I

/
A ''

.^amntfoi^toa.ixymK

BOSTON
PUBLIC
LIBRARY
^^""^^^^ i/LrixL*^

PLANE AND SOLID

ANALYTIC GEOMETRY
 -The

THE MACMILLAN COMPANY

NEW YORK BOSTON
• CHICAGO• DALLAS•

ATLANTA SAN FRANCISCO

•

MACMILLAN & CO., Limited

LONDON • BOMBAY CALCUTTA
MELBOURNE

THE MACMILLAN CO. OF CANADA, Ltd.

TORONTO
 PLANE AND SOLID
ANALYTIC GEOMETRY
BY

WILLIAM F. OSGOOD, Ph.D., LL.D.

PERKINS PROFE880R OF MATHEMATICS
IN HARVARD UNIVERSITY

AND

WILLIAM C. GRAUSTEIN, Ph.D.

ASSISTANT PROFESSOR OF MATHEMATICS
IN HARVARD UNIVERSITY

Ncto god?

THE MACMILLAN COMPANY

1921

All rights reserved

 COPTRTGHT, 1920, 1921,

By the MACMILLAN COMPANY.

Set up and electrotyped. Published July, 1921.

AJ,

J. S. Gushing Co. —
Berwick & Smith Co.
Norwood, Mass., U.S.A.
 PREFACE
The object of an elementary college course in Analytic
Geom-
etry is twofold: it is to acquaint the student with n.-w aii.l
interesting and important geometrical material, and to
provide
him with powerful tools for the study, not only of gtH)mctrv
and pure mathematics, but in no less measure of physics in the
broadest sense of the term, including engineering.
To attain this object, the geometrical material should In-
presented in the simplest and most concrete form, with emphasis
on the geometrical content, and illustrated, whenever possihlr,
by its relation to physics. This principle has been ol>servf<i
throughout the book. Thus, in treating the ellipse, the nu'tluHls
actually used in the drafting room for drawing an ellipse from
the data commonly met in descriptive geometry are given a
leading place. The theorem that the tangent makes ecpial
angles with the focal radii proved mechanically a rope which
is :

passes through a pulley has its ends tied at the foci and is drawn
taut by a line fastened to the pulley. Moreover, the meaning
of foci in optics and acoustics is clearly set forth. Again, there
is a chapter on the deformations of an elastic plane under stress,
with indications as to the three-dimensional case (pure strain,

etc.).

The methods of analytic geometry, even in their siniph-st

forms, make severe demands on the student's ability to compre-

hend the reasoning of higher mathematics. Consequently, in
presenting them for the first time, purely algebraic difiieultics,
such as are caused by literal coefficients and long formal compu-
tations, should be avoided. The authors have follow(>tl this
principle consistently, beginning each new subject of the early
chapters with the discussion of a simple s])('eial, but typical,
case, and giving immediately at the close of the
paragraph
vi PREFACE

simple examples of the same sort. They have not, however,

stopped here, but through carefully graded problems, both of
geometric and of analytic character, have led the student to
the more difficult applications of the methods, and collections
of examples at the close of the chapters contain such as put
to the test the initiative and originality of the best students.
As a result of this plan the presentation is extraordinarily
elastic. It is the treatment of any given topic
possible to make
brief without rendering the treatment of later topics unin-
telligible, and thus the instructor can work out a course of any

desired extent. For example, one freshman course at Harvard

devotes about thirty periods to analytic geometry and the ma-
terialcovered consists of the essential parts of the first nine
chapters. Another freshman course gives twice the time to
analytic geometry (the students having already had trigo-
nometry), taking up determinants and the descriptive properties
of the quadric surfaces, and also devoting more time to the less
elementary applications of the methods of analytic geometry.
The advanced courses in the calculus and mechanics require the
material of the later chapters.In fact, a thorough elementary
treatment of the rudiments of Solid ^Analytic Geometry is in-
dispensable for the understanding of standard texts on applied
mathematics. It is true that these texts are chiefly Continental.
But we never have American treatises which are up to the
shall
best scientific standards of the day until the subjects above men-
tioned are available in simply intelligible form for the under-
graduate.
The subject of loci is brought in early through a brief intro-

ductory chapter, and problems in loci are spread throughout

the book. A later chapter is devoted to a careful explanation
of the method of auxiliary variables. There is a chapter on
determinants, with applications both to analytic geometry and
to linear equations. Diameters and poles and polars in the
plane and in space receive a thorough treatment. Cylindrical
and spherical coordinates and quadric surfaces are illumined by
the concept of triply orthogonal systems of surfaces. The re-
 PREFACE vii

duction of the general equation of the second dejcree in .^patr

to normal forms by translations and rotations is sketched and
illustrated by numerical examples.
The question may be asked: In so extensive a treatment of
analytic geometry should not, for example, homogeneous co-
ordinates find a place? The authors believe that the student,
before proceeding to the elaborate methods of modern geometry,
should have a thorough knowledge both of the material and tlio

methods which may fairly be called elementary, and they felt

that a book which, avoiding the conciseness of some of the
current texts and the looseness of others, is clear because it is
rigorous will meet a real need.
This book is designed to be at once an introduction to the
subject and a handbook of the elements. May it serve alike
the needs of the future specialist in geometry, the analyst, the
mathematical physicist, and the engineer.

Harvard University
AprU, 1921
 CONTENTS
PLANE ANALYTIC GEOMETRY
INTRODUCTION
DIRECTED LINE- SEGMENTS. PROJECTIONS
ART. PAOB
1. Directed Line-Segments 2
2. Algebraic Representation of Directed Line-Se^inients 3
3. Projection of a Broken Line 5

CHAPTER I

COORDINATES. CURVES AND EQUATIONS

1. Definition of Rectangular Coordinates . 7
2. Projections of a Directed Line-Segment on the Axes 9
3. Distance between Two Points 10
4. Slope of a Line 12
5. Mid-Point of a Line-Segment Hi
6. Division of a Line-Segment in Any Ratio 17
7. Curve Plotting. Equation of a Curve IS
•»
8. Points of Intersection of Two Curves .

Exercises on Chapter I . . . 24

CHAPTER II

THE STRAIGHT LINE

1. Equation of Line through Two Points . 27
2. One Point and the Slope Given .
30
31
3.

4.

5.
Intercepts
The
....
The General Equation of the First Degree

Intercept Form of the Equation of a Line

33
35
3r>
6. Parallel and Perpendicular Lines
3S
7. Angle between Two Lines .

41
8. Distance of a Point from a Line
ix
X CONTENTS
ART. PAGE
9. Area of a Triangle 43
10.

11.
Lines ..........
General Theory of Parallels and Perpendiculars.

Second Method of Finding Parallels and Perpendiculars

Identical

.
44
47
Exercises on Chapter II 49

CHAPTER III

APPLICATIONS
1. Certain General Methods . 53
2. The Medians of a Triangle 54
3. Continuation. The General Case . . . . .56
4. The /Altitudes of a Triangle 58
5. The Perpendicular Bisectors of the Sides of a Triangle . 60
6. Three Points on a Line 60
Exercises on Chapter III 61

CHAPTER IV

THE CIRCLE
1. Equation of the Circle . 65
A Second Form of the Equation
2.

3.

4.
Tangents ....
Circle through Three Points
66
69
72
Exercises on Chapter IV 74

CHAPTER V
INTRODUCTORY PROBLEMS IN LOCI. SYMMETRY OF
CURVES
1. Locus Problems 79
2. Symmetry 83
Exercises on Chapter V 86

CHAPTER VI
THE PARABOLA
1. Definition . . . . 88
2. Equation of the Parabola . . . . . . .90
3. Tangents .93
 CONTENTS XI

ART.
FAQB
4. Optical Property of the Paral)ola 95
Exercises on Chapter VI 97

CHAPTER VII

THE ELLIPSE
1.

2.

3.
Definition
Geometrical Construction
Equation of the Ellipse
....
....
4. Tangents
5. Optical and Acoustical Meaninu; of the Foci
6.

7.

8.
Slope and Equation of the Tangent
A New Locus Problem ....
Discussion of the Case e<l. The Directrices
9. The Parabola as the Limit of Ellipses .

10. New Geometrical Construction for the Ellipse.

Representation
Exercises on Chapter VII .

CHAPTER VIII

THE HYPERBOLA
1. Definition . ; . . .
124
2. Equation of the Hyperbola .
125

3.

4.

5.
The Asymptotes
Tangents
....
Axes, Eccentricity, Focal Radii .
127
129
133

New Definition. The Directrices 130

6.

7. The Parabola as the Limit of Hyperbolas. Summary . 139

8. Hyperbolas with Foci on the Axis of y. Conjugate Hyperbola' 141

142
9. Parametric Representation
144
10. Conic Sections
145
11. Confocal Conies
148
Exercises on Chapter VIII

CHAPTER IX

CERTAIN GENERAL METHODS

154
1. Tangents
2. Continuation. Implicit Equations
xu CONTENTS
AKT. PAGE
3. The Equation u + kv = . . . . . • .165
4. The Equation vv = 173
5. Tangents with a Given Slope. Discriminant of a Quadratic
Equation 174
6. General Formulas for Tangents with a Given Slope . . 179
7. Tangents to a Conic from an External Point . . . 185
Exercises on Chapter IX 188

CHAPTER X
POLAR COORDINATES
1.
 CONTEXTS Xlll

CHAPTER XIII

A SECOND CHAPTER ON LOCI. AUXILIARY VARIABLES.

INEQUALITIES
ART. PAOB
1. Extension of the Method for the Determination of Ix)ci 261
2. One Auxiliary Variable 203
3. Coordinates of a Point Tracing a Curve, as Auxiliary
Variables 200
4. Other Problems Involving Two or More Auxihary Variables 272
5.

