UNIV.
86
tangent at
to
THE
CO^V/OS
OF AFULLUNIUS.
meet the conjugate hyperbola in D, and draw PE meeting CP in V. Then CP, CD are conjugate diameters of both hyperbolas, and QF is drawn ordinate-wise to CP.
QV
also parallel to
CV.CT=CP\
or
CP:CT=CV:CP
= CQ:CE; GP.CE=CQ.Cr. PCE, QCT are supplementary ACQT = ACPE.
..
And
the angles
Proposition 54.
[III. 2, 6.]
// we keep
and if
he
EXTENSIONS OF PROPOSITIONS 17
any other point on
to the
10.
87
to
RU be
and
drawn parallel
let
QT
to
in U,
a parallel throu(/h
R
in
meet
QT and
H, F,
W respectively.
RU
Then
in
M.
Then, as in
adding
HM,
Proposition 55.
[III.
3,
7, 9,
10.]
//'
we keep
to
the
same notation as
and
I
H'
sponding
H, F,
etc.
and
if,
further,
RU,
R'W
intersect in
and R'U',
RW in
It will
J, then the
quadnlaterals F'IRF,
lUU'R'
JU'UR.
R
which case F'IRF must meaning the diflfereuce between the triangles F'MI, RMF.]
in the margin, in
be
taken
as
I.
We
have
in figs. 1, 2, 3
[Prop.
.54]
quadrilateral
H'TU'R',
F'H'HF=H'TU'R'~HTUR
= IUU'R' + (IH);
F'IRF = IUU'R'
(1).
88
THE
CONIL'S
OF APOLLONIUS.
FJR'F'=JU'UR.
Fig. II.
1.
In Hws.
4, 5,
we have
so that
GQT = quadrilateral
CU'R'F',
we have
Fig. 5.
Similarly
HFQ = HTUR;
F'lRF^IUU'R
.(1).
Thus
e.g. in fig. 4,
AH'F'Q" ..
In
fig. 6,
F'H'HF = H'TU'R' -
AHTW+ ARUW,
Fig.
(>.
90
= IUU'R'
Then, subtracting (//) from each side in fig. tracting each side from (IJ) in figs. 5, 6, we obtain
4,
(1).
and sub-
FJR'F'
(the quadrilaterals in
triangles
=JU'UR
6 being the differences
(2),
fig.
between the
triangles
JU'W,RUW
same
respectively).
III.
The same
in the case
manner
where P,
6 above.
Cor.
coincides with
Proposition 56.
[III. 8.]
//'
Q, Q' be
latter
to the
PP', QQ' be two diameters and the tangents at P, P', drawn, the former two meeting QQ' in E, E' and the
two meeting
tangent at
PP'
in T, T',
and if
P'
in
through Q'
to the
in K', then
laterals
(E'K), {T'K').
CPE
are equal
[Prop.
53] and
have a
common
vertical angle,
CQ.CT=CP.CE;
..
CQ CE = GP
'.
GT,
EXTENSIONS OF PROPOSITIONS 17
19.
91
whence
QQ'
i.s
EQ = PP'
TP,
AQQ'K'
AQEO = APP'K
And
..
AQQ'K' = APP'K,
CPE, we
obtain
(1).
(EP')
= (TQ')
triangles CP'E',
CQ'T' respectively, we
{E'K)
{T'K')
(2).
Proposition 57.
[III.
-r>,
R,
secondary diameter.)
(I)
Let Gv be the secondary diameter to which the ordiLet the tangent at Q meet it in t, and
the ordinate
parallel to Qt,
Rw meet Qt meet Cv in a.
in
h and
CQ
in
/'.
Also
let
Ri,
Then
[Prop. 19]
(A)
92
..
(2)
Let
R'lu'
h,
corresponding to
be another ordinate, and h', w' &c. points &c. Also let Ru, R'lu meet in i and Riu,
R'u
j.
Ah'Qf = }itiifR',
and
AhQf = htuR.
Therefore, subtracting,
f'h'hf
and, adding (hi),
= iuii'R (hi)
(1).
fiRf=mu'R'
If
we add
{i}) to
each,
we have
(2).
is
fjR'f=ju'uR
[This
is
on the conjugate
hyperbola, and
CwRM
ACuM'-
ARfM=
ACQt,
proof.]
EXTENSIONS OF PROPOSITIONS 17
19.
93
Proposition 58.
[III.
15.]
same properties as
hold, viz.
those
formu-
still
ARMF^ ACMU=
and
ACQT,
F'IRF=IUU'R'.
parallel to
parallel to
Let D'D" be the diameter of the conjugate hyperbola R U, and let QT be drawn and from D' draw DG
;
PE
to
meet
CQ
in G.
Then D'D"
is
the diameter
conjugate to GQ.
Let
sponding
be the parameter in the conjugate hyperbola correto the transverse diameter D'D", and let be the
in the
I CQ = CD", and
.
CD'
CC^.
we have
[Prop, 23]
94
D'C:CG = ^:QT
= ^.CQ:CQ.QT
= CD":GQ.QT.
Hence
or
DV.CG=CQ.QT,
AD'CG= AOQT
Again.
(1).
CM.MU=CQ.QT
= (CQ:
!).(/;
:2)
MF)
(2).
= (p'.D'D").{OQ.QE)
= (p
Therefore the triangles
tively
D'D") .(R3I:
GMU, RMF,
half
of
equiangular parallelograms on
CM
(or
Rv),
RM (or
two of which are similar while the two are connected by the relation (2), have the
(3).
we have,
by subtraction,
or
RJFF = RUU'J.
(4)
F'IRF=IUU'R'
Proposition 59.
[III.
Case
J,
I.
If OP,
OQ
be two tangents to
to
an internal or e.dernal
point, then
Prop. 55.
We
and
..
have then
RJ.Jr = RW'^JW\
RW':JW'=ARUW: AJU'W;
But
..
RJ.Jr
OP' = JU'UR:
.
AOPT
AJFM',
.
(1).
Again
and
m
But
A R'F'M'.
;
R'J.Jr': OQ'
= FJR'F:
AOQE
(2).
96
and
(2),
we have
by Prop.
55,
53.
JU'UR = FJR'F,
and
OPT =
Thus
OQE, by Prop.
:
BJ. Jr
:
or
(b)
R'J.
Jr'.
tively to
OP,
R'r^',
/,
an internal or external
point,
in the
Or-:OQ' = R'I.Ir;:RI.h\.
Hence the proposition
[Cor.
If /, or J,
is
completely demonstrated.
we
Case
t
II.
If
be
a point on
in
t ;
and
the tangent at
meet
GP
tQ'
tq"
= Ri
ir
R'i
ir'.
Using the
we have
Ri.ir
and Mi^
:
= Mi''-MR\
:
Hence
Therefore,
Ri
if
ir
MfR.
QC,
qq' (both
produced) meet in L,
(1).
Ri.ir:tQ'=fiRf: AQtL
Similarly,
R'i
.
.
ir'
R'w" = iuu'R'
:
.:
R'u'w'
R'i
ir'
tq'
= iuu'R' AtqK
:
(2),
where
qK
is
parallel to Qt
in
K.
97
we have
[Prop. 57] [Prop. 19]
f'iRf= iuu'R,
and
tqK =
..
CLt
+ ACQt= A QtL. =
Ii'i.ir':tq\
Ri.ir:tQ'
tQ':tq^
or
Ri.ir:R'i.ir'.
Case
and
III.
If PP'
he
parallel respectively
to the
and
the diameter
PP'
intersecting in I, then
RI.Ir:R'I.Ir' = p:PP'.
If
RW, R'W
:
are ordinates to
PF,
[Prop. 8]
PP' =
RW CW - CP'
:
CW
and Rr,
If OP,
OQ
he tangents to a hyperhola
I,
then
OQ':OP' = RI.Ir.R'I.Ir'.
Using the figure of Prop.
58,
:
we have
OQ'
OQE = RiW
RMF
98
and, in the
OF': A()PT=R'r.Ir':
AR'U'W
AIUW
= R'I.Ir':IUU'R';
whence, by Props. 53 and 58, as before,
()Q':RI.Iv=OP'.R'I.Ir',
or
Proposition 60.
[III.
If Rr, R'r' he chords of conjugate hi/perbolas meeting in and parallel respectively to conjugate diameters PP', DD', then
R0.0r+^^,.RO.0r' = 2CP'
RO.Or
R'O.Or'
1
CP
in
w,
in
K, k
K',
k',
and CD,
respectively.
Draw LPL',
the tangent at P,
in L, L', so that
PL = PL'.
Then
and
.
LP.PL'=CD\
LJ' PL'
:
Now
..
LP CP = K'O
:
OK,
0/..
RECTANGLES
09
[From this point Apollonius distinguishes five cases: (1) where is in the angle LCL', (2) where is on one of the asymptotes, (8) where is in the angle LCk or its opposite, (4) where is within one of the branches of the original hyperbola, lies within one of the branches of the conjugate (5) where hyperbola. The proof is similar in all these cases, and it will be sufficient to take case (1), that represented in the accompanying figure.]
We
:
have therefore
.
Ok'
C'D'
KO
.
Ok + CP'
= K'O
:KO.Ok + CP' W" -0W"-\- R W" - K' W" Ow^ - Kiu' + CP' = K'
.
Ok'
+ K'R
R'k'
= =
R W" RO
.
W"
:
Riv'
- Kw' -
Riv'
.
'+
CP'
Or'
Or'
RK Kr + GP' - RO
.
Or
= RO
whence
2CP'
- RO Or
.
(since
Kr = Rk),
fip2
or
RO
Or ^^,
.
'
we have
RW
and
...
Cid"
Riu''
R'W" +
Cid"
- CD' CF' :
(Rtu'
so that
whence
RO Or' - CD' CP' RO.Or= CD' CP', RO Or' 2CP' RO Or = CD' CP'
.
RO.Or'
RO.Or
CD^-^-CP^-^-^
72
100
Proposition 61.
[III.
If in ai} ellipse or in conjugate hyj^erholas two chords Rr, and parallel respectively to two R'r he drawn meeting in conjugate diameters FP', DD', then
(1)
for the
ellipse
RO' +
Or'
+^^3 {RV +
+
Or")
= 4CP^
-4,
RO^+Or'
or
RO'+Oj''\
^^,
^p,
the hyperbolas
and for
RO' + Or'
R'O'
Or"
= CP' CD\
:
K',
k',
then
K'O'
Or'
= CD' CP\
:
(1)
We
have
for
both curves
CP':CD' =
PW.WP'.RW'
Div'.w'B'
= R'w": = CP' +
W + Dw'
w'D'
INTERSECTING CHORDS.
101
the
for
(taking the
for the
..
upper sign
the
hyperbolas and
lower
ellipse)
CP'
CD'
= CP'
CD' =
CW" +
W.
WP
Cw'
whence,
CP
or
CW" + GW^
+ Cw"
(A),
CP'
CD'
whence
(2)
^pi
We
R'O'
R'O'
= 2CP' -(CW" + CW) Cw" + Cw^ = ^CP' - {RO' + Or') {RV + Or"), RO'+Or'_^ R'O' + Or" =4
: :
jjjji
{B).
Now
and
+ Or" = K'O' + Ok" + 2CD'. - K'O' = R'K" + 2R'K' K'O, Or" - Ok" = r'k" + 2r'k' k'O
. .
= R'K" + 2R'K'.kO.
Therefore, by addition,
R'O^
Or'-^
...
_ K'O' - Ok" = -IR'K' {R'K' + K'O + = 2R'K'.R'k' = 2CD'. R'O' + Or" = K'O' + Ok" + 2CD\ + Ok" + 2CD' RO' +
:
Ok')
whence
K'O'
Or'
= CD'
CP',
HARMONIC PROPEllTIES
(JF
Proposition 62.
[III.
:}(),
31, 82,
:v.i
U.]
TQ,
point
T(j
an asymptote meets the curve in R' and the parallel through to tlie chord of contact in N, then
I.
Let
is
CV meet
to
PL,
which
RW,
R'W
CP.
RW
'TWR
:
are similar,
TW = PL'
..
CP'
It will
begins with two particular cases of the general property in Prop. 64, namely
(<i)
is
parallel to
an asymptote,
i.e.
parallel to
an asymptote,
an aHymptute, or a tangent at
infinity.
103
CV.CT=CP',
or
whence and
It follows
TW=
by
parallels that
WV.
TR = RM
Again
..
GP'
PU =
WV
.
(1 ).
;
W'R"
:
so that
And
W'R'\
6^ = CF.Cr+lF^
TW = WV,
NR' = R'V
(2).
whence, as before,
and
II.
Next
let Q,
P'P
be
the diameter parallel to Qq. Draw the tangent PL, and the ordinates from R, R', as before.
Let TM,
GP
intersect in
K.
and
Hence, adding
GP' GP'
..
PU = KW WK\
GD' =
PW
WP' WR'
:
GP\
But
by similar
triangles.
Therefore
Tw'
104
whence
Tw Cw = CV,
..
or
\ = wV\
(1).
:
TR = RM
Again
rP'
W'P'
R' W"
GP"
..
PU = R'w'^ :w'V'\
w'V = Cw" + cv.cr,
Tw'
wliciicc, as before,
= w'F,
(2).
is
and, by parallels,
III.
NR' = R'V.
an asymptote,
is
The
a tangent at
as follows.
separately
proved
Draw PD,
LM
parallel to
meet the curve in R and the straight line Pi^ drawn through parallel to CL in M. Also draw RE
CL\ and
let
LM
parallel to CL.
Now
..
LP = PL';
or
PD - CF = FL', FP = CD = DL. And FP.PD = ER. RL. [Prop. 34] But ER = LC = 2CD = 2FP: PD = 2LR, LR = RM.
..
Proposition 63.
[III.
35, 36.]
// PL,
in L,
line
and if
PO
LQOQ'
be
drawn meeting
U, then
Ur LQ = QV
:
OQ.
105
Wc
asymptotes as in the
LQ = Q'L'
and
DL = IQ' = CF
Hence
since
{HD)
(>) = (i/C)
{CQ)
[Prop. 34]
Therefore
:
(EW) (DW)
(1).
Now
[Prop. 34]
Therefore, subtracting
CX
from both,
{BX) = {XH), and, adding (XU) to each, (EU) = (HQ). Hence, from (1), since (EW) = (CQ), (MG):(CQ) = (MH):(HQ), or LQ' LQ = Q'O OQ.
: :
(2)
on
omitted
for the
sake
is
Join
in R,
PQ
and produce
it
R.
Draw
PV parallel
in V.
106
LV=VL'.
/. =
(//.':
..
But
N(
QV=
VL'
VQ'.
QV:
= QP.PR' = PQ:QR
:
..
Proposition 64.
[III. 37, 38, 39, 40.]
he
he tangents to
line
// any
straight line he
[or the polar of the point F], this straight line is also divided
harmonically
i.e.
in the figures
drawn below
(1)
(2)
I'OLAllS.