6.
Use
Equation ........
of the Formula

......
Loci of Inequalities
for the Sum of the Roots of a Quadratic
274
277
7. Locus of Two or More Simultaneous Inequalities 279
8. Bisectors of the Angles between Two Lines . 2S1
Exercises on Chapter XIII 283

CHAPTER XIV
DIAMETERS. POLES AND POLARS

1. Diameters of an Elhpse 288

Conjugate Diameters of an Ellipse 290
2.
Diameters of a Hyperbola 293
3. .

Diameters of a Hyperbola 295

4. Conjugate
of a Parabola
299
5. Diameters
299
6. Extremities and Lengths of Conjugate Diameters
302
7. Physical Meaning of Conjugate Diameters
309
8.

9.

10.
Harmonic Division
Polar of a Point
Pole of a Line
....
. . . •
311
315
317
11. Properties of Poles and Polars
320
12. Relative Positions of Pole and Polar
324
13. Construction Problems
320
Exercises on Chapter XIV .

CHAPTER XV
TRANSFORMATIONS OF THE PLANE. STRAIN
330
1. Translations
332
2. Rotations 334
3. Transformations of Simihtude
330
4. Reflections in the Axes
 XIV CONTENTS
ART. PAGE
5. Simple Elongations and Compressions 337
6. The General Afl&ne Transformation 342
7. Factorization of Particular Transformations . . . 3-19

8. Simple Shears 351

9. Second Method of Factorization. Homogeneous Strains 355
Exercises on Chapter XV 358

CHAPTER XVI
DETERMINANTS AND THEIR APPLICATIONS
I. Determinants

2.
1. Simultaneous Linear Equations
Two- and Three-Rowed Determinants
.... .
360
361
3. Determinants of the Fourth and Higher Orders . 364
4.

5.

6.
Evaluation of a Determinant bj^ Minors
Simplified Evaluationby Minors
Fundamental Properties of Determinants
.... 367
370
373
7. Interchanges of Rows and of Columns 376
8. Cramer's Rule 381
9.

10. Homogeneous Linear Equations ....

Three Equations in Two Unknowns. Compatibility 383
387

II. Applications

11. The Straight Line 391

12. The and the Conies
Circle . . . . . . .394
Exercises on Chapter XVI 399

SOLID ANALYTIC GEOMETRY

CHAPTER XVII
PROJECTIONS. COORDINATES
1. Directed Line-Segments 405
2. Projection of a Broken Line . . . ... 405
3. The Angle between Two Directed Lines 406
4.

5. Coordinates ,......
Value of the Projection of a Directed Line-Segment 408
409
6.

7. Distance between Two Points ....

Projections of a Directed Line-Segment on the Axes 413
414

f*?
 CONTENTS XV
ART.
FAOa
8. Mid-Point of a
Line-Segment 415
Division of a Line-Segment in a Given Ratio
9.
Exercises on Chapter XVII .... 117

CHAPTER XVIII
DIRECTION COSINES. DIRECTION COMPONENTS
1. Direction Cosines of a Directed Line .
420
2. Angle between Two Directed Lines 12.')

3. Direction Components of an Undirected Line 427

4. Formulas for the Use of Direction Components 4:i3
5. Line Perpendicular to Two Given Lines 4.30
6. Three Lines Parallel to a Plane .
410
Exercises on Chapter XVIII 441

CHAPTER XIX
THE PLANE
1. Surfaces and Equations 444
2. Plane through a Point with Given Direction of its Normals 447
3. The General Equation of the First Degree . 448
4. Intercepts 4.^)0

5. Intercept Form of the Equation of a Plane . 4.51

6. Plane through Three Points 453

7. Perpendicidar, Parallel, and Identical Planes. Angle between
Two Planes 454
8. Planes Parallel or Perpendicular to a Given Plane 4.5r>

9.

10.
Distance of a Point from a Plane
Point of Intersection of Three Planes
Exercises on Chapter XIX
.... 4r>0

462
405

CHAPTER XX
THE STRAIGHT LINE
1.

2.
Equations of a Curve .....••
....
Line of Intersection of Two Planes
470
17(>

Line through a Point with Given Direction Components 47S

3.
ISl
4. Line through Two Points
5. Line or Plane in Given Relationship to Given Lines or
'^^
Planes
xvi CONTENTS
ART. PAGE
6.

7.

8.
Angle between a Line and a Plane
Point of Intersection of a Line and a Plane
Parametric Representation of a Curve
.... 487
488
490
Exercises on Chapter XX 494

CHAPTER XXI
THE PLANE AND THE STRAIGHT LINE. ADVANCED
METHODS
1. Linear Combination of Two Planes 498
2.

3. Line in a Plane ........

Three Planes through a Line. Three Points on a Line 504
506
4.

5. Two ......
Four Points in a Plane. Four Planes through a Point
Intersecting Lines
507
512
6.

7.
Lines .........
Distance of a Point from a Line. Distance between

Area of a Triangle. Volume of a Tetrahedron

Two
514
516
Exercises on Chapter XXI 519

CHAPTER XXII
SPHERES, CYLINDERS, CONES. SURFACES OF
REVOLUTION
1. Equation of the Sphere 523
2.

3.
4.
General Form of the Equation
Sphere through Four Points
Tangent Plane to a Sphere
....... 524
526
527
5. The Circle . . . .' . 1 . . . .529
6. Cylinders 532
7. Cones 536
8. Surfaces of Revolution 540
Exercises on Chapter XXII 544

CHAPTER XXIII
QUADRIC SURFACES
1. The Ellipsoid 548
2. The Hyperboloids 550
3. The Paraboloids 553
4. Rulings 555
 CONTENTS XMi
ART.
rAQB
5. Parallel Sections .

6. Circular Sections .
r,r,i

7. Tangent Lines and Planes . '.«.>

8. Diameters. Diametral Planes '><i9

9. Poles and Polars .

575
10. One-Dimensional Strains, with Applications 577
Exercises on Chapter XXIII 580

CHAPTER XXIV
SPHERICAL AND CYLINDRICAL COORDINATES.
TRANSFORMATION OF COORDINATES
1.

2.
Spherical Coordinates
Cylindrical Coordinates
.... r,s\

5s7
3.

4. Confocal Quadrics ....

Triply Orthogonal Systems of Surfaces 589
590
5.

6.

7.
Rotation of the Axes ....
Transformation to Parallel Axes .

The General Equation of the Second Degree

592
593
599
Exercises on Chapter XXIV r,04
 PLANE ANALYTIC GEOMETRY
INTRODUCTION
DIRECTED LINE-SEGMENTS. PROJECTIONS
Elementary Geometry, as it is studied in the high school
to-day, had attained its present development at the time when
Greek culture was at its height. The first systematic treat-
ment of the subject which has come down to us was written
by Euclid about 300 b.c.
Algebra, on the other hand, was unknown to the Greeks.
Its beginnings are found among the Hindus, to whom the so-
called Arabic system of numerals may also be due. It came

into Western Europe late, and not till the close of the middle
ages was it carried to the point which is marked by any school
book of to-day that treats this subject.
When scholars had once possessed themselves of these two
subjects —
Geometry and Algebra — the next step was quickly
taken. The renowned philosopher and mathematician, Ren^
Descartes, in his Geometric of 1637, showed how the methods
of algebra could be applied to the study of geometry. He
thus became the founder of Analytic Geometry.*
The "originals" and the locus problems of Elementary
Geometry depend for their solution almost wholly on ingenu-
ity. There are no general methods whereby one can be sure
of solving a new problem of this class. Analytic Geometry,

form of his name,

*Also called Cartesian Geometry, from the Latinized
Cartesius.
1
 '

2 ANALYTIC GEOMETRY

on the other hand, furnishes universal methods for the treat-

ment of such problems moreover, these methods make pos-
;

sible the study of further problems not thought of by the

ancients, but lying at the heart of modern mathematics and
mathematical physics. Indeed, these two great subjects owe
their very existence to the new geometry and the Calculus.
The question of how
make use in geometry of the nega-
to
tive, as well as the positive, numbers is among the first which
must be answered in applying algebra to geometry. The solu-
tion of this problem will become clear in the following
paragraphs.

1. Directed Line-Segments. Let an indefinite straight line,

i, be given, and let two points,A and B, be marked on L.
Then the portion of L which is
bounded by A and B is what is
ACS
1 1 1

called in Plane Geometry a li7ie-

1 ' '
segment, and is written as AB.
Let a third point, C, be marked
—f
A 75
1 '
on L. Then three cases arise,
as indicated in the figure. Cor-
responding to these three cases we have :

(a) AB + BC = AC;
(b) Ab'-CB = AC',
(c) CB-AB= ca.
Three other cases will arise if the original points A and B are
taken in the opposite order on the line. Let the student
write down the three corresponding equations.
A unification of all these cases can be effected by means of
an extension of the concept of a line-segment. We no longer
consider the line-segments AB smd' BA as identical, but we
distinguish between them by giving each a direction or sense.
Thus, AB shall be directed from A to B and BA shall be
directed from B to A, i.e. oppositely to AB. These directed
 :

INTRODUCTION 3

line-segments we denote by AB and BA, to distinguish tlmm

from the ordinary, or undirected, line-segments.
We may, for the moment, interpret the directed line ;>*g-
ment AB as the act of walking from A in B; then BA rej)re-
sents the act of walking from B to A. With this in mind, let
us return to Fig. 1 and consider the directed line-segments
AB, BG, and AC. We have, in all three cases represented by
Fig. 1, and also in the other three

AB + BC=AC,
since walking from A to B and then walking from B to C is

equivalent, with reference to the point reached, to walking

from A to C.
Accordingly, we unify all six cases by defining, as the sum
of the directed line-segments AB and BC, the directed line-
segment AC:
(1) AB-^BC=M\
From this definition it follows that, if A, B, C, and D are

any four points of L,

(2)
AB-\-BC+CD = AD.
For, by (1), the sum of the first two terms in (2) is AC, and,
by the definition, the sum of AC and CD is AD.
Similarly, if the points M, M^, M., • • •, 3f„_i, N are any points
of L, we have

(3) MMi + M^2 + • • •

+ mZX-i + 3/„_i.v = yfy.
line or on
Given two directed line-segments on the same
lines, we say that these two directed line-segments
two parallel
are equal, if they have equal lengths
and the same direction or
sense.

Line-Segments. On
Algebraic Representation of Directed
2.
opposite directions or senses be
the line L let one of the two
positive direction or sense
chosen arbitrarily and defined as the
called the negative direction or sense.
of L and let the other be
•
 ;

4 ANALYTIC GEOMETRY

A directed line-segment AB, which lies on L, is then called

positive, if its sense is the same as the positive sense of X, and
negative, if its sense is the same as the negative sense of L.
To such a directed line-segment AB we assign a number,
which we shall also represent by AB, as follows. If I is the
length of the ordinary line-segment AB, then
AB = I, if AB is a positive line-segment
AB = — l, if AB is a negative line-segment.

If AB = I, then BA = -l', and if AB = — then BA- I, I.

In either case

(1) AB-{-BA = or AB=:-BA.

Since the act of walking from ^ to 5 is nullified by the act
of walking from B to A, we might have arrived at equations
(1) from consideration of the line-segments themselves, instead
of by use of the numbers which represent them.
It is easy to verify the fact that equations (1), (2), and (3)
of the preceding paragraph, which relate to directed line-seg-
ments, hold for the corresponding numbers. Consequently,
no error or confusion arises from using the same notation AB
for both the directed line-segment and the number correspond-
ing to it. We shall, however, adopt a still simpler notation,
dropping the dash altogether and writing henceforth AB to
denote, not merely the directed line-segment or the number
corresponding to it, but also the line-segment itself, stating
explicitly what is meant, unless the meaning is clear from the
context.

Absolute Value. It is often convenient to be able to express

merely the le7igth of a directed line-segment, AB. The nota-
tion for this length is |
AB \
; read :
" the absolute value of
AB.''
The numerical, or absolute, value of a number, a, is
denoted in the same way : \a\. Thus, |
— 3 = 3.
1 Of course,
13| = 3.
 :

INTRODUCTION

Projection of a Broken Line.

3. By the projection of a i)()int
P on a line L is meant the foot, M, of the perjitMidicuhir
dropped from P on L. If P lies on L, it is its own jirojectiou
on L.
Let PQ be any directed line-segment, and let L be an arbi-
trary line. Let 31 and N
be respectively the projections of
P and Q on L. The projection
of the directed line-segment PQ
on L shall be defined as the di-
rected line-segment MN, or the
number which represents MN al-

gebraically. Since MN= — NM,

it follows that
Proj. PQ = -Pioj. QP.

If PQ on a line perpendicular to L, the points

lies and M
N coincide, and we say that the projection oi PQ on L is MN
zero. Such a directed line-segment 3fX, whose end-points are

identical, we may call a nil-segment; to it corresponds the

number zero. It is evident that in taking the sum of a num-
ber of directed line-segments, any of them which are nil-
segments may be disregarded, just as, in taking the sum of a
set of numbers, any of them
which are zero may be disre-

garded.
Consider an arbitrary
broken line PP^P. P„-iQ. •

By its projection on L is

ill M^ ilf il/3 n-2

M^.iN meant the sum of the pro-
J

Fig. 3 jections of the directed line-

segments PPi, P1P2, • • •, Pn-iQ, or

3f3f, + 3f,3f2 -f-

• • • + 3f„_rN.
the same value as 3fN', the
projection on L
This sum has
of the directed line-segment PQ ;
cf. § 1, (3)

JOfi 4- 3Ii3L -\- • + M„_,y= 3fN.

 : :

6 ANALYTIC GEOMETRY

Hence the theorem

Theorem 1. The sum of the projections on L of the segments

PPi, PiP-i, •
-, Pn-\Q 0/ ^ broken line joining P with Q is equal
to the projection on L of the directed line-segment PQ.

If, secondly, the same points P and Q be joined by another

broken line, PPlPl, ••• PL-iQ, the projection of the latter on
L will also be equal to MN:
MM[ + M[M^2-\- " + Ml_^N= MN.
Hence the theorem
Theorem 2. Given two broken lines having the same extremi-
ties '

PP,P2 P^.iQ and PP[Pl'PL_,Q.

Let L be an arbitrary
straight line. Then the sum of the pro-
jections on L
of the segments PP^, P1P2, •, Pn-^Q, of which the
• •

first broken line is made up, is equal to the corresponding sum

for the second broken line.

 : :

CHAPTER I

COORDINATES. CURVES AND EQUATIONS

1. Definition of Rectangular Coordinates. Let a plane be
given, in which it is desired to consider points and curves.
Through a point in this plane take two indehnite straight
lines at right angles to each other, and choose on each line a
positive sense.
Let P be any point of the plane. Consider the directed
line-segment OP. Let its projections on the two directed lines
through be OM and ON. The numbers which represent
algebraically these projections,
that is, the lengths of OM and
ON taken with the proper signs
(cf. Introduction, § 2), are called
the coordinates of P. We shall
denote them by x and y

X = OM, y = ONy
and write them in parentheses
{x, y). The first number, x, is Fig. 1

known as the x-co6rdinate, or

abscissa, of P; the second, y, as the y-coordinate, or ordinate,
of P.
The point called the origiii of coordinates.
is The directed
lines throughare called the axes of coordinates or the coordi-
nate axes the one, the axis of x the other, the axis of y.
;
;
It

is customary to take the coordinate axes as

in Fig. 1, the axis

of X being positive from left to right, and the axis of y,

posi-

But, of course, the opposite sense

tive from below upward.
on one or both axes may be taken as positive, and an oblique
7
8 ANALYTIC GEOMETRY

position of the axes which conforms to the definition is legiti-

mate, the essential thing being solely that the axes be taken
perpendiculm' to each other.
Every point, P, in the plane has definite coordinates, (x, y).
Conversely, to any pair of numbers, x and ?/, corresponds a
point P whose coordinates are {x, y). This point can be con-
structed by laying off OM=x on the axis of x, erecting a per-
pendicular at M and then laying off
to that axis, = y. MP
We might equally well have begun by laying off 0N= y on
the axis of y (cf. Fig. 1), and then erected a perpendicular to

y that axis at and laid off on it N

NP = X. It shall be understood that
the positive sense on any line parallel
to one of the coordinate axes, such as
—X
M the perpendicular to the axis of x at
M, shall be the same as the positive
Fig. 2
sense of that axis. For other lines
of the plane there is no general principle governing the choice
of the positive sense.
The coordinates of the origin are (0, 0).Every point on
the axis of x has as its ordinate, and these are the only
points of the plane for which this is true. Hence the axis of
X is represented by the equation

y = 0, (axis of x).

Similarly, the axis of y is represented by the equation

x = 0, (axis of y).

The axes divide the plane into four regions, called quadrants.
The Jirst quadrant is the region included between the positive
axis of Xand the positive axis of y the second quadrant, the
;

region between the positive axis of y and the negative axis of

X etc. It is clear that the coordinates of a point in the first
;

quadrant are both positive that a point of the second quad-

;

rant has its abscissa negative and its ordinate positive etc. ;

The system of coordinates just described is known as a sys-

tem of rectangular or Cartesian coordinates.
 ; ; ; : ;

COORDINATES. CURVES AND EQUATION'S 9

EXERCISES
The student should provide himself with some squared
paper for working these and many of the later exercises in
this book. Paper ruled to centimeters and sulxlivided to mil-
limeters is preferable.

1. Plot the following points, taking 1 cm. as the unit:

(«) (0, 1) W (1,0); (c) (1,1);

W (1,
- 1) W (-1,-1); (/) (2, - 3)
(9) (0,-2i); {h) (-3.7,0); (0 (-U, -13);
(J) (-4,3.2); {k) (3.24, - 0.87) (0 (-1,1)-
2. Determine the coordinates of the point P in Fig. 1 when
1 in. is taken as the unit of length ; also when 1 cm. is the
unit of length.

3. The same for the point marked by the period in

" Fig. 1."

/ 2. Projections of a Directed Line-Segment on the Axes. Let

^^T^ith the coordinates {x^, ?/i), and P^ : (a^, 1/2)* be any two
points of the plane. Con-
sider the directed line-seg- £-'(^i'2/i)
ment PiP^. It is required
to find its projections on the
axes. ''i?.;(T..,y.,)

To do this, draw the

broken line P^OP^. By Ii^- / ^''

troduction, § 3, Th. 1, the

of this broken Fio. 3
projections
line on the axes are the
directed line-segment 7^iA. Hence,
same as those of the
taking first the projections on the axis of x, we have

Proj. P1P2 = Pi'oJ- ^1^ + ^I'^J- ^^'2

= -Proj. OP, + Proj. OP..

P2 as an ab-
We shall frequently use this shorter notation,
(-Cs, i/2),
* :

breviation for " P2, with the coordinates

(xo, 1/2)."
10 ANALYTIC GEOMETRY

But the terms in the last expression are by definition —Xi

and X2. So
(1) Proj. P1P2 on a^axis = X2 — Xi.
Similarly,

(2) Proj. P1P2 on ^/-axis = 2/2

— 2/i-

The projections of P1P2 on two lines drawn parallel to the

axes are obviously given by the same expressions.