107
Let TF be the diameter bisecting Qq in V. Draw as usual IIRFW, H'R'F'W, EF ordinate-wise to the diameter TF; and draw RU, R'U' parallel to QT meeting TF in U, U'.
(1)
We
have then
R'r:IR' = H'Q':HQ'
Also
RT
= AH'F'Q: AHFQ = H'TU'R' HTUR. [Props. 54, 55] TR' = R' U" R U' = AR'U'W: ARUW;
:
RT
..
TR' =
.
Rr:TR= AR'U']V'
ATHW
..
(2)
We
have
108
Also
..
MV: VR" = HQ':QH" = AHFQ AH'F'Q = HTUR H'TU'R. RV: VR" = HTUR + ARUW H'TU'R' AR'U'W = ATHW: ATH'W
:
= TW':TW'*
= RO':OR";
that
is.
RO
OR' =
RV
VR'.
A THIRD.
Proposition 65.
[III. 41.]
If the
ti&n,
or
Pr
rq
= rQ Qp = qp
:
pR
Let
is
therefore a diameter.
Draw T'TQW
parallel to it
through Q,
are
meeting Pq
in
and
qR
it
in
T.
Then
QW
is
also a diameter.
Draw
the ordinates to
from P, R,
viz.
110
^F passes
all
is
obvious,
and
be ratios of equality.
we
EBF
where
qV
Pr =
Then
(1)
rT,
Tp=pR, qF=FR.
2,
and, alternately,
rP.PT=EP:Pq=l: rP PE = TP Pq
:
= OP PV,
:
rP :Pq=OP: PR,
and
(2)
(1).
since
PU=2rQ,
RW = 2pQ
(3)
OR FR Rq=pR:RT',
Qp = PO:
(2).
and, alternately,
FR Rp^qR: RT = VR RO.
:
pR = PO OR
:
(3).
and
:
(3) that
Pr
rq
= vQ Qp =
:
qp
pR.
Ill
Proposition 66.
[III.
42.]
If
and any
Pr.P'r'
a diameter PP' of a r, r
= GD\
Draw
and DD'
;
PP'
t
and
let
the tangent at
Q meet
the diameters in T,
respectively.
CD
pass through Q,
the proposition
is
CD,
be equal.
If not,
we have
so that
..
whence
Hence, by
parallels, Gt
P'r'
= Pr QV
:
..
112
43.]
If
.
a tangent
to
a Jii/perbola,
LPL', meet
the asymptotes in
LCL
LC CU
is constant.
Draw PD,
PF
parcallel to
LP = PL';
CL = 2CD = 2PF, = 2CF=2PD. LG.CL' = ^DP.PF,
..
CL'
..
which
is
constant for
all
positions of P.
[Prop. 34]
Proposition 68.
[III. 44.]
If
to
the tangents at P,
;
to
respectively in L, L'
M, M',
LM',
L'M
PQ,
the chord
of contact.
We
so that
..
LM' L'M
arc parallel.
It follows that
or,
OL: LP=OM':M'Q;
..
lent of that
not spoken of by Apollonius under any equivaname, but they are determined as the two points
on the axis of a central conic (lying in the case of the ellipse between the vertices, and in the case of the hyperbola within
each branch, or on the axis produced) such that the rectangles
of
<
60'
" if
,
the
conic,"
\p.AA'
or
is
CB"^.
The shortened
<;
fiill
"
description of the
is
as follows
iirl
8e
} ,
:
iav
/ 77}<^
This determines
C.
.
114
of
we
is
falls
by a square.
Thus
in the figures
AA'
The
in the ellipse,
and the
drawn the rectangles AF, CB\ the base AS' falling short base A'S exceeding A'A in the
SF
is
equal to S'A' or
SA
respectively.
of a parabola is not
used or mentioned by
Apollonius.
Proposition 69.
[III. 45, 46.]
If Ar,
then
(1) (2)
r'rS',
A'r'
of the axis of a
r, r'
any point
in
respectively,
7'
the angles rr'S, A'r'S' are equal, as also are the angles
ArS.
(1)
rA.A'r' =^Cn'
= AS .SA', by
rA
definition,
:AS=SA'
A'r'.
115
Hence the
triangles
rAS,
SAY
zArS= zA'Sr';
..
a right angle.
And
(2)
rSV
;
is
a right angle,
rr'
as
rr'S
= =
rS'S, in the
S'r'A',
r'rS'
same segment,
triangles.
by similar
AiS.
In like manner
Proposition 70.
[III. If, in the
47.]
same
ficjures,
OP loill
at P.
Suppose that
OR is
to the tangent
We
shall
show that
S'r'A',
For
.*.
Or'R
82
116
Thereioie
r'R
S'r'
:
r'O
similar triangles,
..
(1).
Again,
[Prop. 13]
if
PN
A'T
r'R
-TA^A'N
:
Hence, from
(1),
= r'P Rr = r'P
Pr, by parallels.
Pr,
and therefore
coincides with P.
It follows that
OP
is
Proposition 71.
[III.
48.]
make equal
OPr
are right
r,
are concyclic
SPr =
In like
manner
S'Or',
and the angles SOr, S'Or' are equal, being the same or opposite
angles.
Therefore
SPr =
S'Pr'.
Proposition 72.
[III. 49,
50.]
(1)
to the
If,
from
either focus, as S,
SY be
drawn perpendicular
will be
AYA'
a right
A A'
as diameter.
The
line
to
to either
of the focal
distances of
be equal in length to
CA,
or CA'.
Il7
Draw
Let the
tions.
iSiF
VA'.
We
(1)
..
have then
the angles rAS,
Y,
rYS
are right
A,
r,
are concyclic,
and
ZAYS=ZArS
=
7''8A', since
rSi^' is
right
=
S, Y,
.".
1^'YA',
r',
it,
A
Therefore
lies
A'
SYr'
=a
right angle.
on the
circle
having
A A'
for diameter.
Draw GZ
parallel to
draw S'K
also parallel to
Now
whence
AS.SA' = AS\S'A',
and therefore CS
AS = S'A',
CS'.
Therefore, by parallels,
PZ=ZK.
since
Again
S'KP =
SP F,
parallel,
= ^S'PK;
..
[Prop. 71]
S'P = S'K.
And
..
PZ = ZK;
at right angles to the tangent, or
is
S'Z
is
But F'
on the
circle
..
having
A A'
for
diameter
GT = CA, or CA'.
And
similarly for
GY.
118
Proposition 73.
[III.
51, 52.]
In an
ellipse the
focal distances
of any point
equal
to the a.xis
if
A'.
We
parallel,
S'K = ST.
Then, since
SG =
GS',
..
= AA'.
[Prop. 72]
&c.
If PP'
point on
in
it,
he a diameter of a central conic, and Q any other and if PQ, P'Q respectively meet the tangents at P',
R,
R, then
PR.P'R = DD'\
120
Draw
the ordinate
:
QF to
Now
PP' = Q
V -.PV.P'V
PP')
.
[Prop,
= (QV:PV).(QV:P'V)
=
Hence
Therefore
(PR
PR
(PR
PF = PR .P'R
PP'\
PR =
PP'
= DD'\
Proposition 75.
[III.
.34,
.-)6.]
TQ, TQ'
point on
it,
beinij
tiuo
tangents
tu
conic,
and
)'
,
any
other
and
:
Qr
where
is
Q'r'
QQ"'
= (PV'
Q'r',
QR
in
.
then
:
{TQ TQ'
QV\
to QQ'.
of a tangent parallel
ETC.
121
Draw through
tively
;
the ordinate
R and moi-ting
it
since
PV bisects
QQ',
bisects
Then,
Now
QU
= QK':RK.KR' = QK':RK.RK'.
But
[Prop. 59]
QL
QL
Q'L'
QL' =
QK
Q'K'
QK\
Therefore,
.
ea:
aequali,
:
Q'L'
LP
PL' =
QK
Q'K'
RK
RK'
Qr.Q'r':QQ''.
Q'r'
QQ" = QL
Q'L'
.
LP
:
PL'
.
.
:
= {QL
122
THE t'OXKM
<JF
APOLLONIUS.
Proposition 76.
Llll.
:.x]
to
opposite branches
t
and meet
in
r,
t,
and
if tq
is
Qr.Q'r':QQ"=tQ.tQ':tq\
let
Let RM be the chord parallel to QQ' drawn through R, and Then QQ', RR', LL' are all bisected it meet tQ, tQ' in L, L'.
tv.
by
Now
But
tQ'
:
[Prop. 59]
Therefore, ex aequali,
tq'
:
tQ tQ'
.
Thus
Qr.Q'r':QQ"=tQ.tQ':tq\
the last two propositions give the
For, since the two tangents and
alone
may be
Qr
.
expressed thus,
Q'r
= (const.).
ETC.
123
Now
,
.
in the
accompanying
first
figure substi-
tuted for Q,
respectively in the
and we have
Q^r
Qy = (const.)
Draw
Rq^,
Rq.^
panillel
respectively
to
T,Q,,
T^Q^
and
meeting Q^Q^ in q^, q^. Also let Rv^ be drawn parallel to the diameter CT, and meeting QJ^^ in v,.
Then, by similar triangles,
Q,r:Rq; =
ClQ.--Q.q:,
Hence
But Rq^
.
Q,r
%
.
Q,q,.
%/
is
Rv^' =
..
T,Q,
T^Q,
:
, V\ by similar
triangles
Rq^ Rq^
jRy/
.
= (const.).
constant, as proved.
Also QiQ^
constant, and
Q{i'
Q^/"' is
It follows that
-flv,"
:
Qii,
Q/y/
(const.).
But Rv^ is the distance of R from Q,^.^, the chord of contact measured in a fixed direction (parallel to 0T^)\ and
R from the tangents measured in a fixed direction (parallel If the distances arc measured in any to the chord of contact).
Qj^,,
QjQ'j'
124
THE
CO.yiCS
OF Al'OLLOXlUS.
other fixed directions, they will be similarly related, and the constant value of the ratio will alone be changed.
is
it
if
meet three
by
tw^o of
The
at Q^, Q
If QiQjQgQ^ be meet at
an inscribed quadrilateral, and the tangents ,, the tangents at Q^, Q^ at jT^ and so on,
suppose Bq^, Rq^ drawn parallel to the tangents at Q^, Q^ respectively and meeting Q^Q^ in q^, q^ (in the same way as
Rq^
,
Rq^ were
drawn
let similar pairs of lines Rq^, Rq^' QxQi)' ^i^d drawn to meet Q,Q^ and Q^Q, respectively.
Also suppose Rv^ drawn parallel to the diameter GT^, meeting Q,Qj in
I'j,
and so
on.
Then we have
Q^^U
Qs^s
^'2
^V
k, are
Qsqs'Q.q:
= K-R<[
constants.
Hence we derive
Rv^'-Rv:
'Q.qrQs9:'Q.q/Q.q.
where k
is
some constant.
triangles Qtq^qt',
But the
SIS
Q^qsi
^^.c.
all
Hence
all
the ratios
constant;
Rv^.Rv,
^
^,
ETC.
12;
directions
But Bv^, Rv^, Rv^, Ri\ are straight lines drawn in fixed (parallel to CT,, etc.) to meet the sides of the
Hence
The beginning
of
Book IV.
of Apollonius'
work contains
proves
the
The method
it
of proof adopted
is
ad absurdum, and
thought
may, however, be observed that one of them [IV. 9] gives to a conic from an external
point.
R,
in
and RR'
it
both ways
meet the conic in P, P'. Then P, P' are the points of contact of the two tangents from T.
INTERSECTING CONICS.
Proposition 77.
[IV. 24.]
No
of them
If
way
is not.
common
to both,
q'Q'PQ of a conic be common Take Q' any other point on the conmion portion and Bisect QQ' in 1^ and draw the join QQ'.
possible, let a portion
to two,
and
let
them diverge
at Q.
diameter PV.
Draw
line
rqv(j'
parallel to QQ'.
Then the
will
through
parallel to
QQ'
we
shall
have in
one of them qv
= vq', and in
..
the other rv
is
= vq'
rv
= qv, which
impossible.
the
There follow a large number of propositions with regard to number of points in which two conies can meet or touch
all
much
space.
given
all
the
proofs
is
by which
established.
each
The
and " hyperbola " in the various enunciations do not (except when otherwise stated) include the double-branch hyperbola but only the single branch. The term " conic " must
terms
be understdod as including a
circle.
INTERSECTING CONICS.
127
Group
I.
opposite directions
will
[IV. 35.]
in
ADBEC be
A B,
BC.
Then AB,
BC
the same parts as the concavity of ABC. And for the same reason they contain an angle towards
the same parts as the concavity of
ADBEC.
contrary to
in the
same direction
vhich
is
the hypothesis.
If a conic meet one branch of a hyperbola in two and the concavities of the conic and the branch are in the same direction, the part of the conic produced beyond the chord of contact will not meet the opposite branch of the
2.
points,
hyperbola.
[IV. 36.]
of intersection will cut both
of
within the
opposite branch.
If a conic meet one branch of a hyperbola, it will not 3. meet the other branch in more points than two. [IV. 37.]
The
it
conic,
being
one-branch
curve,
must
have
its
meets
in
two points,
meet the
The
true
The same
is
first
branch.
its
[IV. 30.]
128
touches must be on
Avill
common
be
Whence
If
two points, the opposite branch of the first hyperbola [IV. 41.] will not meet the opposite branch of the second.
of concourse
It
will
fall
will
not
across
the
opposite
of
asymptotal
the
opposite
angle
and
will
not meet
\vill
either
branches.
Therefore neither
in
two
[IV. 42.]
D, E.
INTERSECTING CONICS.
the
129
For
it
same argument
as
in
the
last
proposition.
DE
will
hyperbola, and
therefore not
branches
DE.
and therefore
meet either of the branches opposite to the Hence those branches are separated by DE cannot meet one another which contradicts
:
the hypothesis.
Similarly, if the
two branches
DE touch, the
result will
be
if it also
point, then the opposite branch of the first hyperbola will not
[IV. 45.]
i.1,
being
th(
points of meeting
ith
the
first
branch and
9
H. C.
130
before, neither
AC
nor
BC
will
ACB.
A
it
in
points,
by hypothesis. Hence D will be within the angle formed by AC, produced and will not meet C or AB.
does,
8.
BC
hyperbola with
concavity in
first will
the
opposite
direction, the
The
figure
D and
and
to
it is
seen in a similar
hyperbolas are
branches,
of the
common tangent
the
first
Group
II.
Proposition 78.
No
1.
hyperbola
to
be
alone
excluded.
INTERSECTING CONICS.
If possible, let there be five points of intersection
131
, B,
C,
Join
if
AB,
DC and
let
produce them.
Then
T.
they meet,
them meet at
A,
Let 0, 0' be
conic,
taken on AB,
If
lines joining
conies.