EXERCISES
1. when Pi is the point (a) of Ex. 1, § 1, and Pg
Plot P1P2
is (b). Determine the projections from the foregoing formulas,
and verify directly from the figure.
2. The same, when
i) Pi is (e) and P2 is (/) ;

ii) Pi is (c) and Pg is (d) ;

iii) Pi is (i) and P2 is (Z).

3. Distance between Two Let the points be Pi,

Points.
mth the coordinates (xi, and P2 (0^2, 2/2)- Through P,
2/i)> :

draw a line parallel to the axis of x and through P2, a line

parallel to the ^xis of y let Q >

\y ^(^2' 2^2), denote the point of intersection of

^y^
^^ these lines. Then, by the Pytha-
I
Q gorean Theorem,
^^•'^""^''^^^
X (1) P^P2^ = P.Q^ + QPi\
«^
Fig. 4

(2) D^ = {x2-x,y+{y2-y,)\
where D denotes the distance between P^ and P2. Hence •>

(3) D=^{x2-x,f+{y2-y,f.
In the foregoing analysis, we have used P^Q (and similarly,
QP2) in two senses, namely, i) as the length of the ordinary
line-segment P^Q of Elementary Geometry ii) as the algebraic ;
 :

COORDINATES. CURVES AND EQUATIONS 11

expression x^-x^ for the projection P,Q of the directed line

segment P^P^ on a parallel to the axis of x. Since, however,
these two numbers differ at most in sign, tln'ir squares are
equal, and hence equation (2) is equivalent to equation
(1).
In particular, P^P^ may be parallel to an axis, e. j/. the axis
of X. Here, y^ = 2/1, and (3) becomes

D = V(a;2 - x,f.
The student must not, however, hastily infer that

i) = iC2 — Xi.
It may be that x^ — x^ is negative, and then *

D = — {X2 — x^.
A single formula which covers both cases can be written in
terms of the absolute value (cf. Introduction, § 2) as follows

(4) .
D=\x2 — Xi\.

EXERCISES
J 1. Find the distances between the following pairs of points,
expressing the result correct to three significant hgures. Draw
a figure each time, showing the points and the line connecting
them, and verify the result by actual measurement.

(a) (2, 1) and (- 2, - 2). (6) (- 7, 6) and (2, - 3).

(c) (Ic:, 5) and (-2, 5). (d) (7, 3) and (12, 3).

(e) (4, 8) and (4, - 8). (/) (- 1, 2) and (- 1, 6).

J 2. Find the lengths of the sides of the triangle whose ver-

tices are the points (,— 2, 3), (—2, — 1), (4, — 1).

^ 3. How far are the vertices of the triangle in question 2

from the origin ?
* There is no contradiction here, or conflict with the ordinary laws of

algebra. For, the V-sign always calls for the positive stiuan^ root,
that —
being the definition of the symbol, —and we must see to it in any
given

case that we fulfill the contract.

 :

12 ANALYTIC GEOMETRY
•^
4. Find the lengths of the diagonals of the convex quadri-
lateral whose vertices are the points (4, 1), (1, 3), (—3, 1),
(-2,-1).

4. Slope of a Line. By the slope, X, of a line is meant the

trigonometric tangent of the angle, 6, which the line makes
with the positive axis of x

(1) X = tan 6.
To find the slope of the line,
with the coordinates (xi, ?/i),
let Pj,
and P2 (^2, 2/2) ^^ the extremities
y^ :

of any directed line-segment P1P2

Fig. 5
on the line. Then

(2) tan =
PlQ X^ — Xi
or

(3)
X2 — »cj^

If, instead of P1P2, we had taken its opposite, ^2^1? we

should have obtained for X the value (pi — 2/2) /(-'^i — ^2)- ^^^
this is equal to the value of X given by (3). Thus, A is the
same, whether the line is directed in the one sense or in the
opposite sense. Hence we think of X as the slope of the line
without regard to sense.

Variation of the Slope. Consider the slopes, X, of different

lines, L, through a given point, P. When is parallel to the L
axis of X, X has the value zero. AVhen L rotates as shown in
the figure, X becomes positive and increases steadily in value.
As L approaches the vertical line L', X becomes very large,

increasing without limit.

When L passes beyond L', X changes sign, being still nu-
merically large. As L continues to rotate, X increases alge-
braically through negative values. Finally, when L has again
become parallel to the axis of x, X has increased algebraically
through all negative values and becomes again zero.
 COORDINATES. CURVES AND EQUATIONS 13

When L is in the position of L', is 90° and tan ^ =X is

undefined, that is, has no value. Hence V has no slope. One

often sees the expression : tan 90° = oo, and, in accordance with
it, one might write here, A = go.This does not mean that IJ has
a slope, which is infinite, for " infinity " is not a number. It
is merely a brief and
symbolic way of describ-
ing the behavior of X for
a line />, near to, hut not
coincident ivith L' ;
it

says that for such a line

A is numerically very
large ; and further that,
when the line L ap-
proaches L' as its limit,
A increases numerically
without limit, — that is,
Fig. 6
increases numerically be-
yond any preassigned number, as 10,000,000 or 10,000,000 !, and
stays numerically above it.

The Angle In measuring the angle from one line to an-

6.

other, it is essential, first of all, to agree on which direction

of rotation shall be considered as positive. We shall take
always as the positive direction of rotation that from the posi-
tive axis of X to the positive axis of ?/ so that the angle from ;

the positive axis of x to the positive axis oi y is -f 90°, and

not - 90°.
The complete definition of 6 is, then, as follows :
Tlie slope-

angle 6 of a line is the angle from the positive axis of x to the

direction of the line. There are in general two positive values

for less than 360° if the smaller of them is denoted by 0, the
;

other is 180° -f 0. Which of these angles is chosen is imma-

terial, since tan (180° -f 6) =,tan 6 this result is in agreement ;

with the previous one, to the effect that the slope pertains to
the undirected line without regard to a sense on it.
14 ANALYTIC GEOMETRY

The student should now draw a variety of lines, indicating

for each the angle and assure himself that the deduction of
^,

formula (3) holds, not merely when the quantities X2 — x^ and

2^2 — 2/1 S'l^e positive, but also when one or both are negative.

Right-Haiided and Left-Handed Coordinate Systems. For the

choice of axes in Fig. 1, the positive direction for angles is

the counter-clockwise direction. But for

^ X such a choice as is indicated in the
present figure, —a choice equally legiti-
mate, — it is the clockwise sense which is

positive.

y
The above formulas apply to either
Fig. 7 system of axes. The first system is
called a right-handed system ; the other,
a left-handed system. We shall ordinarily use a right-handed
system.

Problem. To draw a line through a given point having a

given slope. In practice, this problem is usually to be solved
on squared paper. The solution will be sufficiently clearly
indicated by an example or two.

Example 1. To draw a line through the point (—2, 3) hav-

ing the slope — 4.
Proceed along the parallel to the a^-axis through the given
point by any convenient distance, as 1 unit, toward the left.*
Then go up the line through this point, parallel to the ?/-axis,

by 4 times the former distance, here, 4 units. —

Thus, a sec-
ond point on the desired line is determined, and the line can
now be drawn with a ruler.
If the given point lay near the edge of the paper, so that
the above construction is inconvenient, it will do just as well
to proceedfrom the first point toward the right by 1 unit, and
then down by four units.

* The student will follow these coiistructions step by step on a piece

of squared paper.
 COORDINATES. CURVES AND EQUATIONS 15

Example 2. To draw a Hue through the poiut (1.32, 2.78)

haviug the slope .6541.
Here, it is clear that we cauuot draw accurately enough to
be able to use the last significant figiJre of the given slope.
Open the compasses to span 10 cm. (if the squared paper is
ruled to cm.) and lay off a distance of 10 cm. to the right on a
parallel to the a'-axis through the given point. This parallel
need not actually be drawn. Its intersection, Q, with the cir-
I cular arc is all that counts, and this point, Q, can be estimated
and marked. Its distance above the axis of x will be 2 cm.
and 7.8 mm. The error of drawing will be of the ^rder of the
last significant figure, namely, more than yL mm. and less than
.5 mm.
Next, open the compasses to span 6 cm. and 5.4 mm. Put
the point of the compasses on Q, and lay off the above dis-
tance, 6.54 cm., on a parallel through Q to the y-axis and
above Q. The point R, thus found, will be a second point on
the desired line, which now can be drawn.

EXERCISES
fljThe points Pj, P2, P3, with the coordinates (2, 5), (7, 3),

(—0, on a line. Show that the value for

7) respectively, lie
the slope of the line as given by equation (3) is the same, no
matter which two of the three points are used in obtaining it.
2. 'Find the slopes of the sides of the triangle of Ex. 2, § 3.

3. Find the angles which the sides of that triangle make

with the axes, and hence determine the angles of the triangle.
^ 4. Show that the points (- 2, - 3), (5, - 4), (4, 1), (- 3, 2)

are the vertices of a parallelogram.

5. Draw a line through the point (1, — 2) having the
slope 3.

--"Q. Draw a line through the point (—2, — 1) having the

slope — 1^.
7. Draw a line through the point (—1.32, 0.14) having the
slope - .2688.
 : :

ANALYTIC GEOMETRY

% - -«/. Mid-Point of a Line-Segment. Let Pi, with the coordi-

y- nates (xi, 2/1)5 and P2 (x2, y-y) be the extremities of a line-seg-
:

ment. It is desired to find the

f|-(^2'2/2) coordinates of the point P which
bisects P1P2.
\
Let the coordinates of P be
{x, y). It is evident that the
directed line-segment PiP is

equal to the directed line-seg-

Fig. 8 ment PP2. Hence the projec-
'
tion of P^P on the axis of or a;,

X — oji, must equal the projection of PP2 on that axis, or ^2 — a?

*C ~~~ Jb't - •- •^O ^~~ w*
Hence
Xi 4- X9

Similar considerations apply to the projections on the axis

of y, and consequently

2
We have thus obtained the following result : The coordinates
(cc, y) of the point P ichich bisects the line-segment P1P2 are given
by the equations

(1) X _^l+^2
^ 2

EXERCISES
1. Determine the coordinates of the mid-point of each of
the line-segments given by the pairs of points in Ex. 1, § 3.
Draw figures and check your answers.
"^
-3v. Find the mid-points of the sides of the triangle mentioned
in Ex. 2, § 3, and check by a figure.
3. Determine the coordinates of the mid-point of the line
joining the points (a + b, a) and (a — b, b).
4. Show that the diagonals of the parallelogram of Ex. 4,
§ 4 bisect each other.
 :

COORDINATES. CURVES AND EQUATIONS 17

6. Division of a Line-Segment in Any Ratio.* Let it be re-

quired to Und the coordinates (x, y) of the point P which

divides the line-segment PxP^ in an arbitrary ratio, mi/m2 : t

PP2 "TH

Obviously the projections of P^P and PP2 on the axis of x

must be in the same ratio, m^/rthi, and hence
X — Xi _mx
X2 — X m2
On solving this equation for x, it is found that

^ _ m^x^ + Wia?2
m2 4- Wi
Similar considerations, applied to the projections on the
axis ofy, lead to the corresponding formula for y, and
thus
the coordinates of P are shown to be the following

mo -f- Wi ' wi2 + iih

If nil and m2 are equal, these formulas reduce to those of

§5.
External Division. It is also possible to find a point on P
the indefinite straight line through P^ and P2 and lying outside
the line-segment P1P2, which makes

PiP^mi
PoP~ m^'
where m^ and mg are any two unequal positive numbers. Here,

x^ — X _mi
X2 — X m<i
* This paragraph may well be omitted^ till the results are needed in
later work.
I The given numbers mi and m^ may be precisely the lengths PiP
and PP2 but in general they, are merely proportional respectively to
;

or
them, i.e. they are these lengths, each multiplied by the same positive
negative number.
 :

18 ANALYTIC GEOMETRY

On solving this equation for x and the corresponding one for

y, we find, as the coordinates of the point P, the following

^_ m2Xi-miX2 moVi — miT/g .

m2 — wii ' m2 — mi
The point P is here said to divide the line P1P2 externally in
the ratio mi/m2 ; and, in distinction, the division in the earlier
case is called internal division. Both formulas, (1) and (2),
can be written in the form (1) if one cares to consider external
division as represented by a negative ratio, m^jm^^ where, then,
one of the numbers mj, m^ is positive, the other, negative.

EXERCISES
1. Find the coordinates of the point on the line-segment
joining (—1, 2) with (5, — 4) which is twice as far from the
first point as from the second. Draw the figure accurately
and verify.

2. rind the point on the line through the points given in

the preceding problem, which is outside of the line-segment
bounded by them and is twice as far from the first point as
from the second.
3. Find the point which divides internally the line-segment

bounded by the points (3, 8) and (— 6, 2) in the ratio 1 5, and :

lies nearer the first of these points.

4. The same question for external division.

7.Curve Plotting. Equation of a Curve. Since the subject

of graphs is now very generally taught in the school course

in Algebra, most students will already have met some of the

topics taken up on the foregoing pages, and moreover they
will have plotted numerous simple curves on squared paper
from given equations. Thus, in particular, they will be famil-
iar with the fact that all the points whose coordinates satisfy
a linear equation^ i.e. an equation of the first degree, like

(1) 2x-^y-l = 0,
 ;; ,

COORDINATES. CURVES AND EQUATIONS 19

lie on a straight line, though they may never have seen a

formal proof.
A number of points, whose coordinates satisfy equation (1),
can be determined by giving to x simple values, computing
the corresponding values of y from (1), and then plotting the
points (x, y). Thus
if = 0,
X = — 1, and the point is
?/ (0, — J)
if = 1, y = ^, and the point is
a; (1, |)

if = 2, y = 1, and the point is

a; (2, 1)

if X = — 1, y = — 1, and the point is (—1, — 1);

etc.

Of course, if it is known that (1) represents a straight line,

— i.e. that all the points whose
coordinates satisfy (1) lie on a
straight line, — it is sufficient
to determine two points as above,
and then to draw the line
through them.
This process of determining a
large number of points whose
coordinates satisfy a given equa- Fig. 9
tion and then passing a smooth
curve through them is known as " plotting a curve * from its

equation."

TJie mathematical curve\ defined by an equation in x and

y consists of all those points and only those points ichose coordi-
nates, ichen substituted for x and y in the equation, satisfy it.

Suppose, for example, that the equation is

(2) y = x\
The point (2, 4) lies on the curve defined by (2), because, when
* In Analytic Geometry the term curve includes straight lines as well

as crooked curves.
t This cui've is sometimes called the locus of the equation.
20 ANALYTIC GEOMETRY

X is set equal to 2 and y is set equal to 4 in (2), the resulting

equation,
4 = 4,
is true. We say, equation (2) is satisfied by the coordinates of
the point (2, 4), or that the point (2, 4) lies on the curve (2)
On the other hand, the point (—1, 2), for example, does not
lieon the curve defined by (2). For, if we set a; =—1 and
y = 2, equation (2) becomes
2 = 1.
This is not a true equation; i.e. equation (2) is not satisfied
by the coordinates of the point (—1, 2), and so this point does

not lie on the curve (2).

Equation of a Curve. A curve may be determined by simple

geometric conditions as, for example, that all of its points
;

be at a distance of 2 units from the origin. This is a circle

with its center at the origin and having a radius of length 2.

It is easy to state analytically the condition which the coor-

dinates of any point (x, y) on the circle must satisfy. Since
by § 3 the distance of any point {x, y) from the origin is

Va^^ '+ y\

the condition that (x, y) be a point of the curve is clearly this,

that
Voj^ + 2/- = 2,
or that
(3) a;2 + 2/2 = 4.
Equation (3) is called the equation of the curve in question.

Tlie equation of a curve is an equation in x and y ivhich is

satisfied by the coordinates of every point of the curve, and by

the coordinates of no other point.

In this book we shall be engaged for the most part in find-

ing the equations which represent the simpler and mote im-
portant curves, and in discovering and proving, from these
equations, properties of the curves.
 ^

COORDINATES. CURVES AND EQUATIONS 21

Nevertheless, the student should at the outset have clearly

in mind the fact that any equation between x and y, like

y — ^i y = log X, y = sin x,

represents a perfectly definite mathematical curve, which he

can plot on paper. Moreover, he is in a position to determine
whether, in the case of a chosen one of these curves, a given
point lies on it. He will find it desirable to plot afresh a few
simple curves, and to test his understanding of other matters
taken up in this paragraph by answering the questions in the
following exercises.

EXERCISES
A. What does each of the following equations represent?
Draw a graph in each case.

y(d) x = 2', ^^
(c) x-y = 0- (e) 2x-3y-^6 = 0',
(b) 22/ +3= ;
^
(d) 2x + 5y = 0; ^f) 5x + Sy - 4 = 0.
Plot the following curves on squared paper.
2. .
y = x^.
Take 2 cm. or 1 in. as the unit of length. Use a table of
squares.

3. y'^- — x.
Take the same unit as in question 2 and use a table of square
roots.

4. Show that, when one of the curves of Exs. 2 and 3 has

been plotted from the tables, the other can be plotted from
the first without the tables.

Work the corresponding exercises for the following curves.

^^5. y = x^. 6. y = y/x. Ji. y"- = ofi. 8. y^ = x\

^ Plot the
9. curve
y = logiox
from a table of logarithms for values of x from 1 to 10, taking
1 cm. as the unit. *.——
 ;

22 ANALYTIC GEOMETRY

10. Which of the straight lines of Ex. 1 go through the

origin ?

11. Show that the curve

(tt) y = smx
goes through the origin.
Do the curves

(p) y — tan x, (c) y = cos »,

go through the origin ?

12. Do the following points lie on the curve

xy = 1?
(a) (-1,-1); (&)(-l,l); W (|, f)

W (-1,-1); W (i, -2); (/) (0,1)-

13. Find the equations of the following curves.

(a) The line parallel to the axis of x and 8 units above it.

(6) The line parallel to the axis of y and 1| units to the

left of it.

(c) The line bisecting the angle between the positive axis
of y and the negative axis of x.

(d) The circle, center in the origin, radius p.

(e) The circle, center in the -point (1, 2), radius 3.

Ans. (x-iy-\-(y-2y. = 9.

Points of Intersection of Two Curves.

8. Consider, for ex-
ample, the problem of finding the point of intersection of the

lines
L: 2x-3y = 4.,

U: 3.r + 42/ = -ll.

Let (a:i, y^ be the coordinates of this unknown point, Pi-

Any point P, with the coordinates (x, y), which lies on L, has
its x and y satisfying the first of the above equations.
Hence,
in particular, since Pi lies on L, x^ and y^ must satisfy that

equation, or

(1) .
2a^i-3j/i = 4.
 :

COORDINATES. CURVES AND EQUATIONS 23

Similarly, a point P: (x, y), which lies on L', has its x and
y satisfying the second of the above equations. Hence, in
particular, since Pi lies on L', «i and 2/1 must satisfy that equa^
tion, or

(2) 3x,+ 4.y, = -ll.