DC such TE cuts
that
TD
it will
meet each
will
and the
the intersections to
be tangents to the
point except E.
Then
since
it
common
:
Therefore
and
where 00',
= ET:TP' =
:
ET
IP
EI:IPy
;
intersect at /.
ratios cannot hold simultaneously
But these
If
therefore the
AB,
DC
ellipses or circles.
Bisect
AB,
DC
at
M, M'
MM'
is
then
a diameter.
parallel to
AB
or
DC,
meeting
is
MM'
in
P, P'.
Then, since
MM'
a diameter of both,
NP = NE = NP',
which
is
impossible.
conies do not intersect in
Thus the
2.
points.
meet
[IV. 38.]
fact
that [Group
I.
3]
the conic
92
132
3.
hyperbola in
hyperbola
will
two points, the opposite branch of the not meet either branch of the second. [IV.
43.]
The
'
ApoUonius Group
is
I.
Let
HOH'
first
Now
of the second hyperbola, Avill not meet and therefore AB separates the latter from the asymptote OK'. Similarly DC separates the former branch
the
from OK.
4.
branch of the
[IV. 44.]
The proof
fifth
is
If
point and
determined as before,
separate
points,
ET
secting branches
in
whence
the
harmonic
[IV. 46.]
IXTEIISECTING CONICS.
133
in
Let the
tirst
B, C\ and (if
AB,
DE
through
C
If
by the diameter;
which
(6)
is
impossible.
AB,
Bisect
AB,
DE be not parallel, let them meet in 0. DE in M, M', and draw the diameters MP,
in the respective hyperbolas.
MP'
',
will
be parallel to
^0,and
the tangents at
Q parallel
BO.
L^t the tangents at P,
Q and
P', Q'
meet
in T, T'.
Let
R,
R',
CRR' be
parallel io
AO
in
and
DO
in 0'.
Then
= T'P" T'q\
:
[Prop.
5i)]
It follows that
whence
which
is
impossible.
Therefore, etc.
The two branches of a hyperbola do not meet the 6. two branches of another hyperbola in more points than four.
[IV. 55.]
134
hyperbola and
Then
if
(a) if
in
meet B' at
point, for
in tAvo points and B' in one point, A' cannot meet in one [Group I. 5], and it can only meet in two points A could not have met B' if A' met
all
(which
(c)
it
does)
A meet in two points and A' meet B, A' Avill not in more points than meet B' [Group I. ], and A' cannot meet two [Group I. 3]
if
{d)
if
meet
in
will
or B' in
I.
6]
the branches A,
direction,
and
cut
in four points,
;
A'
A'
in three points,
will
And
Proposition 79.
Two
1.
touch at one
The
proposition
is
[IV. 20.]
The proof
follows the
method of
Pr(^i).
78 (1) above.
INTERSECTING
2.
CONICVS.
135
in
and meet the other branch of the second hyperbola two points, the opposite branch of the first will not meet either branch of the second. [IV. 47.]
in one point
The
is
If one
second in one point and cut the same branch in two other
points, the opposite
78
(4).
Avill
The proof
5.
follows the
method of Prop. 78
If one branch
of a
its
will
The proof
case (4).
6.
follows the
method of Prop. 78
If a hyperbola with
bola vith two branches in one point, the hyperbolas will not
meet
in
[IV. 56.]
The
methods em-
ployed in Group
8.
in
Proposition 80.
No
at
1.
tiuo iJoints
can intersect
The
proposition
is
true of
all
136
,
is
B.
Then,
if possible, let
them
also cut at G.
(a)
and
does not
lie
between
property
(6)
if
is
proved by
the chord
G by
a diameter
is
between
G, for
and
B, draw
TVirom
middle point of
absurd.
Then
to
conies,
And
an absurdity.
any other
point,
[IV. 40.]
I, 4.
If one
not meet
[IV. 51.]
AB
AG,
BE
in
A, B.
Draw
the tangents at ^,
T,
G.
If possible, let
GD,
Join GT.
therefore
Then
falls
is
AB, and
GT
ATB.
BT
GT
falls
on the side
GT
passes
through
it foils
A TB
which
is
impossible, since
ATB.
INTEUSECTING CONICS.
4.
137
If one branch
of i)ue
if
[IV. 52.]
This
is
in
two
do not
intersect.
[IV. 53.]
This
6.
If a hyperbola with
bola with two branches in two points, the hyperbolas will not
point.
[IV. 57.]
The
Group
conies,
1.
V.
parabola cannot
[IV. 30.]
more
TP =
V.
parabola,
if it fall
CV:CP = CP:CT
= GV-CP:CP-CT = PV:PT.
Therefore
PV>PT.
parabola
And
falls
3.
for the
P'V=P'T:
is
impossible.
ellipse or circle.
32]
The
proof
is
1:38
same
centre.
[IV. 33.]
Proved by means
5.
oiGV.CT= CP\
circle
If an ellip.se have double contact Avith an ellipse or a having the same centre, the chord of contact will pass through the centre. [IV. 34.]
Let
be
the
(if
A,
meet
in T,
and
let
V
If
Then
TV
is
a diameter.
possible, let
be the centre.
GT=CF\
not meet,
which
i.e.
is
absurd.
5 do
they are
parallel.
AB
'\&
[V. 1, 2, 3.]
If in an
to the aa;is
meet the
a hyperbola he d7'awn perpendicular and equal to one-half its parameter, and if CM ordinate on the curve in H, then of any point
ellipse or
AM
A A'
PN
PN' =
2 (quadrilateral
).
Let
to
AL
be twice
Join
AM,
i.e.
let
AL
parameter.
A'L meeting
PN in
R.
Then A'L
is parallel
CM.
Therefore
Now
[Props.
2,
3]
..
=
In the particular
ca.sc
2 (quadrilateral
is
).
C and A'
in the
.,.
,
where
between
fuKJvz
'
140
ellipse,
triangles,
and
P'N" =
Also,
if
CA
CN'H
).
be the end of the minor axis of the quadrilateral becomes the triangle CAM, and
ellipse,
the
BC'^2ACAM.
[The two
pruptisitions.
l;ist
by Apollonius
in separate
Cf. the
p. 40.]
Proposition 82.
[V. 4.]
7/i
a pardbola, if
to the
he
AE is
the
side.
e(jual to
minimum
he
strairjht line
from
curve,
AE
and, if
PE
increases as
on either
PE'=AE' + AN-\
Let
AL
Then
PE'=2AE.AN+EN'
=^2AE.AN + (AE'- ANf =^AE'+AN\
141
as
AN,
AE,
i.e.
moves
the
minimum
proposition,
value of
PE
is
or
AE
A
is
to the curve. in
[In
this
as
the succeeding
is
is
propositions,
between
and E,
where
coincides with
and
PE
is
therefore perpen-
AN
greater than
AE, and
he proves the result separately for each. The three cases will for the sake of brevity be compressed, where possible, into one.]
Proposition 83.
[V.
5,
G.]
a point on the axis of a hyperbola or an ellipse such is the least to half the latus rectum, then to the curve; of all the straight lines which can he draimi from moves increases as and, if he any other point on it,
If
he
that
AE is
equal
AE
PE
further
from
on either
.
side,
and
'
^4^^ [= ^E" + AA
to the
e'
^N']
hyperhola)*.
straight line
Also in the
to the curve.
A'
is the
maximum
from
Let
AL
let^X be
bisected at if, so
let
that^iT/=
J.^".
Let
PN (produced if
necessary) meet
CM
in
EM
and
EN = NK.
side of the
equation
base .-'' similar to that contained by the axis (as base) and the sum (or difference) of the axis and its parameter. A similar remark applies to the similar expression
142
Now
and
..
PN^ =
\\'=2;
2 ((luadrilateral
),
[Prop. 81]
'
(1)
NORMALS AS MAXIMA
MINIMA.
143
Proposition 84.
[V. 7.]
If any
point
that
AO
<
hpa, then
OA
is
the
is
minimum
straight line
from
to the cin-ve,
and
OP
(if
increases as
Let AEhQ set off along the axis equal ami join PE, PO, PA.
Then
so that
and a fortiori
>\ >,
PE > AE,
PO>AO.
P'E > PE.
OPP' >
..
so that
..
ZEPP'>ZEP'P;
and a fortioH
OFP.
OP'>OP,
and so
on.
Proposition 85.
[V. 8.]
I7i
a imrahola, if
inid if
he
a point on
the
AG>\pa,
be taken between
and G such
NG
then, if
2'
to the axis
NP is dravm perpendicidar
is the
in P,
the
PG
minimum
straight line
from G
carve [or
normal at P].
144
P'G increases
as P'
mon
furthei' from
in either direction.
Also
P'G'
= PG'-\-NN'\
Wchave
Also
Therefore, adding,
P'G' =
Thus
it is
clear that
PG
is
the
minimum
as P'
normal at P].
And P'G
increases with
NN',
i.e.
/* in either direction.
Proposition 86.
[V. 9, 10, 11.]
///
a hyperbola or
an
ellipse,
if
G be any point
on
A' (within
AG>^,
so that
and if
GN
be
measured towards
CA'],
145
through
line
minimum
nonnal
at
straight
;
from G
be
PG is the PG is the
P'G
P]
ai^d,
if
P'
the curve,
side.
Also
P'G'
- PG' =
NN"
"^-4^4^
[=e\NN'
where P'N'
is the
146
Also P'G'
P'G'
=
Thus
it
HP
PG
CA AM ' GA
is
NN'
straight line from as
follows that
the
minimum
to
P'G
will
increases with
NN'
P' moves
further from
in either direction.
In the
ellipse
GA'
is
be the
maanmum
to the curve, as
easily
Cor.
centre, the
two
minimum
manner
to
CP
increases continually as
moves from
to
A.
Proposition 87.
[V.
12.]
If G
be
a point on
the axis
to the
of a conic and
is the
GP
be the mini-
mum
if
straight line be
any point
to
from G on PG,
P\ and
then
OP
minimum
straight line
from from
to the cui^je,
[or to A'].
Since
FG > PG,
zGPP'>zGPP.
NORMALS AS MAXIMA
Therefore, a fortimn,
ANT) MINIMA.
147
OP ' >
or
OF ,
[&c. as in Prop. 84].
for the three
by red actio ad crbsurdum, the convei-se of the propositions 85 and 86 just given. It is also proved that the normal makes with the axis towards the nearer vertex an acute angle.]
Proposition 88.
[Y. 16, 17, 18.]
If E'
be
a point on
to
the
minor axis of an
ellipse at
a distance
then
from
is the
equal
BE'
to the
GA'
or
E'B
he
maximum
side.
curve ; and, if
any
other point on
E'P
diminishes as
on either
Also
E'B'-E'P^Bn'.''
-^
[='.
'^^]
ApoUonius proves
^<BB',
the
(2) whei-e
^=BB',
is
^>BB'.
The method
first
of proof
is
given here.
*'"^"""'~?<"'
"vT"
;,'>
-----y<
102
148
By
Prop. 81 (which
is
we
have,
if
Bm =^ = BE',
Also
.'.
E'm
in h,
k respectively,
"=21'.
PE'^=2AmBE'-2Amhk.
BE'*=1AviBE'.
..
But
BE"-PE"=^2Amhk
= mi
^
.
(hi
ki) = mi
(hi
mi)
mB-CB
folloAVS.
Proposition 89.
[V. 19.]
If BE'
to
be
measured along
or
the
minor axis of an
ellipse equal
CA'^~\
BO > BE',
to the
OB
is
maximum
otJier
from
curve; and, if
be
any
point
on
it,
OP
diminishes continually as
from
to B'.
The proof
follows the
method
149
Proposition 90.
[V. 20, 21, 22.]
If g
he
a point on
the
minor axis of an
or
and
if
Gn
he
measured
to-
luards
so that
Cn:ng = BB':p^[=CB':CA'l
then the perpendicular through
to
BB'
two points
such that
Pg
is the
maximum
straight line
from
to the curve.
on either side
to
,,
or B',
and
Pg'
rg -nn
CA'-CBn
^^,
150
Now
Pn^=
ng^
..
2 (quadrilateral
mBnh),
2A}ing:
Pg'=2(mBnJi + Ahng).
P'g'
Similarly
2 {mBn'h'
hi'g).
By
subtraction,
Pg^-Py=2/S},h'l
hi (h'i
.
ki) hi)
= hi.{h'i
hi'
[Em - BC\
V
BG
-nn
whence
it
^^
the
follows that
Pg
is
maximum
to the curve,
the
area described.
Cor.
1.
It follows
be any point on
Pg
produced
beyond the minor axis, PO is the mammum, straight line that can be drawn from to the same part of the ellipse in which Pg is a maximum, i.e. to the semi-ellipse BPB', and if OF be
semi-ellipse,
OP' diminishes
as
centre C, the
maximum
C
Pg
to the ellipse
if is
is
perpendicular to BB',
centre, the angle
viz.
CA
or GA'.
Also,
if
g be not the
a
PgB must
[(jr
be acute
maximum
and,
if
Pg
is
maximum
a normal],
(hi: ng
= GB': GA\
[This corollary
is
151
Proposition 91.
[V.
2.S.]
If g
in G,
he on
tlie
minor axis of an
to the curve,
ellipse,
and
gP is
a nicucimum
straight line
from g
and if
gP
GP is a minimum
to the cin've.
the
maximum
We
have
Cn
ng
= BB'
CB''
pb
[Prop. 90]
[=
CA']
= p: A A'.
Also
Gn
ng
= =
PN
NG
ng
= NG
CN.
..
NG'.CN=pa:AA',
minimum
straight line
or
PG
is
from G.
[Prop. 86]
Proposition 92.
[V. 24, 25, 26.]
conic,
in which
it
meets
A A', or
meets
minor
axis.
152
This
that
is
THE
(JOXICS
OF APOLLONIUS.
at once proved
PG,
lines
.straight
by reductio ad ahsiirdum on assuming A A' in G, H) are minimum to the curve, and on a similar and
assumption
for
Proposition 93.
[V. 27, 28, 29, 30.]
conic,
whetJter
regarded
minimum
or as
A A'
from its intei'section with the axis a maximum from its intersection with BE (in the
straight line
'
case of
an
ellipse), is
perpendicular
to the
tangent at P.
axis
meet the axis of the parabola, or the Let the tangent at hyperbola or an ellipse, in T. Then we have to A A'
prove that
TPG
is
a right angle.
(1)
= ^,
..
and
NG = ^',
NG
pa
AN
NT,
so
that
TN.NG=pa.AN
= PN\
a right angk
And
the angle at
..
is
TPG
is
a right angle.
158
ellipse
PN':CN.NT
=
.'
:
[Prop.
U]
= NG CN,
= TN.NG:CN.NT.
..
PN^ = TN.NG,
..
is
right
TPG
a right angle.
If Pg be the maximum straight line from g on the (3) A' in G, PG is minor axis of an ellipse, and if Pg meet a minimum from G, and the result follows as in (2).
all
three
If
GP
is
let
GK be
perpendicular to
it.