Thus it appears that the two unknown quantities, x^ and y^,
satisfy the two simultaneous equations, (1) and (2). Hence
these equations are to be solved as simultaneous by the
methods of Algebra.
2aJi-3?/i= 4, 4

To do this, eliminate ?/i

by multiplying the first equation
through by 4, the second by 3, and then adding

llx^ = — Vi, or iCi = — 1.

On substituting this value of x'l in either equation (1) or (2),

the value of ?/i is found to be: y^ = -2. Hence P, has the

coordinates (—1,-2).
The equations (1) and (2) are the same, except for the sub-
scripts, as the equations of the given lines, L and L'. Hence
we may say : To find the coordinates of the point of intersection
latter as simul-
of two lines given by their equations, solve the
taneous equations in the unknown quantities, x and y, by the
methods of Elementary Algebra.
The generalization to the case of any two curves given by
their equations is obvious. Ths equations are to be regarded
as simidtaneous equations between the unknown quantities, x and

y, and solved as such.

The student should observe that the letters "a:"and"2/"
have meanings when they appear as the co-
totally different
ordinates of a variable point in the equation of a curve, and when

they represent unknown quantities in a pair of simultaneous

equations. In the first case, they are variables, and a pair of

values, (x, y), which satisfy equation L will not, in general,
the
satisfy L' In the second case, x and y are constants,
 24 ANALYTIC GEOMETRY

coordinates of a siyigle point, or of several points ; but of

isolated and not variable points.

EXERCISES
y Determine the points of intersection of the following curves.
Check your results by plotting the curves and reading off as
accurately as possible the coordinates of the points of
intersection.

1. The straight lines (a) and {d) of Ex. 1, § 7.

2. The straight lines (c) and (e) of Ex. 1, § 7.

3. The straight lines (e) and (/) of Ex. 1, § 7.

f,
G
V^
f

I
2/2= Ax,
a; + ^ = 3. Ans. (1, 2), (9, - 6).

'

---A
'
\xy = 6. \x + y = 0.
\x^+y^ = 25, g \f~ + 6x = 0,
[4a:'^ + 36?/'^ = 144. "
\2x-\-y = 7.

g I
+ y^ = 2,
aj2

jic?/ = l. Ans. (1,1), (—1,-1).

10. Show that the curves

^ " y = ^ogio^, x-\-y = l,

intersect in the point (1, 0).

/ 11^ Show that the curves

^-^ = 25, - = 0,
ic^ + 2/2 3 a; 4 2/

intersect in the point (4, 3), and also in (— 4, — 3).

EXERCISES ON CHAPTER I

V
1. .Show that the points (2, 0), (0, 2), (1 + V3, 1 + V3) are
/he vertices of an equilateral triangle.
( 2.'; Prove that the triangle with vertices in the points (1, 8),

(3, 2), (9, 4) is an isosceles right triangle.

 :

COORDINATES. CURVES AND EQUATIONS 25

3. Show that the points (-1, 2), (4, 10), (2,3), and
(— 3, —5) are the vertices of a parallelogram.

4. Given the points A, B, (7 with coordinates (—7, —2),

(- V-j ^)> (5) 3)- % proving that

show that the three points lie on a line.

( 5. /Show that the three points of the previous problem lie

on-^ine by proving that AB and AC have the same slope.

V^ 6. Prove that the two points (5, 3) and (— 10, — 6) lie on

a line with the origin.
7. Prove that the two points (x^, t/i), (.^2, 2/2) ^ie on a line

with the origin when, and only when, their coordinates are
proportional
^1 '•

yi = ^2- Hi-
^ r 8. ^Determine the point on the axis of x which is equidis-
^( t^irTirom the two points (3, 4), (—2, 6).

9. If (3, 2) and ( — 3, 2) are two vertices of an equilateral

.ngle which contains within
y/\triangle withir it the origin, what are the co
ordinates of the third vertex ?

X^IO. If (3, —1), (— -ij —3), (1, 5) are th'ree vertices of a

' parallelogram and the fourth lies in the first quadrant, find
the coordinates of the fourth. Ans. (8, 7).

11. If P is the mid-point of the segment PxP^-, and P and

Pi have coordinates (8, 17), (—0, — 3) respectively, what are
the coordinates of P2 ?

12. If P divides the segment P1P2 in the ratio 2:1, and Pi

and P have coordinates (3, 8) and (1, 12) respectively, deter-
mine the coordinates of P2. Ans. (0, 14).

13. Find the ratio in which the point B of Ex. 4 divides

the segment AC of. that exercise. Ans. 2 : 3.

14. A point with the abscissa 6 lies on^the line joining the
two points (2, 5), (8, 2). Find its ordinate.
 26 ANALYTIC GEOMETRY

Suggestion. Determine the ratio in which the point divides

the line-segment between the two given points.
15. Prove that the sum of the squares of the distances of
any point in the plane of a given rectangle to two opposite
vertices equals the sum of the squares of the distances from it
to the two other vertices.
Suggestion. Choose the axes of coordinates skillfully.

16. If D is the mid-point of the side BG of a triangle ABC^

prove that
AB"- + AC = 2 AD-" 4- 2 BD\
17. Show that the lines joining the mid-points of opposite

X.
\ sides of a quadrilateral bisect each other.
18. Prove that the lines joining the mid-points of adjacent
sides of a quadrilateral form a parallelogram.
19. Prove that, if the diagonals of a parallelogram are equal,
the parallelogram is a rectangle.
^20^ If two medians of a triangle are equal, show that the
triangle is isosceles. fi ;
^ ryC

>.
 CHAPTER II

THE STRAIGHT LINE

i

1. Eftitfrnon of Line through Two Points. Let Pj (x^, y{) :

and-i2 : (^'2? 2/2) ^^ two given points, and let it be required to

find the equation of the line through
them.
The slope of the line, by Ch. I,

§4, is

?/2 - Vl
Xo Xi
FlQ
Let P, with the coordinates (x, y),
be any point on the line other than P^. Then the slope of

the line is also given by

y-Vi .

X — Xi .

Hence

^
X — Xi X2 — Xi
Conversely, if P : (x, y) is any point whose coordinates
satisfy equation (1), this equation then says that the slope of
the line PiP is the same as the slope of the line P1P2 and
hence that P lies on the line P1P2.
more desirable form of equation (1) is obtained by multi-
A
plying each side by {x - x{)/{yo - 2/1). We then have :

X — _^i _ ?/ —yi .
(I)
a?2 — ^1 2/i — 2/1

Equation (I) is satisfied by the coordinates of those points and

only those points which lie on the line PiP^. Consequently,
27
28 ANALYTIC GEOMETRY

by Ch. equation of the line through the tivo given

1, § 7, (I) is the

points.

Example 1. Find the equation of the line which passes

through the points (1, - 2) and (- 3, 4).

Here
Xi = l, 2/1 =—2 and iCg = — '% 2/2 = 4.
By (I) the equation of the line is

^-i -y-izij) or ^^^^=y±^.

-3-1 4 -(-2)' -4 6

On clearing of fractions and reducing, the equation becomes

Sx-{-2y + l = 0.
Let the student show that, if (o^i, y{) had been taken as
(_ 3^ 4) and (0^2, 2/2) as (1, — 2), the same equation would have
resulted.

Example 2. Find the equation of the line passing through

the origin and the point (a, h).
Here, (aji, 2/1) = (0, 0) and (0^2, 2/2) = (a, ^), and (I) becomes

? = 2/, or hx-ay = 0.
a b

._ Lines Parallel to the Axes. In deducing (I) we tacitly as-

sumed that
2/0
— 2/1 T^ and Xo — Xi^O'^
for otherwise we could not have divided by these quan-
tities.

If y^^y^^ 0, the line is parallel to the axis of x. .

Its

equation is, then, obviously

(2) 2/ = 2/1-

Similarly, if x^ — x^ = 0, the line is parallel to the axis of y

and has the equation *

(3)
• x = x^.
These two special cases are not included in the result em-
 THE STRAIGHT LINE 29

bodied in equation (T). We see, however, that they are so

simple, that they can be dealt with directly.*

Example 3. Find the equation of the line passing through

the two points (—5, 1) and (—5, 8).
It is clear from the figure that this line is parallel to the
axis of y and 5 units distant from it to the
y
left. Accordingly, the abscissa of every point <-5,8)

on it is —5 ; conversely, every point whose

abscissa —5
is lies on it. Therefore, its equa-
<-5,l)
tion is

x = — 5, or x-{- 5 = 0. Fig. 2

exercises!
Draw the following lines and find their equations.

1. Through (1, 1) and (3, 4). Ans. 3x - 2^/ - 1 = 0.

2. Through (5, 3) and (- 8, 6).

^. Through (0, - 5) and (- 2, 0). Ans. 5x+2y + 10 = 0.

A. Through the origin and (—1, 2).

•"5. Through the origin and (— 2, — 3). \

/e. Through (2, - 3) and (- 4, - 3). Ans. i/ + S = 0.
* It is not difficult to replace (I) by an equation which holds in all

cases, — namely, the following :

(I') iy2-yi)(x-xi) = (x2-xi)(y-yi).

We prefer, however, the original form (I). For (I) is more compact
and easier to remember, and the special cases not included in it are best
handled without a formula.
and (^2, 1/2) in (I), the
t In substituting numerical values for (Xi, r/i)
student will do well to begin with a framework of the form
X- _y —

and then fill in each place in which Xi occurs ;

next, each place in which
M?/i occurs ;
and so on. When Xi or yi is negative, substitute it first in

parentheses ; thus, if Xi = — 3, begin by writing

x-(-3) _y-> .

-(-3) -
30 ANALYTIC GEOMETRY

Through (0, 8) and (0, - 56).

Through (5, 3) and parallel to the axis of y.