Then
GP >
is
GK.
Hence a
GP > GQ,
and
GK
GP
is
minimum.
Therefore &c.]
Proposition 94.
[V. 31, 33, 34.]
(1)
maadmum
conic
and
OP
be
to the conic,
straight line
conic at P.
PT
drawn
at right angles to
PO
154
(2)
any point on OP produced outside the conic, drawn from 0' to meet the conic in one point but not produced so as to meet it in a second point, O'P vnll be tlie minimum; and of the rest that which is nearer to it will be less than that which is more remote.
then,
of
(1)
First, let
OP
be a maocimum.
Then,
if
does not
touch the conic, let it cut it again at Q, and draw PQ in and the curve in R.
OK to
meet
OPK is
a
right,
OPK >
OR > OP
:
OKP.
which
is
OK > OP,
OP OP
is
and a fortiori
impossible, since
maximum.
Therefore
TP
Secondly, let
TP
cut the
curve again in Q.
From any
point between
draw a straight
line to
and draw
ORK perpendicular
line
\h
meeting
it
at
in R.
OKP
:
a right angle.
is
Therefore
which
impossible, since
NORMALS AS MAXIMA
(2)
MINIMA.
155
tangent at P, as
Let 0' be any point on OP produced. Dmw the PK, which is therefore at right angles to OP. Then draw O'Q, O'R to meet the curve in one point only, and in K. let O'Q meet
PK
is
O'P.
>
O'P, and
O'P
the angle
RP, RQ. Then the angle O'QR is obtuse, and therefore O'RQ is acute. Therefore O'R > O'Q, and so on.
Proposition 95.
[V. 35, 86, 37, 38, 39, 40.]
(1)
axis
'
as
meet the lucis of a parabola or the the normal at of a hyperbola or ellipse in G,the angle PGA increases or G moves further and further from A, but in the
If
PGA
will
ahuays be
less
same
AA'
as
Two normals at points on tJie same quadrant of an AB, will meet at a point luithin the angle ACB'.
Suppose P'
are
is
(1)
Then,
since
minimum
.56
THE
(a)
(JOXJCS
OF APOLLONIUS.
and
'> .
''>\
17
[Prup.
2.S]
AL AM= CA AL
:
:.
(2)
It follows at once
on one side of
(3)
lines
A A'
from (1) that two normals at points \\\ meet on the other side of A A'.
from
Regard the two normals as the maximum straight g, (/', the points where they meet the minor axis of
the
ellipse.
Then
On
n'g'
BE
:
:
pi,
[Prop. 90]
= Cn ng
..
On'
Cg = On
..
Gg.
Og,
But
whence
toAvards
it
On >0n,
Og >
must
cross at a point
lies
before
Therefore
A
(2) above,
lies
And, by
the
'.
below
AG;
therefore
lies
within
Proposition 96.
[V. 41, 42, 43.]
(1)
In a parabola or an
ellipse
any normal
PG
will meet
In the hyperbola (a), if AA' he not greater than pa, no (2) normal can meet the curve in a second point on the same branch but (b), if AA'>pa, some normals luill meet the same branch again and others not.
(1)
For the
and
will
PG
viz.
meets a diameter (the axis), it that through the point of to PG, i.e. the diameter bisecting
will
18
(2)
(a)
CL
be
the
.
asymptotes, and
to
let
the
tangent at
meet them
FO
~.
Let
Then, by hypothesis,
CA
:
-^
AM,
:
and
CA
AM = CA'
CA
If^AL;
[Prop. 28]
..
CLA
is
ACL
or
ACL'.
[Prop. 95]
But
..
CZ^ >
PGiV
/.ACL'>ZFGN.
ACL'
together with the angle
adjacent to
FON will
FO
Therefore
will
will not
meet
CL
(b)
Suppose
C^ > ^il/
or
Then
.AC.
LA .AM>LA
Take a point
on
AC,
159
in
to
P, and
PG
is
then the
minimum from G
to the curve,
and
NG .CN=pa:AA'
= CA
Hence
..
AL, from
above.
^ACL'=Z ACL =
PG, CL'
are parallel
PGN;
at points
between
and
make with
the
PGN, and
normals at points
PGN.
will not
and
meet
;
but
normals bevond
again.
160
Proposition 97.
^
in
and if
he joined to
any
othei'
point
on
conic
(ill
that
further supposed in the case of the ellipse three lines OP^, OP^, OP cut the same half of the aads),
(it
heincf
then
y 1 )
OP
if
cannot he a normal
to the
curve
(2)
at
OP
PG
he
the
normal
P,
and
I.
is
does not
P^ and P^.
parabola.
P,-A^,,
^,
161
Draw
OM
and measure
MH
Then
and
Therefore
MH = A\G
MH
4-
Therefore
Similarly
Therefore
. \\
:
by similar
triangles.
,^.) ^^.]
:
^'
= P^N^ P^N^
whence
N^h\
HN^ =
\^
TN^
and
If
TN^ = HNJ
be a variable point and
^"
PN
now
three cases
or
HiY^
(1),
or ^^Y,, but
or
< TN^
or
HN^
(2),
HN^
(3).
(1), (2),
we have
TN^:TN>HN:HN^_
<HN'.HN.^
(1), (2),
(3).
>HN:HN^
* It will be obser\-ed that there are three sets of points P,
N, K, in the
figure denoted
by the same
;
letters.
This
is
mind that attention must be confined to one at a time as indicated in the course of the proof.
different cases
it is
and
H. c.
11
162
If
NP meet P^P^ in F, we
P^N,
:
FX>HN
HN,
in (2)
(1)
and
(8),
(2).
<HN.HN,
But
in (1)
and
(3)
FN < PN
(1)
Therefore, a fortiori in
the cases,
:
PN>HN
HN,
and
(3),
(2).
<HN:HN,
(l)and
Thus
P^N^.N,H>PN.NH
(3), (2).
<PN.NH
Hence
Therefore
and
<PN.NH
MO .PN>NH:HM
.,.., ^^^\,hy(A)aho.e.
(S))
(1)
and
(3),
<NH:HM
and
(2)
MO'.PN = MK:NK.
Therefore
-.
MK NK> NH HM <NH:HM we obtain MN NK > MN whence <MN:HM HM or NG > NK in (1) and 80 that and < NK in
: :
(1
and
(3), (2),
(1)
and
(3),
(2),
(3),
(2).
is
proved.
II.
P.^
meet
at 0,
in
and draw
OM perpenfor
the axis.
Divide
CM
:
(internally
the
CH
and
to
let
HM = AA'
pa [or CA'
CB'],
OM
be similarly divided at L.
parallel to
Draw
HVR
parallel
OM and
LVE, Oi?P
CM.
1G.3
EL
in
T,
and
let
FiN^,
P.N. meet
it
in
U..
the axes in K,
on the curve.
Join
OP
meeting
meet P^P.
in
Q and EL in U. 112
. ;
1G4
Now
the ellipse,
OiY.
N,G, =
'
.p
= GH: HM.
CM:GH=GG.:GN^ = GG,~GM:GN,-GH
= MG,'.HN,
= MG,: VU,
Next
so that
(A).
Thus
FG:N,U,==GG,
N,G,
Again
FG.
GM EG. GH = {FG
:
(B),
and
FG GM = Fg, MG,
.
.
'.
,
Fg,:
GM = FG
il/(?o
..
or
manner
..
L\V: U,V=PM,:P,U,
whence
and
= TU., TUi, by U,U, U,V= U,U,: TU, .:TU,= VU,l TU,= VUj
: :
similar triangles,
.^.
^
^'
AN < AN'^;
t/^,F
hence
^: 7 > i7F
PJ\^.
UU,:TU>UU,:U,V;
:
2\U^.QU>UV: UJ,
by similar
triangles.
Therefore
a fortiori
UJ>QU.
UV,
>PU.UV,
165
\ .1\= CE
V, iVuin above,
.
= LO.OR,
..
CE.LO = uR.EV;
<
xiN^.
LO.OR>FU.UV.
(2) that
Suppose
yliY>^iY,
but
Then
..
U^U:Tl\>l\U: UV,
whence
by similar
triangles.
Therefore
{a fortiori)
PU UV >P^U^.UJ
.
>LO.OR
Lastly (3)
let
AN be
> AN,.
Then
..
whence
or
and afortioH
..
>PU
UV\
LO.uR>PU.UV,
and (3)
as in (1) above.
Thus we have
and
for (2)
LO.OR>PU.UV, LO.OR<PU. UV
to refer
to
That is, we shall have, supposing the upper symbol (1) and (3) and the lower to (2),
LO-.PU^ UV:OR,
i.e.
LS-.SU^ UV:LV;
..
LU: US^LU-.LV,
and
LV^US.
166
FO'.LV^FO:
or
SU,
or
Fk PU,
:
CM:MH^Fk:PU',
..
FC:
CE^Fk:PU
^FkTFC:PU+CE
J Ck J CK
:
i\r/f
FE
or
i.e.
:
jB:6'
CiV^
iVZ,
CN NK^FE:
=
EC,
2^^':j).
But
(7i\r:i\r(;
yl^':^^;
..
NK^NG;
i.e,
when
lies
is not between P^ and P^ NG, and when between P, and P.^, NK>NG, whence the proposition
NK<
follows.
Cor. 1. In the particular case of a quadrant of an ellipse where P, coincides \vith B, i.e. Avhere coincides with g^, it follows that no other normal besides P,f/i, Bg^ can be drawn through g^ to the quadrant, and, if be a point between A and
P, while Pg^ meets the axis in K,
,
NG > NK.
NK.
But
if
is
lie
[This
in Prop.
95
2.
C(JU.
ellipse
normals at
i^oints
on one quadrant of an
167
Four
7ioi'mals at points
or (2)
meet in one must (1) separate one normal from the three must separate two from the other two, or (3)
must
lie
on one of them.
In cases (1) and (3) a contradiction of the preceding proposition is involved, and in case (2) a contradiction of
Prop. 90 (3) which requires two points of intersection, one on
axis.
Proposition 98.
[V. 49, 50.]
In amj
conic, if
be
any point on
axis
AM
is
and if
be
any point on
the
drawn
to
to
opposite
and
can
be a normal.
Let
let
OP
PN be the ordinate at P.
We
have in the parabola, since
AM'i^^
NM<^,
Therefore, a fortiori,
i.e.<NG.
< NG.
ellipse
AA' jh
:
is
not greater
and
..
CN:NM>CA :AM;
CN NM > AA' pa >CN:NG;
: :
..
NM<NG,
and a fortiori
Therefore
\: < NG.
OP
is
not a normal.
Proposition 99.
[V. 51, 52.]
If
the
AM measured
{but in
such that
(a)
if
;
OM > y,
hut,
which
if
OP
;
any straight
K,
ordinate
and
PG
the
normal at
(b)
0, and, if
if OM=y, only one normal can he so drawn through OP he any other straight line drawn to the curve and
< NG,
as before
0,
is
if
he
OM < y,
and, if
less or
OP
two normals can be so draiun through any other straight line drawn to the curve,
NK
greater than
NG
according as
iiornials.
OP
is not,
or
is,
inter-
I.
is
a parabola.
Measure
at
.V, so
tlmt //.V,
= 2.Y,.4
PR()P(JSITI()NS
160
where P^N^
(a)
is
iV,.
Suppose
OM > y.
Then
..
N^H.HM\ OM:P^N^>N^H:HM,
y.P^N^ =
or
HM
HM,
henee
so that
il/iV,
iY,/i^,
iVjii,
i.e.
iV^A<f
Therefore OP^
is
N^K^ <N^G^.
in
Next let be any other point. Join OP meeting the axis K, and let the ordinate PN meet the tangent at P, in Q.
70
Then,
since
\=2\ = \,
if
^iV< AN^
avc
have,
If
AN>AN^,
\>;
NJ>NH;
.'.
N^N:NH>N^N:NJ,
HN^
whence
thus
or
P^N^:
TN^:TN>HN.HN^, QN>HN:HN^,
\HN>TN:
>
TN^
P,N^
QN
and a fortiori
or
>PN:P^N^,
a fortiori
.'.
P^N^.NJI>PN.NH',
But
.
P^N^.N^H>PN.NH.
.
or
i.c.
by hypothesis
MK:KN>NH:HM,
triangles.
by similar
Therefore, componendo,
whence
Therefore
(b)
MN NK > MN NK < HM ov ^
:
HM,
OP
is
< NG.
Suppose
OM = y, and
N^K^ = HM=
N.G.
^
and P,0
If
is
is
a normal.
P,N^.N^H>PN.NH,
and PjiYj N^JI
.
is
Therefore
and
it
OP
is
NK < NG.
(c)
Lastly,
if
J/ <
7/,
or
OM:P^N^<N^H:HM, OM.MH<P^N^.N,H.
Let N^li be measured along iV,P, so that
OM.MH=RN,.N,H
171
Thus
Let
to the axis,
and with
AH,
R.
HL
as
passing
through
of the hyperbola,
..
above
or
and, componendo,
MK KN = NH HM, MN NK = MN HM
: :
..
NK=HM=^^ = NCr,
normal.
in 0,
and
PO
is
normal.
Similarly
P'O
is
and the
rest of
OM=y,
is
is
the
the centre of
y be the coordinates of
0,
so
that
AM=x,
and
if
4a=^j,
HM=2a,
N^H = l{x-1a\
Also
or
af = AN,.N^H' = ^\{x-2a)\
i/'
or
27(/
is
(.r
2(0',
which
172
II.
THE
CUXIC.S
OF APOLLONIUS.
lui
ELLIPSE.
:
Wo
AM >
-^^
so that
CA AxM< AA'
:
jh-
173
CH*, and
Take two mean proportionals OiV,, CI between let P^N^ be the ordinate through iV,.
Take a point
CA
and
on
OM
(in
the hyperbola) or on
OM
Draw
both
ellipse)
such that
OL
LM = AA'
and CE,
pa.
LVE, OR
in
axis,
HVR
Join
P^N^ meet
EL
in
U^.
0P
P^N^ = (CM
first
..
MH)
{HN^
iV.C).
(a)
Suppose
that
OM > y
OM'.P^N^>y:P^N^.
.
But
OM
and
y
:
P,i\^,
= (Oil/ ML)
:
(il/X
P^N^)
= {OM:ML).{N^U^:P^N^),
PjiY,
= (Ci
il/^)
{\
i\^,C')
= (Oi:i/X).(iyiV^j:i\'',C);
..
N^U^:P,N^>HA\:Nfi
P^N^.N^H<CN^.N^U^.
(1),
or
Adding
P^L\.U,V<CH.HV
<LO.OR,
But, for a normal at Pj,
Prop. 97]
.
CH HM=OL LM.
:
the proof of
P^U,.UJ=LO.OR.
Therefore
Prop. 97]
P^O
is
of finding
duction.