Through (5, 3) and parallel to the axis of x.

Through (a, h) and (6, a). Ans. x -{- y = a-\-h

Through (a, 0) and (0, 5).
^^^ ^ + ^ =1.
a b

2./0ne Point and the Slope Given. Let it be required to find

the equation of the line which passes through a given point
A •
(^b 2/i)^^d ^^s a given slope, A.
If P : (x, y) be any second point on the line, the slope of the
line will be, by Ch. I, § 4,

y-Vi
X — 3/j

But the slope of the line is given as A. Hence

X— Xi
or
(II) y-y^ = X(x- iCi).

The student can now show, conversely, that any point, whose
coordinates (x, y) satisfy (II), lies on the given line. Hence
(II) is the equation of the line passing through the given point
and having the given slope.

Example. Find the equation of the line which goes through

the point (2, — 3) and makes an angle of 135° with the posi-
tive axis of X.
Here, A =—1 and (xi, 2/1) = (2, — 3), and hence, by (II), the
equation of the line is

2/ + 3 = -l(a;-2),
or X + + 1 = 0.
?/

Slope-Intercept Form of Equation. It is frequently conven-

ient to determine a line by its slope A, and the ^/-coordinate

of the point in which it cuts the axis of y.

 THE STRAIGHT LINE ^ U 31 ^

Here, a^i = ; and, if we denote y^ by the letter 6, (II) be-

comes ^j "*
(III) , y = \x + h. "^ .^^
This is known as the slope- ijiterce^jt form of the equation
of a straight line ; h is known as
the intercept of the line on the axis
oiy.

Example. Find the equation of the

line which makes an angle of 60° with
the axis of x and whose intercept on the "^
j,j^ .^

axis of 2/
— 2.
is

Since X = V3 and 6 = — 2, the equation is

•"v"
,.^^i.>:: EXERQISES
Draw the following lines and find their equations.
, Through (—4, 5) and with slope — 2.
Ans. 2x-\-y -\-Z = 0.
2. Through (3, 0) and with slope f.

Through (|,
— i) and with slope — |.
4. Through the origin and making an angle of 60° with the
axis of X.

yb. Through (— 4, 0) and making an angle of 45° with the

/^xis of y.

With intercept 1 on the axis of y and with slope — |.

Ans. 3x + 2.y-2"'=0.
't) With intercept ^ on the axis of y and making an angle
oOO° with the axis of x.

With slope — 1 and intercept —c on the axis of

/^. y.

9. With slope a/h and intercept h on the axis of y.

Ans. ax—hy-\- b-=0.

The General Equation of the First Degree. Let there be

3.
given an arbitrary line of the plane. If the line is parallel
 : :

32 ANALYTIC GEOMETRY

to neither axis, its equation is of the form — an equa-

(I), § 1,

tion of the first degree in x and y. If the line is parallel to

the axis of a;, its equation is of the form y = yi, — a special

equation of the first degree in x and y, in which it happens
that the term in x is lacking. Similarly, if the line is parallel
to the axis of y, its form x = Xi,
equation an equa-
is of the —
tion of the first degree which lacks the term in y. Conse-
quently, we can say The equation of every straight line is of the
:

first degree in x and y.

Given, conversely, the general equation of the first degree in

x and y, namely
(1) Ax + By+C=0,
where A, B, C are any three constants, of which and B are A
not both zero * this equation represents ahvays a straight line.
;

Tlie Case B^O. In general, B will not be zero and we

can divide equation (1) through by it

^a; + y + -=0,
B B
and then solve for y:
y
^ =
An X
B B
But this equation is precisely of the form (III), § 2, where

B' B
Therefore, it represents a straight line whose slope is — A/B
and whose intercept on the axis of y is — C/B.
Tlie Case B = 0. If, however, B is zero, the equation (1)
becomes .^ ^
Ax-\-C=0.
Now, A cannot be zero, since the case that both A and B are
zero was excluded at the outset. We can, therefore, divide by
A and then solve for x

A
* In dealing with equation (1), now and henceforth, we shall always
assume that A
and B are not both zero.
 THE STRAIGHT LINE 33

This is >.
tms equation of a straight line parallel to the axis
of y,\i C^ Cr.^^y C = 0, it is the equation of this axis.
This completS the proof that every equation of the first
degree represents a straight line. In accordance with this
property, such an equation is frequently called
a linear equation.

Exa7nple. What line is represented by the

equation

If we solve for y, we obtain

Fig. 4
Hence the equation represents the line of slope
—2 with intercept — i on the axis of y. From these data we
may draw the line.

EXERCISES
Find the slopes and the intercepts on the axis of y of the
lines represented by the following equations. Draw the lines.
^^^^ 4a;-|-2^— A. 2x — y = l.
^>^^.
2. + 5 = 0.
7x-^Sy yf^y^O.
3. =
2x— 5y 0. ^~'~T^x = S — y.
Find the slopes of each of the following lines.

— x + 2y = 7. Ans. 2?/- 3 = 0.
= 3ij.
A 9.
x=y-}-l.
3 — 2x = 5y.
2x
x = oy -\-l.
hx-\- ay = ah.
/^yiO. 2x- 32/ = 4.
4. In the pre)?6ding paragraph we learned to
Intercepts.
plot the line represented by a given equation, from the values
of its slope and its intercept on the axis of y, as found from
the equation. It is often simpler, however, in the case of a
line which cuts the axes in two distinct points, to determine
from the equation the coordinates of these two points and then
to plot the points and draw the line through them.
 .

34 ANALYTIC GEOMETRY

The point of intersection of a line, for example,

(1)
2x-Sy + = 0, 4.

with the axis of x has its ^-coordinate equal to 0. Conse-

quently, to lind the a>coordinate of the point, we have but to

set 2/ = in the equation of the line and solve for x. In this
case we have, then,
2x + 4 = 0, or x = — 2.
Similarly, the x-coordinate of the point of intersection of
the line with the axis of y is 0, and its ^/-coordinate is obtained
by setting a; =
in the equation of the line and solving for y.
In the present case this gives
_32/ + 4 = 0, or y =
-1-

The pointsof intersection of the line (1) with the axes of

coordinates are, then, (- 2, 0) and (0, |). ^Ye now plot these
points and draw the line through them.
We recognize the number | as the
intercept of the line (1) on the axis of y ;

X the number —
2 we call the intercept on
the axis of x. We have plotted the line
Fig 5
by finding its intercepts.
(1), then,
In general, the intercept of a line on the axis of » is the
ic-coordinate of the point in which the line meets that axis. The
intercept on the axis of y is similarly defined. These defini-
tions admit of extension to any curve. Thus, the circle of Ch. I,_.

§ 7, has two intercepts on the axis of x, and — 2.

namely, +2
An axis .or a line parallel to an axis has no intercept on that
axis. Every other line has definite intercepts on both axes,
and these intercepts determine the position of the line unless
they are both zero, that is, unless the line goes through the
origin.
EXERCISES
Determine the intercepts of the following lines on each of
the coordinate axes, so far as such intercepts exist, and draw
the lines.
 THE STRAIGHT LINE 35

/i) 2x-^Sy-6 = 0.
36 ANALYTIC GEOMETRY

/TV triangle
tr has its vertices at the points (0, 1), (—2, 0),
(f^^. Draw the triangle and find the equations of its sides

(produced). Use formula (IV), when possible.

6. A triangle has its vertices at the points (a, 0), (b, 0),

(0, c). Find the equations of the sides (produced).

7. A line goes through the origin and the mid-point of that
side of the triangle of Ex. 5 which lies in the first quadrant.
Find its equation.

8. Find the equations of the lines through the origin and

the respective mid-points of the sides of the triangle of Ex. 6.

/G. Tarallel and Perpendicular Lines. Parallels. Given two

lines oblique to the axis of y, so that both have slopes. The
lines are parallel if, and only if, they have equal slopes. For,
if they are parallel, their slope angles, and hence their slopes,
are equal ; and conversely.
Example 1. To find the equation of the line through the
point (1, 2) parallel to the line

(1) 3x-2y + 6^0.

The slope of the line (1) is f The required line has the
.

same slope and passes through the point (1, 2). By (II), § 2,
its equation is ,

y-2==i(x-l),
or
Sx-2y-hl = 0.

If the given line is it has no slope and

parallel to the axis of y,
hence the method of Example 1 is inapplicable. But then the
required line must also be parallel to the axis of y and its
equation can be written down directly. For example, if the
given line is Sec + 8 = 0, and there is required the line parallel
to it passing through the point (—8, 2), it is clear that the
required line is parallel to the axis of y and 8 units to the left
of it, and consequently has the equation x = — S, or a; -|- 8 = 0.

Perpendiculars. Given two lines oblique to the axes, so that

both have slopes, neither of which is zero. The lines are per-
 :

THE STRAIGHT LINE 37

pendicular if, and only if, their slopes, Ai and Ao, are negative
reciprocals of one another :

(2) A2 = -f, or Ai = -f, Ai^O, X2^0.

A\ A2

For, if the lines are perpendicular, one of their slope angles,

Oi and $2, may be taken as 90° greater than the other, viz.

^0 = 6'i -f-
90°,
and hence
Ao = tan (9. = tan (Oi + 90°) =-cotO^ = — = -~,
tan^i Ai
or

Ao = •
•

Conversely, if this last equation is valid, the steps can be

retraced and the lines shown to be perpendicular to each
other.

Example 2. To find the equation of the line through the

point (1, 2) perpendicular to the line (1).
The slope of (1) is f Hence the required line has the slope
.