174
Next let be any other point than P^, and let U, N, have the same relation to that U^,N^, K^ have to P,.
Also, since
U^N^
,,> HN^
i\\P^
N<0 by
:
be taken on
i/',iV,
such that
^\:
and draw wuu^v
.
parallel to
(2),
so that
CA = CA
CI,
:
CT
But CX^
and
is
a third proportional to
CH,
CA\ :CA = CI
..
CN^ =
:
CH
CT
CI;
CH
CN^ = CN^
And
since
= CH-^ CN^ 6'iY, ~ CT = HN^ N^T. CH CN^ = P^u^ P^iV^, n^N^ i\'',P, = HN^ iV^C, from (2) above
:
: :
\\
:
P^u^
P^N^
thus
<^,
wu <
u^v,
AN>AN^,
imi^^> uv
;
and
WjW
uw >
:
u^u
u^v,
:
.".
uu^
uv
uu^
wu^,
whence u^w
.'.
P,,
uw > Qu >
nv
uv
n^v.
whence
w?i,
:
(where
thus
,
:
u^v
in Q);
riv
vu
wiL
Qi('
wu^
Qi<
".
.
> Pu
But, since
thus P,u^.u^v>Qu.uv
.
uv,
afortioH.
> Pu
uv,
a fortiori,
and the proof proceeds as in the first column, leading to the same result,
rectangle
i<,iV,
iV, ff,
PU.UV<LO.OR,
\'01.
175
= CH.Hv; CH.Hv>Pu.uv,
uU. UV,
the hyperbola,
rectangle
PU.UV<CH.Hv + uU.UV
for
or
PU.UV<CH.Hv-uU.UV
for the ellipse,
..
in either case,
a fortiori,
or
PU.UV<CH.HV, PU.UVkLO.OR
Therefore, as in the proof of Prop. 97,
PO
is
not a normal,
hx\tNK<NCT.
(b)
Next suppose
OM = y,
=y
P,^\,
and
^^:^^ =
..
:\0:
Ave
CN^.N^U^ = P^A\.N^H.
U^N^.N^H,
have
Adding
or subtracting
P^U^.
U,V=CH.HV=LO.OR,
normal at P,.
0.
and
be any other point on the curve, it will be shown as before that U^W= U^V, because in this case the lines WV, wv
coincide
;
also
:
UU,
and
UW>
UV>
UU,
U,V
in the case
where
where
UW<
U^V,
UU^
UU, -.U^W
in the case
afortion,
PU UV<LO.OR
.
PO
is
NK < -tYG.
176
(c)
OM < y,
wc
shall
177
and we
Let
shall derive
LO.uR<PJJ^.UJ.
S be taken on 1\\ such that LO .OR = SU^. through ' describe a hyperbola whose asymptotes are
i7,
V,
and
in
VW and
F// produced. This hyperbola will therefore meet the conic two points P, P', and by the property of the hyperbola
PO, P'O
The
of
curvature at P.
To
= CM,
CHa^
Also
GH
_
I
(1).
_j^_C^HN^
GN ^
^
and
where the upper sign
N"^ +^'';
=1
(3),
And,
lastly,
GN,
From
and
,
(4)
(4).
a.GH
^
;
..
GA\'
= a\CH
(.5).
from
(2),
P^N^
=
H. C.
ft*
^.^
+
iV
'
.by(3).
1
178
whence
But, from
b\l.,
^ ^^,)
.
(6).
CH = a
ax
,
= -^^yt
a'
.
whence
;'
[-^^^
(3),
(7).
(7),
(
by the aid of
by
ax
\i
a*6V
{ax)i
WbV
+ {byf
y^
'
{a'
6^)1]
Proposition lOO.
[V. 53, 54.]
If
(a)
the
be
ellipse,
then
if
OB BG <^ A A'
BA, BA'
pa,
and
be
quadrants
major axis in K,
PO
(6)
NK < NG
normal only besides OB can be drawn to either of the tivo quadrants as OP, and, if P' be any other point, N'K' is less or greater than N'G' according as P' is further from, or nearer to, the minor axis than P.
if
OB BC < A A'
:
pa, one
separately thus.]
jiC
:
We
have
..
OB BC < On:
: :
On
nC, or
CN:NK> AA'
:
pa,
whence
and
179
0'B:BC<AA':pa.
Take a point
on O'B such that
O'n
:
nC = AA'
-.pa-
CONSTRUCTION OF NORMALS.
Proposition lOl.
[V. 5o, 5G, 57.]
If
is
any point
{where
beloiv
the axis
A A'
of an
ellipse,
and
AM > AC
tlirough
is the
can
and
C, but never
Produce
OM to
p^,
IH
parallel
CONSTRUCTION OF NORMALS.
This
will
181
For, drawing
in I),
meet the
ellipse in
AD,
IL produced
we have
AH:HM>CH:HM
> AA'
> OL
..
:
j)a
LM:
or
AH.LM>OL.HM, AD.DI>UL.LL
it
between
point P,.
Produce OP^ both ways to meet the asymptotes and draw R'E perpendicular to the axis.
Therefore
in
R, R',
OR=P^R', and
AA'
:pa
consequently
EN^ =
iMH.
Now
OL:LM
:
= ME
Also
..
EK^, by
similar triangles.
AA':pa =
AA'
since
EN^ = MH.
N^K^ = N^G^, and P^O
is
Therefore
a normal.
Let
OP
meets
AC in
K.
Produce
axis in K'.
BC
to
meet OP,
in F,
[at P,,
B] meet
in F,
FP
is
not
Therefore,
fortiori,
NK > NG.
1.]
And,
if
is
between
and
< NG.
182
Proposition 102.
[V. 58, 59, GO, 61.]
If
extremity
be
is
any point outside a conic, but not on the ct-xis iuhose A, we can draw a normal to the curve through 0.
MH
equal to
to
draw
side as 0, and,
a point P, and,
OP
the axis in
and
HR
in R,
we have
= NK.
Therefore
NK^P^.
and
PK
In
is
the
on
CM
or on
CM
])(1(.(1,
and L on
67/
OM or OM produced, so :HM=OL:LM=AA'
that
Pa-
CONSTRUCTION OF NORMALS.
183
Then draw
through
HIR
and
ILW
184
OP
will
be a normal.
:
For we have
since
(1)
HN = MK
=
= MO
:
LR',
OR = PR',
and therefore IL
UR'.
Therefore
MK
.
=
:
MC
:
CH,
CH HM = OL
LM.
Therefore, alternately,
MK:MC=NH:HG
In case (2)
or
[so that 0,
(A).
OL
LM = CH HM,
:
OL.LI=GH.HI,
C
are on opposite branches of the
same rectangular
hyperbola].
Therefore
or,
PU
OL = LI lU,
:
by similar triangles,
UR'.R'L^LI-.IU,
whence
..
and
before (A).
GK GM=GN:GH,
:
GN.GK = GH:GM,
so that
GN:NK=GH: HM
= AA':pa\
..
NK = NG,
and
oy
is
the normal at P.
CONSTRUCTION OF NORMALS.
(3)
185
coincides with C, or
For the hyperbola, in the particuhir case where is on the conjugate axis, wc need only
in L, so that
divide
OC
LP
parallel to
hyperbola in 1\
for
AA' .pa=uL: LC = OP
:
and
[The particular case
is
= CN.NK, NK^NCr.
that in which the hyperbola used
two straight
lines.]
Proposition 103.
[V. 62, 63.]
If
he
an internal
point,
(J
a normal
to the conic.
186
the preceding
The
is
obvious
and
or ellipse
MK.HN=OM:
=
:
OL
..
CM CH. CM CH = CM MK CH HN
:
= CK:CN
..
NK:CN = HM.CH
= 2^a :AA'\
..
NK=NG,
and
PO
is
a normal.
AND
MINIxMA.
Proposition 104.
[V. 64,
(J5,
66, 67.]
If be a jjoint below the axis of any conic such that either no normal, or only one normal, can be drawn to the curve through
which cuts the
then
aa-is {betiueen
and C
OA
is
which
If
least of the lines OP cutting the axis, and that nearer to OA is less than that which is more remote.
is the
must have
AM>^,
and
also
(a)
OM must be
parabola
ij.P^N^
(6)
= N^H:HM:
OM > y,
for
to
between
< NG
for
for
any point
which N^K^
= N^G^.
188
Also
for
than P^
it
is
still
true that
I.
< NG.
P,
NK < NG.
Draw
the
Let
tangents
A F,
since
PT.
OA
AL,
is
is
obtuse.
pendicular at
to
AO,
and
as
falls
< NG,
PG
perpendicular
PT, the
angle
(1)
OPT
is
acute.
if possible,
Suppose,
UP= OA.
as centre describe a circle.
With
OP
as
radius
and
but
AL
will lie
wholly without
in
it.
as R.
li
is
RU
ORU
it
acute.
RU
But
falls
wholly
Therefore
(2)
OP
is
not equal to
OA. OA.
Suppose,
if possible,
OP <
189
as centre
and
UP
is
as radius
AM
And an
absurdity
Therefore
i.e.
OP
is
neither etjual to
(J
nor
loss
than OA,
()A
It
< OP.
if
OP < OP'.
If the tangent
TP
OP,
be produced to
T', the
angle
OPT'
is
OPT is
viz.
acute.
ialls
PE,
will
apply to P, P'.
Therefore
II.
OA <
OP,
OP <
OP', &c.
Where
drawn from 0, the above proof applies to all points between A and P, (excluding P, itself) and also applies to the comparison between tAvo points each of which is more remote from A
than
P.
190
It only
(a)
OP
between
and OP^,
)
(a)
straight line
Suppose first, if possible, that OP = OP^, and let Q be any point between them, so that, by the preceding proof, OQ > OP. Measure along OQ a length Oq such that Oq is as centre and Oq as greater than OP, and less than OQ. With This circle radius describe a circle meeting OP^ produced in p^.
in
an intermediate point R.
OQ
is less
than Oq
which
is
absurd.
Therefore
OP
is
Again suppose,
on OP, a length
absurdity
is
if possible,
OP >
OP^.
Then, by taking
less
0;j,
greater than
OP^ and
than OP, an
Therefore, since
OP is neither equal
OP<OP^.
(b)
If
OP'
exactly similar
more remote from than proof will show that OP^ < OP'.
lies
is
07-*,,
an
completely established.
Proposition 105.
(Lemma.)
If two
conic meet in T,
and
if
TQ <
TQ'.
propcjsition
for
is
The
hyperbola and
the case where Q, Q' are on one quadrant of an ellipse: for the angle TVQ' is greater than the angle TVQ,
and
QV=Vq.
191
TQ
is less
ellipse,
In the case where Q, Q' are on different quadrants of an produce the ordinate Q'N' to meet the ellipse again
Join q'C and produce
in q.
Q'N' Let
= N'q',
ellipse in R.
Then
RM be the ordinate of R.
NOAV
..
RM>QN;
CQ > CR,
>0Q';
..
zGVQ>zGVQ'
TQ<TQ'.
and, as before.
Proposition 106.
[V. 72.]
below the axis of a jxirabola or hyperbola If from a point draw two normals OP^, OP^ cutting the axis (P, being nearer to the vertex A than P^), and if further be any othei point on the carve and UP be joined, then
it
is possible to
192
(1)
the lines
behueen
is the
greatest of
all
OP, and
OP^ on each
side is
more remote;
P.^,
if
lies
OP^
is the
OP^
is less
than the
more remote.
By
but
Prop. 99,
if
NK < NG.
we
Therefore,
find
in Prop. 104,
between and P,, OP is not a normal, by the same proof as that employed moves that OP increases continually as
is
from
as
have therefore to prove that OP diminishes continually moves from P, to Pj. Let be any point between P, and Pj, and let the tangents at Pj, meet in T. Join OT.
Then, by Prop. 105,
Also
since
A We
towards P,.
< TP.
obtuse.
Therefore
Similarly
OP^.
if
P'
is
a point between
andP,,
OP'kOP.
That OP increa.ses continually as moves from P, further away from A and P^ is proved by the method of Prop. 104.
is
established.
.
7
.
193
Proposition 1
[V. 73.]
If
it is
sucJi that
ivhole
of
ABA', then, ifOP^ be that normal and P, is on the quadrant AB, OP^ nill he the greatest of all the straight to the semi-ellipse, and that which is nearer lines drawn from
the semi-ellipse
to
OP^ luill be greater than that which OA' will be the least of all the straight
is
lines
the semi-ellipse.
It follows from
Props. 99 and
101
thcat,
if
OM
be per-
between C and A', and that OAI must be greater than the length y determined as in
must
lie
Prop. 99.
Thus
to
A' than
is
is, it
is
is
nearer
that
104
OA'
the least of
all
OP
increases con-
tinually as
and P, we use the method of For any point P' between Prop. 106, drawing the tangents at P' and B, meeting in T.
u. c.
13
194
Thus we derive
OB < 0P\
increases continually as
For the part of the curve between P, and A we employ the method of reductio ad absurdum used in the second part of
Prop. 104.
Proposition 108.
[V. 74.]
be a point below the major ao^is of an ellipse such that If two normals only can be draiun through it to the whole semiellipse ABA', then that normal, OP^, which cuts the minor a^is
is the
greatest of
from
to the semi-ellipse,
and
which
is
more
least
remote.
to the
nearer vertex A,
is the
if be nearer to A than to is the centre of curvature, which on the quadrant AB, and that OP^ is one of the only two
99 that,
is
on the
quadrant
A'
also
0M=y determined as in
OP
Prop. 99.
moves from A to P, by the method of Prop. 104, as also that OP increases as moves
from P, to B.
That
from
it
OP
increases as
to P^,
established by the
method employed
195
Proposition 109.
[V. 75, 76, 77.]
//'
he
ABA'
A),
at points
AB and
P^ on
the
to the vertex
OP^is
the
greatest of
all lines
drawn from
to
to
points
OP^ on
more remote
lines from
to
(2)
OP^
is the
greatest of all
to
points on the
on either side
semi-ellipse
is
from
OP^
of the two
established by the
p^
method of
Part
(1)
is
proved
107.
by
the
method of Prop.
It
remains
to
prove
(3).
a|
We
GN^
have
.N^G^
= A A'
p^
= CN^
< MN^
whence
^, a
Nfi,
;
fortiori,
MG,
, < MG,
OM.P^N^<OM:P,N^,
or
in
jt),
we have
at once, on producing
OM to
;
P,R>PA
so that
..
Op,
> OP,
a fortiori
0P^>
OP,.
132
196
we have
OB
OB
is
to the curve,
If
be on the minor
axis,
the
all
to the curve,
When
unequal.
if,
the same
number
of
ordinates being
drawn
all
to the axis at
proportional
distances
The
is
segment of a
circle or
conic
4.
called the
base
of the segment.
The diameter
it
of the
segment
is
is
other in such a
each parallel
is
abscissa in each.
198
2.]
In two parabolas,
if the ordinates to
a diameter in each
parabolas are
are inclined
to the respective
the corresponding
equal.
(2) If the ordinates to a diameter in each of two hyperbolas or two ellipses are equally inclined to the respective diameters,
parameters are
at once established
by means of the
fundamental properties
(1)
QV' = PL.PV
QV* =
for for
(2)
PV.VR
1
proved in Props.
3.
Proposition 111.
[VI. 3.]
Since an ellipse
is limited, tvhile
proceed
to
infinity,
other curves.
be equal to
a hyperbola.
For,
if
AN, AN'
have
for the
parabola
AN
possible, because
which
is
im-
A'N'.
[Here follow
the
six
symmetrical
form of a conic,
re-
produced.]
199
(2)
is similar to the "figure" on a diameter of the other and the ordinates to the diameters in each make equal angles ivith the diameters respectively.
the "figure"
on a diameter of one
(1)
The
result
is
FN'=Pa.AK
Suppose the diameters to be axes in the first place (conjugate axes for hyperbolas, and both major or both minor
(2)
axes for ellipses) so that the ordinates are at right angles to the
diameters in both.
Then the
fore,
ratio pa AA' is the same in both curves. Thereusing capital letters for one conic and small letters for the
:
AN
:
an equal to
AA'
:
aa',
we have
at the
But
because
or
PN^ AN. A' =pn' an.na'. AN. A' AN^ = an na' /^^ A'N AN= a'n an PN'':AN'=pn':an\ PN AN = pn an,
:
. : : :
..
',
is
make
equal
Draw tangents at P, meeting the axes in T, t respectively. Then the angles CPT, cpt are equal. Draw AH, ah perpendicular to the axes and meeting CP, cp in H, h and on GH,
;
through A,
a.
described in R,
200
on
PP' ]'
,
are similar,
.
or
Prop.
1 4]
whence
and, since the angle
AV
A VC
VR = av
vr
{a),
is
etpial to the
angle avc,
it
follows
201
/,
i
circles,
and
the middle
points
VA AI = va
and, since
ai
ZKVI=
KVI,
kvi are similar.
kvi,
the triangles
vi
are divided at
-i4,
in the
same
ratio
the triangles
KVA,
ZAKV= Zakv:
202
KGA =
Therefore, since
triangles
kca.]
the angles
at
F,
j)
are also
equal, the
CFT,
and
it
will follow
FN':CN.NT = 2^n'-cn.nt,
whence the
its
ratio of
'
to its
to
[Prop. 14]
Proposition 113.
[VI. 14, 15.]
A
ellipse
parabola
;
is
neither similar to
is
a hyperbola nor
to
an
and a hyperbola
Proposition 114.
[VI. 17, 18.]
(1)
If FT,pt be tangents
t
to tivo
axes in T,
if,
respectively be
further,
so that
FV,
measured along
F,
FV:FT = pv:pt,
and if QQ', qq
respectively:
be the chords through
V,
parallel
to
FT, pt
then
the
segments
similarly situated.
(2)
And,
conversely,
if
the
simiUirly situated,
FV: FT = pv
and
the
tangents
are
203
Draw
P,
in
H,
and
let
PL 2PT = OP PH\
:
and
where
0,
pi
2pt
op
-.pit,
AH, PT and
the
ah, pt.
Therefore
PL, pi
are
pv.
the parameters of
ordinates
[Prop. 22]
to the diameters
PV,
Hence
(1)
since
zPTA=Zpta,
,
a
Z0PH=
:
Zoph,
= op OP PL PT = pi
:
ph
pt.
But, by h}^othesis,
..
and, since
QV
pi,
is
qv
between pv,
QV:PV=qv
.pv.
204
Similarly,
if
V,
v'
be points on
PV, pv such
:pv',
that
PV: PV'=2)v
and therefore
it
PL
PV =pl
pv',
V,
v'
are in the
same
(Def. 6.)
If the
=
and
:
Zpta,
:
PV PT = pv
Now
the tangents at P,
2)t.
tively,
PTA,pta
by similar segments,
QV: PV=qv
while
..
:
pv,
:
PL QV = QV PV,
PL:PV=pl:pv.
and pi
qv
= qv
:pv\
But
PL 2PT = OP PH)
: :
pi
"Ipt
op
ph
ph,
'
and
by similar triangles.
Therefore
II.
UP
PH= op
:
PV PT = pv
curves be
pt.
If
the
hyperbolas
or
ellipses,
suppose a
and
let
the ordinates
PN, pn
be
drawn
qq',
to the
We
remembering that the chords arc here QQ', and do not pass through A, a.
Since the conies are similar, the ratio of the axis to is the same for both.
its
(1)
parameter
205
[
FX' CN
:
NT = pn'
en
nt.
Prop.
4]
PTN,
Hence
and
Therefore also
CPT=
OP OP
:
cpt
Therefore
But
whence
Also,
PH = op PH = PL
:
ph.
:
2PT\
'
op
:
ph=pl
.2pt
:
PL PT = pl
by similar
triangles,
pt.
PT :GP=pt:ep;
..
PL:CP=pl:cp,
or
PL: PP'=pl.pp'
Therefore the
(A).
similar.
Again,
so that
we made
PV PT =pv PL: PV = pl
:
pt,
pv
(B).
We
and
derive,
be proportionally divided in the points v, the ordinates through these points are in the same ratios.
that, \{
QV:PV=qv:pv,
V,
are equal.
If the
segments are
and we have
QV:PV=qv
206
QV:
Q' V"
qv'
q'v"
and
so that
From
and
these equations
it
follows that
:vv')
'
:
py
:
VV'=pv'
whence
..
vy'j
pv
v'p
P'
V VP -.PV'^ p'v'
. :
pv'*.
But
..
PV':Q'V'=pv":q'v'*;
P'F'. F'P
Q'V"=p'v'.v'p
q'v'^.
But these
parameters.
PP', pp'
to their respective
pp
on the
Therefore
PN""
C'iV
NT = pn'
C7i
nt,
CPT
is
equal
CPT,
CTP
and
it
is
ctp.
:
.
Now,
vp'
qv^,
follows that
whence
CPT,
:
cpt,
CP PT = cp
or
..
pt,
is
CONICS.
207
Proposition 115.
[VI. 21, 22.]
If two ordinates he drawn to the axes of two parabolas, or the major or conjugate axes of two similar ellipses or two similar on hyperbolas, as PN, P'N' andpn, p'n, such that the ratios
AN
and AN'
segments PP',
pp
will he similar
also
PP'
will
not he similar to
any segment in
by
ttvo
and
vice versa.
line.s
of the
and accordingly
it
is
unnecessary to
reproduce it]
Proposition 116.
[VI. 26, 27.]
If any cone
making hyperbolic
On
drawn,
line
and
3, it will
be seen
p'pm
parallel to also
P'PM and
will
dme
parallel
to
DME;
i.e.
p'pm
hyperbola or
ellipse,
be parallel to
dme,
to
DME.
the
ordinates
to
Therefore
the
PL,
pp.
crrvr.
208
PL PP'
.
and
i)l
And they
2)1.
PL PP'
.
cannot be equal to
[Cf. Prop.
pp.
110(2)]
[A
Prop.
similai
1,
PL
a constant ratio.
PROBLEMS.
Proposition 117.
[VI. 28.]
to
to
am
is
the
a.xis
and
BC
OBC
be a
BC.
Measure
along
al
:
OB
such that
OA =
B(f
BO
0(1
H. C.
14
210
THE
COXIt'S
OF APOLLONIUS.
DraAV AM parallel to OC meeting BG in M, and through AM draw a plane at right angles to the plane OBC and cutting
the circuhvr base in
Thi'u T)E
is
DME.
AM,
and the section
perpendicular to
is
DAE
is
AM.
is
Also [Prop.
1],
\ AL
AL:AO = BG'
whence
.BO. 00,
is
AL = aI,
[Prop. 110].
No other parabola with vertex on OB can be found which is equal to the given parabola except DAE. For, if another such parabola were possible, its plane must be perpendicular to the
plane
OBC
:
and
its
axis
must be
parallel to 00.
If
A' were
we should have
if
A'L'
A'O = BG^
BO 00 = AL AO.
.
Thus,
AL
or
Proposition 118.
[VI. 29.]
to
to
a given
{A necessary
of the square on the axis of the cone to the square on the radius of the base must not be greater titan the ratio of the transverse
a.vis
to its
parameter.)
aa', al are
the
Suppose 07"
BP < aa'
al,
vhere I
is
Let a
circle
and produce
01
meet the
circle
again in D.
212
the
that
AL = AJ.^=al.
sections are each equal to the given
For
to
( 1 ), if
OH were
parallel
OH
either Avith
OQ
or OQ'.
This
is
OH
meet
BO
if
in
H,
in R,
and the
circle
which
is
impossible, since
aa':al=OI
II.
.IE=OH:HR.
we
shall
If
and OQ,
or OQ will
:
= aa'
al,
have
01 ID = aa'
:
al,
is
this
If
OP BP > aa'
:
al,
is
For,
if possible, let
ON
be
drawn
meeting
:
BG
in
N.
.
aa'
al
ON''
:
BN
NO,
But
0>.
NC. which
is
absurd.
I'RORLEMS.
2\:\
Proposition 119.
LVL
In a given right cone
In this
F, F'
ciise
80.]
to
to
a given
ellipse.
we
OBG
and suppose
if
taken on
BO
produced
in
circle in Q, Q',
at.
OF:FQ=OF':F'Q' =
Then we
',
^1,/!,' in
the angle
is
BOG
parallel to
OQ
and A^A;
to
OQ.
through
A A',
A^A^'
each
The proof
follows the
proposition.
214
Proposition 120.
[VI. 31.]
To find a rir/ht cone similar to a given one and containing a given parabola as a section of it.
Let
let
OBC
AN
is
AL
and
AN
make
the angle
OBC.
Let and on
AM be taken of such a length that AL AM= EG BO, AM as base, in the plane MAN, describe the triangle
: :
AM
similar
to
the
triangle
AM
AM.
The cone
For
therefore
=
AN,
',
EM
is
parallel to
also a parabola.
Now
or
.'.
AL:AM = BG:BO
= AM:AE, AM' = EA.AL;
PROBLEMS.
21
ot"
Hence
the cone
AL
is
also
is itself
and
AM
is
having
There can be no other right cone similar to the given on., its vertex on the same side of the given parabola, and
iis
a section.
through the axis of this cone a plane cutting the plane of the
given parabola at right angles.
in
intersect
lie in
AN,
F must
the
plane of ^^lY.
Again,
cone,
if
AF,
FR
FR
must be
parallel to liN, or to
EM, and
so that
must
Let
AM
meet
FR
in R.
Then,
the cone
if
FAR
:AF = AR':AF.FR
= AM':AE.EM = AL:AE.
Therefore AL',
is
AL
FA R.
Proposition 121.
[VI. 32.]
a given one and containing a OBC be the given cone and D the centre of its base BG, and if A A', AL be the axis and parameter of tJie given hyperbola, a necessary condition of possibility is that the ratio OB' DB'^ must not be greater than
To find a nr/ht cone similar
to
{If
the ratio
AA' AL.)
:
216
',
the given
in the
/p
puom.KMs.
: :
217
on EI such that FI IH = AA' AL, and Take a point draw the chord QQ' of the circle parallel to AA'. Join A'Q, AQ, and in the plane of the circle draw AR making with AQ an angle equal to the angle OBG. Let AR meet A'Q produced in R, and QQ' produced in N.
through
Join
FQ
meeting
^^'
in
K.
OBC, and
AR'iH
parallel to
QAR is
OBG.
Q and
FQA.
is
We
have, by construction,
AA'
AL = FI
=
-.IH
:
FK
KQ, by
parallels,
QKAN,
A'K:KQ=^QN:NR,
and
KA:KQ = QN
A'
.
:
A,
:
whence
KQ' = QN'
AN
NE.
It follows that
AA':AL = QN':AN.NR.
Therefore [Prop, 2]
section
of the cone
hyperbola.
axis
The two hyperbolas accordingly have the same and parameter, whence they coincide [Prop. 110 (2)]; and
the cone
QAR
218
of
Q and No
proceeding as before.
is
same
side of the
For,
its
if
right
angles.
The
plane
thus
described
must
is
then
pass
lie in
must
FQA.
lie
similar to
must
FP
meet A'A
in
T)
..
which
II.
is
impossible.
Suppose that
OD'
so that
'
:
= AA' AL,
:
FI IE = AA' AL.
:
apex
In this case Q, Q' coalesce Avith E, and the cone with as diameter perpendicular and base the circle on
AG
to the plane of
III.
FQA
is
If
UD-: DB''>AA'
AL, no
possible, let
FT:TP = AA'.AL
But Hence
Therefore, etc.
or
FI
is
IE.
FT TP < FI
:
IE, which
absurd.
PROBLEMS.
21)
Proposition 122.
[VI.
:VA.]
to
it.
a given
ellipse as
section
of
As
on
through
'
perpendiciUar to the
and
in the plane so
circle
drawn describe
Bisect
AA'
as base a
segment of a
to the angle
BOC, the
Draw two
and such
lines
FK, FK'
to
FK KQ = FK'
:
K'Q'
= A A' AL.
:
AN parallel to QF, meeting in N. Join AQ, A'Q, and let A'Q meet AN in R.
DraAv QiV parallel to AA', and
Q as apex and
as
bii.se
the circle
on
of
AR
as diameter
and
in
AFA'.
This cone
will
is
one of
its sections.
220
..
And
zAQR=zAFA'
=
BOG.
are similar,
QAR, OBG
:
and likewise
QAR, OBG.
A A' AL = FK KQ,
:
by construction,
),
by parallels,
Therefore [Prop. 3]
section
ellipse.
of the cone
AL QAR
is
And
AL
is
In like manner another similar right cone can be found with apex Q' such that the given ellipse is a section.
No
same
Draw
in
PM
M.
if any, must lie on the arc A FA'. A'A, and A' parallel to FP, meeting Join AP, A'P, and let A meet A' in S.
P,
parallel to
The
shall
triangle
:
PA'S
will
have
A'
TP^ = FT. TP
TP\ in
the
same way
We
and
must
therefore have
AA'
this is impossible,
AL = FT: TP
because
AA'.AL = FK:KQ.
123 (Lemma).
1.]
[VIL
In a parabola*, if along the aocis a^uay from
PN be an
ordinate and
to the
AH be
vieusnred
and equal
latus rectum,
AP' = AN.NH.
[=AN{AN + p)]
This
is
PN^ = p AN,
.
by
adding
AN^
Proposition
124 (Lemma).
2,
[VII.
:l]
If
if
A A'
be divided at
H, internally for
so
the hyperbola,
and
then,
exte^'n ally
PN
*
that
AH
HA' = p. AA\
AP':AN.NH=AA'.A'H.
Though Book VII.
is
central conic, one or two propositions for the parabola are inserted, no doubt
in order to show, in connection with particular propositions about a central
conic,
of the parabola.
222
Pro(iuce
to
K,
so that
thus
AN.NK.AN.A'N
= PN':AN.A'N = p: A A'
=
[Prop. 8]
.
AH
:
= '^*;
A'H, by construction,
or
NK:A'N = AH.A'H.
It folloAvs that
Hence
..
A'
A'N=AA' A'H;
:
: :
or
Thus
But
AN.AK:AN.NH = AA':A'H. AN.AK=AP\ since AN.NK = PN\ Therefore AP^ AN.NH = A A' A'H.
: :
The same
ellipse
proposition
and
;>
true
\ AA'
a.xis
of an
223
Proposition
125 (Lemma).
[VII. 4.]
If in a hyperhnUt or an
A' in T, and if
meet the
aa-ift
OD
PT,
then
CD' =
NT
CN.
Draw AE, TF
let
at right angles to
in
0.
CA
to
meet
if
PT
Then,
we have
^.PT=OP:PE.
Also, since
[Prop. 23]
CD
is
parallel to
PT,
it is
conjugate to CP.
(1).
Therefore
^.CP = CD'
Now
OP :PE=TP:PF;
..
%.PT = PT.PF,
.PF = Pr
.(2).
From
(1)
and
(2)
we have
:
CD^ =
PF:GP = NT -.CN.
224
Proposition
126 (Lemma).
5.]
[VII.
In a parabola, if
he the latus rectum,
he the
parameter of
X the principal
ordinate,
and if
AL
p^AL + 4>AN.
meet
PT
in
E, and
let
PG,
at right angles to
PT, meet
the axis in G.
similar,
22.')
The property
is
halt'
th.
it
latus
rectum)
by regarding
as
the
Cf.
perpendicnlar through
Prop. 85 where the normal
line
to the tangent at
is
that point.
from
to the curve.]
Def.
If
AA'
be divided, internally
and
H'
such that
A'H
where pa
is
AH= AH'
to
A'H'=AA'
the
p^,
A A',
then
are
AH,
pa
in
proportion)
called
A A' may
minor axis of an
ellipse.
Proposition 127.
[VII.
//"
6, 7.]
AH, A'H' he the " hoDwlogues" in a hypei'bola or an ellipse, and PP', DD' any two conjugate diameters, and if AQ he draivn parallel to DD' meeting the curve in Q, and QM he perpendicular to AA' then PP" DD" = MH' MH.
,
: :
let
the tangent at
meet
A A'
in T.
Then, since
in V),
A'C= CA,2a\aQV= VA
CV.
(where
GP
meets
QA
1
A'Q
is
parallel to
Now
And,
also
.CD' =
NT CN
:
[Prop.
2]
=
by similar
AM
A'3i,
by similar
triangles.
triangles,
CP':Pr = A'Q':AQ\
whence, ex
:
aeqiiali,
CP' CD' =
(A'Q'
AQ')
: .
X
H. c.
MH
AQ').
'
226
THE
C'OXICS
OF Al'OLLONIUS.
227
.'.' = ':',
:
and
Also
-.
).
It follows that
or
This result
may
PP'
where
is
= MH' MH,
:
Proposition 128.
[VII. 8,
9,
10, 11.]
last
and
A A''
{PP'
:
+ DD'f = A'
.
AA''
:
AA'' {PP"
(1)
We
have
..
AA" PP" = CA' CT' AA" PP" = CN. GT GP' = A'M. A' A A'q\
:
: : : :
[Prop. 14]
by similar
triangles.
Now
A'Q'
A'M.
[Prop. 124]
A'M. A' A
A'Q'
= A'M. A'H
A'lM MH'.
.
AA":PP" = A'H:MH'
= A'H.MH': MH'\
(a),
152
228
Again,
PP" DD" = MH' MH {), [Prop. 127] = MH"':MH.MH' PP' '.DD' = MH' \/MH MH' (7).
:
...
Hence
and
MH"
:
{MH'
^MH.MH'f
A A"
(2)
{PP DD'f =
We
Therefore by
(a),
:
ex aequal
.
(3)
From {),
MH MH'. MH + MH'.
Therefore by
:
(a),
ex aequali,
:
Proposition 129.
[VII. 12, 13, 29, 30.]
/??
and
of the squares on any two conjugate diameters is equal or difference respectively of the squares on the axes.
sum
Using the
propositions,
figures
we have
p
construction,
Therefore
ellipse),
(a).
220
AA'':FF" = A'H:i]IH',
and, by
means
:
of
()
in the
same
proposition,
:
MH MH' = MH'.HH'.
we
obtain
:
we have
at once
Proposition 130.
[VII. 14, 15, 16, 17, 18, 19, 20.]
Tlie
from
the preceding
proposition,
(1)
For
the ellipse,
A A" FF"
:
DD" = A'H
if
2CJ/;
denote the parameter
the ellipse
to
of the ordinates
(2)
(3) (4) (5) (1)
:
AA"
p'
= A'H. MH'
.
MH\
MH'f,
We
have
AA"'
and
FF'^
= A'H MH',
:
FF" FF"
:
'
{}]
the
ellipse.
AA":FF"--
JJJJ"
= A'H
2CM/.
230
(2)
"
..
(3)
By
Prop. 127,
..
PP"
And
..
{PP'
AA"
(PP' pf = A'H
:
(4)
and
as before,
= MH'.MH;
..
[Prop. 127]
AA'':PP'.p = A'H:MH.
:
(5)
and
AA" :PP" = A'H. MH' MH", as PP" PP" p' = MH" MH" MH\
:
before,
MH" + MH\
Proposition 131.
[VII. 21, 22, 23.]
In a hyperbola, if
AA'
^^.J
BB',
then, if
PP', DD'
he anij
;
respectively
and
ratio
PP'
DD'
continually
"^
>
J
as
(or increases
moves
farther
from
on either side.
Also, if
AA' =
BB',
PP' = DD'.
2^1
to
Of the
AA' >
AA'kBB'.
Taking then the
from
\
figure
respectively,
it
follows
PF'' .DD'* =
that
MH'
MH
:
[Pn.p.
27]
Also
AA
''
:
BB"'
= A A'
pa
construction,
and
while
AH'
AH ^^> MH'
\
MH,
"^
MH'
i.e.
MH
continually as
moves further
(or increases]
from A,
as Q, or P,
Therefore
AA"
\
BB'\^.^
^
i
PP" DD'\
:
and the
latter ratio
as
^4.
(or
mcreasesj
And
the same
is
AA' = BB',
then
AA'=pa, and
both
and //'
coincide with G.
AH = AH' = AG,
MH = MH' = GM,
and PP'
= DD'
always.
Proposition 132.
[VII. 2+.]
if
be
A A'
any
be the imijor,
and BB'
moves from
to
B.
232
THE
COXIC.S
OF APOLLUNIUS.
.
We
have
CA' CB' = AN
:
''
PN'
..
AN.NA'>PN\
CA' > CP\
and, adding
C'iV"^
to each,
or
AA'>PP'
(1).
Also
where
DM
is
Therefore
and, adding
BM
MB' <
DM\
;
CM\
..
BB'kDD'
(^)
P^P^, D^D^ be another pair of conjugates, P, buing further from A than P, D, will be further from
Again,
if
than D.
And
But
and
A' AN^ N,A' = PN' P^N;". AN^.N^A'>AN.NA'; p,n;'>pn\ AN^ N^A' - AN A' > P^N^' - PN\
AN.
..
But, as above,
and
AN^ N^A > P,N^\ AN^ N^A '-AN. A = GN' - CiY," CN' - GN^' > P^N;" - PN'
. ' .
'
..
thus
or
GP'
> GP^\
(3).
PP'>PJ\'
In an exactly similar
DD' <D^D;
(4).
238
We
(2),
AA'.BB'>PP'.DB',
and, by (3) and (4),
D^D^'.
/),
Cor.
It
is
are the
parameter
corresponding to
A A',
Pct<p,
P<p
etc.
Proposition 133.
[VII. 25, 26.]
(1)
In a hi/perbola or an ellipse
PP\ DD'
are
any conjugate
diameters
other
than
the axes.
In the hyperbola PP' + DD' increases continually as (2) moves further from A, while in the ellipse it increases as moves from A until PP', DD' take the position of the equal
conjugate diameters,
(1)
lulien
it
is
maximum.
~
AA"
or
BB" = PP"
(
DD"
.
[Prop. 129]
{AA'
-\-
DD
'),
..
Similarly
it is
AA' ~ BB' > PP' ~ DD' AA' + BB'<PP' + DD'. proved that PP' + DD' increases
]
as
moves
further from A.
For the
ellipse
:
:
..
{A A"
But
*
AA' BB' > PP' DD' + BB") {AA' + BB')' > {PP" + DD") {PP' + DD'f.* [Prop. 12!>] AA"+BB" = PP" + DD": AA' + BB'<PP' + DD'.
:
:
..
ApoUouius draws
no intenuediute
stt'pe.
234
Similarly
it
moves from
that
PP' + DD'
increases as
Proposition 134.
[VII.
-27.]
II
AA''-BB'>PP' -DD',
luliei'e
PP',
DD'
Also, as
moves from A,
tinually,
PP' - DD'
and
PP',
DD'
is
For the
ellipse
the
proposition
clear
AA"
and
It follows that
~ DD",
latter
diminishes continually as
moves further
come before
regards the
really used
(so
far as
Proposition 135.
[VII. 28.]
In every hyperbola or
.
ellipse
.
as
the
DD'
coincide
Wc
so that
have
..
[Prop. 133]
235
[Prop.
2!)]
manner
it
will
be shown that
PP
DD'
increases
PP',
DD'
Apollonius]
increase continually as
is
the proposition
obvious.
Proposition 136.
[Vll. :u.]
If PP', DD'
and if tangents
be
the
parallelogram
LL'MM' =
red.
A A'
BB'.
Let
such that
length
PO
PO' = aN.NT.
Now
or
..
. 0::
.
:
[Prop.
4]
Hence, alternately,
CA'
or
PO.CT = CA GB GT PN,
.
CT.CN:PO.CT=GA.GB.GT.PN
(1).
236
Again,
so that
CD' = NT CN,
: :
[Prop. 125]
CPT
'DC = NT
is
CN.
proportional between
2
for
mean
Also
PO
is
CN and
:
iVT.
Therefore
2ACPT
And
or,
(CZ)
2ACPT=CT.PN;
..
(CL)
side,
= CA CB,
.
quadrupling each
CJLL'MM'^AA'.BB'.
Proposition 137.
[VII.
3:},
S^,
:3.5.]
Supposing pa to be the parameter corresponding to the axis A A' in a Jujperhola, and to be the parameter corresponding to a diameter PP',
(1)
if
A A' he
< p, and
increases
continually as
2ii7
if
A A'
'-^'
,
then
p,,
<
p,
and
(0)
increases as
if
AA' <
found a diametei'
,^
on
p^='2P^P^.
increases as
and
from Po
(1)
ill
either direction.
(o)
'=,
we have
PP'=p =
and PP', and therefore away from A.
p,
I)D\ moves
increases continually as
(b) If AA'>pa, AA'>BB', PP' DD', and therefore PP' p, moves away from A. But PP'
: :
and, as
in
Prop.
131
(1),
diminishes continually as
increases.
Therefore
in-
creases
(2)
all
the more.
Suppose
AA'<pa
but
^^.
A
;
Let
draw
A'Q
parallel to
CP
in Q,
and draw
the ordinate
QM.
A'H
HA = AH'
H'A' =
A A'
pa,
238
Thcreiuic
THE
We
And
have now
" ' AH
:
COXIC'S
pa'
OF AI^OLLONIUS.
.
AH'
(a).
iif
2AH'.
..
or
It follows that
or
AM
^F',
Therefore, componendo,
<MH':AH'
whence
(7),
:
or, alternately,
A'
(2),
and by the
:
Therefore
or
pa.
Pa<PAgain,
is
is, and if Pj be a point further from A than parallel to CP^, and il/, is the foot of the ordinate Qil/,,
if
A'Q^
then, since
AH
:|*
2AH',
MH < 2MH'
also
Thus
This
is
is
(,
M^
for
M.
We
to
()
above, or
whence
or
so that
/)
A'
proposition
is
proved.
LENGTHS OF
(3)
DIAMKTKllS.
29
Now
Mo
let
^^'
.1/,,
be
less
than
2
Take a point
related to
such that
HH' =
Q,
:
'^, and
are to
il/.
let Q,.,
1\ be
in the
Then
PoPo' Po
= M,H' M,H.
that
[Pn.,,. 127]
It follows, since
HH' = H'M,
P,= 2P,P:.
Next,
let
^l,
and
Q,
corresponding points.
Then
since
M,H'.H'M<HH'\ MH'<M,H'.
to each side the rectangle
Add
have
and we
(MM+HAI)MH'<iMH\
This again corresponds to the relation
substituted for
above, with
, M^
for
The
is
..
.,'
or
AA'':p:>AA":2f.
Therefore
>pn
And
in like
jj
increases as
7^
moves
from Po to ^.
Lastly, let
than P^
is.
In this case
H'M >
.
'Mo,
and we have
AA"
and
p'
<
AA"
po\
p>Po
In the same way we prove that
increases
ai>
moves
further
away from
and A.
is
Hence the
proposition
established.
240
Proposition 138.
[VII. 36.]
In a hyperbola
corresponding
to
witli
A A'
and
AA' -pa>PP''P,
and PP' diminishes continually as
whence
A'H HA = AH' H'A' = AA' p, A A" (A A' ~ paf = A'H. AH' HH".
: : : :
A A"
But
..
AA'-
Hence
Similarly,
if
M^ be
further from
than P,
are,
we have
A'H.M^H'>A'H.MH',
and
it
follows that
PP''-p>P^P^' ^p,,
and so
on.
Proposition 139.
[VII. 37.]
In an ellipse, if P^Po, Df^D,'hethe equal conjugate diameters and PP', DD' any other conjiigate diameters, atid if po, p, Pa, Pb
he the
parameters corresponding
to
PqPO, PP',
A A', BB'
respectively, then
(1)
AA' ~ Pa
is
the
maximum
value
all points
hetween
tinually as
moves from
to
Po,
241
BB' -
pi,
is
the
maximum
2\,,
points
between
and
to
ami
'
value of
PP' -
for
all
diminishes continually
as
passes from
(3)
Po,
BB'-pu>AA'-pa.
results (1)
The
(3)
and
Since pb
once that
BB'
-^
pi,>
y),
and
pt,
> A A',
it foiiow.s
at
Proposition 140.
[VII. 38, 39, 40.]
(1)
less
than
I j),
PP' + p >AA'+pa,
luhere
PP'
is
the
corresponding
If AA' <^p)a,
there
is
on
each
side
;
of the axis a
= ^Po
(f>i(l
,^^'+
is
than
side
PP' + p, where PP' is any other diameter on the same moves away from of the axis. Also PP' + p increases as
The construction being the same
AA'-^pa.
as before,
P..
(1)
(ft)
we suppose
PP'
increases as
moves from
PP' +
Suppose
AA' <pah\\t
..
-i^^pa',
AH'^\AH\
AH'-^liAH + AH'),
-^{AH ^ AH')\
thus
and
{AH +
Hence
4>{AH+AH')AM
AM:AJI',
;
(.1 //
C,
lt>
242
and, componendo,
Now
(3iH + MH'f -{AH + AHy = 2AM(3IH + MH' +AH +AH')
..
It follows that
or
..
MH' AH' < (MH + MH'y (AH + A'H.MH' {MH + MHJ < A'H .AH': {AH + AH J
:
AA'':{PP' + pf<AA":{AA' +
py
Hence
Again, since
4
if
And,
Qi
,
than
is,
and
il/,
points corresponding to Q,
for
,
M, we
il/j
as before (substituting
'.'
We derive
(2)
A and
M),
:
PP'+p< P,P^+p,
is
established.
so that
We
have
AH'< \HH'.
MoH' = ^MoH.
Make H'M^
Then
and
Next, since
equal to
so that
P,P:
^Jo
= M^H' .M,H=l:S,
PoPo'
-/-'^'
f. = i -^^o-H",
so that
Now
suppose
to
be
..
4il/^'.
243
(M^H +
.
}'),
;
..
{M,H + MH')
4il/il/
>{MJI+MH')AMM..
Therefore, componendo,
(
-.
or
MM..
Mil',
')\
,': MH'>(M,H+MH')
Hence
A'H.MoH'
{M,H+M,Hy
:
>
AA"
and
Again,
if
(' +
i>o)'
PP' + p>1\P:+p,.
Pi be a point betveen
and A, we have
.
^MH',
P,P;+p,>PP' + p,
and so
on.
Lastly, if
> MH, we
.
shall
(
have
J/ //
(MH + MM')
If to both
sides
,'.
bo added
the
of this
inequality there
rectangle
(MH + ,,') ^fMM^, they become respectively (MH + M,H') ^MH' and (.1/// + MH')\
.
+ p<PP'+/>,
established.
and so
on.
is
IG 2
244
Proposition 141.
[VII. 41.]
In any ellipse, if PP' be any diameter and PP' -\-p> AA' -{>, and PP' + ]) is the less the A. Also +.
its
parameter,
is to
'->'
nearer
Q,
as before,
A'H.HA = AH''.H'A'
= AA''.p, = p,:BB'. A A" {AA' + 2^ = '-' HH" = A'H.AH' '.HH'"AA'^:BB" = AA':pa = A'H:A'H' = A'H.A'H':A'H" BB" {BB' + p^f = A'H" HH"
:
Then
Also
(a).
\
i.
J
and
Therefore, ex aequali,
().
AH' > A'H', AA' + pa<BB' + pi,. Again AA":{PP' + pf = A'H.MH' HH'\ [Prop. = A'H M,H' HH'\ and A A" +
From
(a)
(/ ,
130
(3)]
vhere Pj
is
between
it
follows, since
AA'-\-2)a<PP'+P,
P,P,'
'
+ + ,, + p,<BB' + p,,
,,'
follows.
24.")
Proposition 142.
[VII. 42.]
//;
luith
parameter
p,
AA'.pa<PP'.p,
and PP' .p increases as
moves away from A.
We
and
while
have
A'H
A'll
HA = A A"
AA'.pa,
[Prop.
1
= A A'' PP'.p,
:
30 { 4 )]
AH<MH
..
AA'.pa<PP'.p,
moves from A,
so does
;aid, since
MH increases as
PP'.p.
Proposition 143.
[VII. 43.]
In an ellipse AA'.pa< PP'.p, where PP' is any diameter, moves aivay from A, reaching a and PP'.p increases as
maximum
The
luhen
is
coincides with
or B'.
result
PP'.p = DD'\]
Proposition 144.
[VII. 44, 45, 4.]
In a hyperbola,
(1) (2)
if
if
Pa, or
Pa, but
then
AA" + Pa'<PP" + p\
where PP'
is
moves
away from A
24U
(3) if
side of the
and
," + "
is to
other diameter.
Also PP''-
+ p^
PP'
PoPJ.
(1)
Let
if
Then,
as
increases
moves further from A [Prop. 137 moves further from which increases as
..
AA' <PP',
AA"-^pa'<PP"+p\
moves further from A.
Let
A A'
be
:
less
j^a
A A" ^
:
(A A' ~ paf.
Then, since
AA'
AH = AH'
A'H',
2AH"^HH'\
and
niH'.AH' >HH'\
Adding
2AH .AH' to each side of the last inequality, 2{MH + AH')AH'>2AH.AH'+HH" >AH' + AH",
or
..
2{MH+AH')AM:2{MH + AH')AH',
<
2
:
AM
AH',
Therefore, componendo,
MH':AH'<2{MH + AH')AM + AH' + AH".AH' + AH'\ and MH' + MH" = AH' + AH" + 2AM {MH + AH'), MH' AH' < MH' + MH" AH' + AH", so that .MH' MH' + MH" < A'H .AH' AH' + AH"; or AA":PP" + p'<AA":AA" + pa'. [Prop. 130(5)]
:
..
Thus
Again, since
and
(if
HH",
for
and
il/,
(3)
Let
'
247
than
^{AA' -
/))',
so that
2.Air<III['\
Make 2MoH''
Now
so that
,"
A
.p.
27]
Next,
if
be between
and Po>
"*
and
= HH", 2M,H'.MH'<HH'\
2.1/oif
Adding
2
2MH.MH'
to each side,
we prove
that
'-+<"-' + /.
Again,
if
Pj be between
and P,
,',
same
'^+/</^+
Similarly
Lastly, if
AM > AM^,
2MH'.iMoH'>HH", AM, > AM,
and,
if
2M,H'.MH'>HH";
whence we derive
in like
manner that
iV,
PP''+/>PuP..'-' +
PJ\"+p;'>PP"+p\
and so
on.
248
Proposition 145.
[VII. 47, 48.]
In an
(1)
if
A A"
when
-if
\{AA' + pa)\
coincides with
then
"
and
maximum
(2)
of the
and
side.
^
accis
if
AA'^ > ^{AA' + paf, then there luill be oii each side a diameter PqPo such that PqPo'^ = ^{PoPo + 2>)\ -p^ in the same - pn will then he less than
''^
(1)
Suppose
AA":i(^^{AA'+2)af
:
Also
+ A'H" = AA" AA" +Pa' BB" = pi, BB' = A A' p^ = A'H A'H' AA"
A'H. AH'
:
'
and
BB" (BB"
:
+pi,')
= A'H" A'H' +
:
A'H''
hence, ex aequali,
A'H'
A'H' + A'H'\
and, as above,
AA"
Again,
..
{AA'-'+pa')
AA":i^^{AA'+pJ,
2A'H.AH'^HH'\
whence
Subtracting
2 ',
2 A'
.
'
:
<
".
we have
..
and, since
2A'M.MH'<MH' + MH" (1), 2A'M.AM 2A'M. MH', or AM MH', >2A'M.AM -.MH' + MH", 2.1'.!/ AM + MH' + MH" = A'H' + A'H",
: .
'
24!)
we have, compunendo,
..
whence
A A"'
AH' MH' > A'H' + A'W' MIP + MH'\ A'U.MW MIP + MH'\ (^1^1'" + ^v) > A A {PP" -f /),
:
:
'=
[.
130
(.->)]
Again, either
(a)
< M,H\
ur J/i/.^
M,H'.
Let
J/i/-^
MH<M,H'.
+
J/7/'^
Then
and
..
>
J/.y/^
+ M,H'\
2
J/jiT'
{MJi' - iV//)*
JAUi 2{M,H'-
MH)
.
MM,
.1/. //',
..
+ J/, H'\ But il/if^ + il/^'* - (M, H' + M,H") = 2 {CM* - CM;') il/J/i 2(/^' - MH) + M,H" + M,H" = J//P + il///"
2 (i/iZT'
:
> MM,
MH)
i/,^^
thus, componendOy
we have
: :
and
so that
(b)
yl^'^
"
PP"^
+/ >
^1^'^
'^ +^
MJP + MJP\
[Prop. 130 ()]
+ '<,,"+'
If
MH<^M,H\
MH' + MH"
and
it
^ M, H' + M, H'\
as before, that
:
results, in the
:
same way
MJP + Mjr,
PP''+p-<PJ\''+p;'.
.'
.
BB'+po',
i^ /^"
and
*
MJP
.V,
= A A"
+ ih\
As
in (1) ftbove,
.1/,//
(.!/,//'
J/i/),
u fortiori.
2')0
THE COMCS
shown
in the
(JF
Al'OLLONIUS.
it is
same
nianiier that
(2)
Suppose
so that
2AH">HH'\
equal to
Make 2M,H"
HH",
so that
.
CH'
Hir:Mjr = M,H'
:
-OH'
CH',
whence
and
If then (a)
= HH' - M,H' Mjr ~ M,H CM, = //Zf M,H', //^' (7J/o = M,H M,H'
'
: . .
AM < AM,,
^GMo.CH'>2MH.M,H'.
to each side,
Adding 2MMq.M,H'
4Cifo
.
CH' +
2il/il/o
il/oiT'
It follows that
<
2 (6'i+ CM,)
Now
2 (6'/
so that, componendo,
MH'
and
:
il/o//'
<
,' + ,\
,,"
-\-
p^'
<
PI'"'
+ p\
^il/j
< AM,
1HH'.CM>2M,H.MH',
and we prove,
in the
same manner
as above,
pp''-^p^<pj>:-^^p;\
F,KN(!THS
OF
.
DIAMK IKKS.
"Jol
And. since
in like
2////'
M/,
>
II
J/. //
',
mauner
Lastly
(6),
H AM > AM^,
etc.
Proposition 146.
[Vll.
Ill a Ityperhola,
41),
50.]
(1) if A A' >pa, then A A" - Pa' < PP" - /, where PP' L amj diameter, and PP" - / moves farther from A i)icreases as also PP" ~ p' > AA" ~ pa AA' but < 2 (^ A'^' ~ p^ AA') (2) if A A' < Pa, then
;
. .
'
also
as
moves
away
2 (^1^'^
:
pa AA').
.
(1)
As
..
usual, A'
^i7 = AH'
^'//'
A'H.AH'
'
:
^if'^ - ^//^
Now
..
iVif
il///'
^ii'
<MH.AH;
:
^^' < MH' + AH' + AH < {MH' + MH) HH' (.1//' + AH HH',
:
i.e.
<
Hence
A'H.MH'
..
^1/^'*
:
il/^"^<
A'H.AH'
:
.1//'^'
^1//*
AA"
if
PP'" - p*<
A A"
yl^""
/;'-
[l*iop
130
(5)]
or
^^'^-_p''<PP"^-yr.
Again,
AM^>AM,
..
Mil,
252
THE
aoxic.s of apollonius,
wo
find
and
so on.
Now,
if
FO
be measured along
PP'
eciual to
}),
PP"^p'=2P0.0P' + 0P";
..
PP'' ~
/ > PP'
But
PP'.uP' = PP"-PP'.PO
[Prop. 12!)]
..
PP"
~ ^/ > ^1^'^ ~ Pa
^^'
but
<
2 (yl^'^
~ p
A A').
(2)
If^lJ.'<jj,
il/i/':^/i'>il/i/:yliJ;
..
J\IB'
^i/" >
MH' + MH
:
^iT' +
AH,
and
^'i/
i.e.
.
MW
^'i/ ^ii'
.
^^'^ ~
AH\
we
PP"~p'<AA"'-pa\
Similarly
and so
on.
if
Lastly,
PP' be produced
to
so that
PO = p,
[Prop. 129]
AA"-pa.AA' = PP"'-p.PP'
= PP'.OP'.
And
PP"
--
if
= PP" - PO' =
>2PP'.0P'
2'. + '
2 (4.4'^
or
>
-^.
^1^1').
.ENOTIIS
OF COX.irOATE
lHAMKTF.ltS.
2.')^
Proposition 147.
[VII. 51.]
In an
(1)
ellipse,
if
PP'
he
that
PP' > p,
AA"--p'>PP"^jf,
and PP'^ - p^ diminishes as
(2)
if PP'
he
PP' <
p,
..
i.e.
<
Therefore, alternately,
A'H.AH' AH"
:
Hence
and
Also,
if ^lil/j
AA"--pa'>PP"'-2}\
>AM,
we
shall
have in the
s;\jik
way
'-
A'H.MH': A'H.Mjr<MH''~
and therefore
(2)
Mil-.
MJI
MJl\
- p^, and so
on.
must
between
if
is
254
Then,
if
M^ corresponds
AM,
we have
MH'>M^H',
..
A'
'
and
.
CM <
CM^;
:
ili,
' > CM
>
CM^
>2CM.HH':2CM^.HH',
i.e.
MH'
MH"
M^H' ~ M^H'\
moves nearer
to
B, being a
maximum when
coincides with B.
camiiripor:
phintkd y
j.
c.
f.
clay, at
the university
press.
SOME PUBLICATIONS OF
Diophantos of Alexandria;
Greek Algebra. By T. Cambridge. Demy 8vo.
L.
7s. (id.
a Study
in
the
History
of
of,
8vo.
by
J.
Gow,
Litt.D.,
Demy
10,*. G(/.
Geometrical Conies.
By
F. S.
Macaulay,
8vo.
4s.
M.A., A.ssistant
Crown
ed.
By H.
F.
Baker, M.A.,
in
Mathematics.
Royal 8vo.
Prcjis.
Treatise on the Theory of Functions of a Complex Vaiiable. By A. R. Fousvrn, Sc.D., F.1\.S., Sadlerian l*rnfe.-^>..r .>f
Pure Mathematics in the University of Cambridge
Trinity College.
anil
Fellow of
Royal 8vo.
21s. Net.
By
E.
W. HonsoN,
Demy
8vo.
A Treatise
By
E.
W. Buown, >..,
Fellow of Christ's College, Caml)ridge, I'rofessor of Applieil Mathematics in Haverford College, Penn.sylvania. Royal 8vo. 15s.
Mathematical Papers
of Cambridge.
ot
Demy
TI. ITT.
4to.
To
completcil in 13 vols.
2.')*.
Volumes
I.
each.
[\1. X.
In the I'reu.
-^
Ml
-"^
DAY
USE
'^
w^
'^^N
DEPT.
VERSITY OF
CUIFORNU
LIBRARY Of THE
UNIVERSITY OF CALIFORNIA
Mm
/
THE
U
:'/
'
GAUFORNm
T,V
~(
^M
Qi^
THE
m
CUIFHiP|l},
.,
UNIVERSITY OF
'LIBRARr-JIf. <1
^^
4
jlN-iVERSITT
OF
CALIFORIN
OV_/f^)
\
/
:^/
QJJ