— -|. We have, then, to find the equation of the line through

the point (1, 2) with slope — f. By (II), § 2, this equation is

y-2 = -l{x-l),
or
2a; + 32/- 8 = 0.
If the given line is parallel to no slope or its an axis, it has
slope is zero. In either and the method of
case, equation (2)
Example 2 are inapplicable. But then the required line must
be parallel to the other axis and it is easy to write its equation.
Suppose, for example, that the given line is 2?/ — 3 = 0, a —
and that the required line per-
line parallel to the axis of x, —
pendicular to through the point (3, 5). Then this
it is to go
line must be parallel to the axis of y and at a distance of
3 units to the right of it. Consequently, its equation is
a; - 3 = 0.

The methods of this paragraph are applicable to all problems

 38 ANALYTIC GEOMETRY

in whichit is required to find the equation of a line which

passes through a given point and. is parallel, or perpendicular,

to a given line.

EXERCISES
In each of the following exercises find the equations of the
lines through the given point parallel and perpendicular to
the given line.

Line Point
y^^ Vl. ^x - 82/ = 5, (- 1, - 3).
* h ^ — 2y — 5
x-2y-l
Ans. x = 0', 2x-\-y-\-5 = 0.
X2. x-y = l, (0,0).

f3.5x + 13y-S = 0, (2,-1).

/4. Sx-\-5y =0, (5,0).
5. ,2a; = 3, (5, -6).
6.'
V2?/ + = 0, TT (-2,0). Ans. y = 0', a; + 2 = 0.
( 7. 1 - = 0,
a; . (0, tt).

8. Find the equations of the altitudes of the triangle of

§ 5, Ex. 5.

9. Find the equations of the perpendicular bisectors of the

sides of the triangle of § 5, Ex. 5.

10. Show that the equation of the line through the point
(xi, 2/1) parallel to the line

(3) Ax-{-By = C
is Ax + By = Axi + Byi.

11. f Show that the equation of the line through the point
(^1) yi) perpendicular to the line (3) of Ex. 10 is
Bx — Ay = Bxi — Ay I.

/
7. Angle between Two Lines. Let Li and L2 be two given
lines, whose slopes are, respectively,
Ai = tan Oi, and A.2 = tan $2.

To find the angle, <^, from Li to L2.

 :

THE STRAIGHT LINE 39

Since
<t>
= O2 — Oi,
it follows from Trigonometry that

, ,
= tan — tan 61—
O'y
tan ^ ,
1 + tan 61 tan 0^
and hence that

tan = A2 — Aj Fig. 6
(1) <f>
1 + A1A2
The angle <f>
is the angle from Li to Lo. That is, it is the
angle through which Li must be rotated in the positive sense,
about the point A, in order that it coincide with Zo. In par-
ticular, we agree to take it as the smallest such angle, always
less, then, than 180° < < 180° * :
<^

and X2 are perpendicular, then, by (2),

If Li § 6, A2 =— 1/A.i

and 1 4- XiX^ = 0. Consequently, cot c^, which is equal to the

reciprocal of the right-hand side of (1), has the value zero, and
so (^ = 90°.
Example. Let Li and L^ be given by the equations,
Zi: 4.T-22/ + 7 = 0,
7^2 12x + 4y-5 = 0.
Here Ai =2 and A2 = — 3, and (1) becomes

tan <^ = -3 =L
1-6
Hence the angle <^ from L^ to Xo is 45°.

In deducing (1) it was assumed that L^ and

L2 both have slopes. If this is not the case,
at least one of the lines is parallel to the axis
of y and no formula is needed. The angle
<^ may be found directly. Suppose, for ex-
Fig. 7 ample, that Xi and L^ are, respectively,

X -f 2 = and x — y = l.
* The figure shows Ly and L2 as intersecting hnes, but formula (1) and
the deduction of it are valid also in case ii and Xo are
parallel. In this
 :

40 ANALYTIC GEOMETRY

Then L^ is parallel to the axis of ?/, and Lc^ is inclined at an

angle of 45° to the positive axis of x, since A2 = 1. Conse-
quently, <^ = 135°.
EXERCISES
In each of the following exercises determine whether the
given lines are mutually parallel or perpendicular, and in case
they are neither, find the angle from the first line to the second.

'x + 2y=iZ, x-\-2y=4:.

2x^y-{-5 = 0, 4a;- 22/ - 7 = 0.
^S. x — y = l, X = 2.
-\-y

4^ a; + 2?/ -h 11 = 0, 6x-3y-4. = 0.
5.'3x — y = 0j 2x-\-y = 0.

/^ x-\-2y-[-l = 0, 2x-^y-l=^0.
7. 4:X-\-3y = 3,^ ^' 9x-Sy = 5,
S. 2x-Sy = ly aj-3 = 0.
9. x-^y = 0, y = o.
10. 2a; - 3?/ -f 1 = 0, Sx -4.y -1 = 0.
11.1 By the method of this paragraph determine each of the
angles of the triangle whose sides have the equations

x-2y-6 = 0, 2x + y-4. = 0, Sx-y-{-3 = 0.

Check your results by adding the angles.

\ 12. Prove that if L^ and i, ^-i'^ represented by the equations

A: AiX-{-B,y-^C, = 0',' .
.

L2 ^2^ + ^2y + C2 = 0,
then tan = A^B^-A.B^
cf>
A1A2 + B1B2
What can you say of Li and L2 if A1B2 — A2B1 = ? If
A,A2 4- 51^2 = ?

case, we take the angle from Li to L2 as 0° — , not as 180"^, as is conceiv-

able. Hence arises the sign < (less than or equal to) in the place in
which it stands in the double inequality.
 : ;

THE STRAIGHT LINE 41

13. Show that the formula of Ex. 12 for tan <^ is valid even
if one or both of the lines has no slope, i.e. is parallel to the
axis of v.

(\/ /
8. /Distance of a Point from a Line. Let P:(xi, 2/1) be a
^ v^' giy^n point and let

\^; L: Ax-{-By-{-C=0
\ be a given line. To find the dis- ^(xi,yi)
V tance, D, of P from L.
\V^ Drop a perpendicular from P
'

on the axis of x, and denote the

point in which it cuts L by Q.
The abscissa of Q is a^i. Denote
its ordinate by y^. Then Fig. 8

QP=yi-yq'
Since Q : (x^, y^) lies on i, its coordinates satisfy the equation
of L ',
thus
Axi^By^-{-C=0.
Solving this equation for y^, we find

= Ax^-{- C
y<i
B
Hence
gp^Axj^^-^yj^{-_C
(1) ^ B
be the slope-angle of L 8.nd form the product QPcos 6.
Let
One or both of the factors of this product may be negative,
according to the positions of P and L.* But always the
numerical value of the product is equal to the distance D :

(2) i)=|QPcos^|.
This is clear in case P and L are situated as in Fig. 8

* There are four essentially different positions for P and L, for L may
have a positive or a negative slope, and P may he on the one or on the
other side of L,
 :

42 ANALYTIC GEOMETRY

the student should draw the other typical figures and show
that for them, also, (2) is valid.
Since the slope of X is

X = tan = — A
we have
sec2 6 = 1 + tan^ 6 = A^ + B'
Consequently,
cos =± B
(3)

It is immaterial to us which sign in (3) is the proper one,

For, according to (2), we have now to multiply together the
values of QP and cos 6, as given by (1) and (3), and take the
numerical value of the product. The result is the desired
formula ^ > . ^ ^\

V^^ + B^
or
(4)
j^^^ Ax, + By, + O ^

Va^ + b^
where, in the second formula, that sign is to be chosen which
makes the right-hand side positive.

Example. The distance of the point (3, — 2) from the line

3a; + 4?/- 7 =
IS

|3.3-f4(-2)-7| _ |-6| ^6^-^,

.
,V32 + 42
—
V25 5
'"'

The deduction of formula (4) involves division by B and

hence tacitly assumes that B=^0, i.e. that L is not parallel to
the axis of y. The formula holds, however, even when L is
parallel to the axis of y. For, in this case it is clear from a
figure that
^

and (4) reduces precisely to this when B = 0.

 : :

THE STRAIGHT LINE 43

EXERCISES
In each of the first seven exercises find the distance of the
given point from the given line.

Point Line
(5, 2), 3a; + 6 = 0.
-4?/ u47is.
2f.
(2, 3), 5a;H-12?/ + 2 = 0.

UT. (6, - 1), 3a;-?/ + l = 0. ^715. 2V10, or 6.32.

4. (3, 4), 3a; + 5 = 0.

^. (-2, -5), 7/=0.

6. Origin, a;+ — 1 = 0.
?/

7.» Origin, 3a; + 22/- 6 = 0.

8. 1 Find the lengths of the altitudes of the triangle with

ces in the points (2, 0), (3, 5), (— 1, 2).

^Y
Sy^rea of a Triangle. Let a triangle be given by means of
its^ertices (xj, y^, (a^g, 2/2), (^s/^/s)-
y (^ziVs)
To find its area.
Drop a perpendicular from one
of the vertices, as (xg, 1/3), on the ^^i.Vi}
opposite side. Then the required
area is <^i^yi)

Fig. 9
where D denotes the length of
the perpendicular and E, the length of the side in question.
By Ch. I, § 3, we have

D is the distance of (ajg, 2/3) from the line joining (oti, 2/1)

and {X2, 2/2)- The equation of this line, as given by (I) or (I'),

§ 1, may be put into the form

(2/2
- 2/1)^ - (^2 - ^\)y - ^ilh 4- 2^2.^1 = 0.

Consequently, by (4), § 8, we find

I) — ± ^y^~ yi)^3- fe - ^\)lh - ^i?/2 + x^yi

-Vix2-Xiy + {y2-yiy
 44 ANALYTIC GEOMETRY

Thus
A = ± i[(^2 - 2/1)^3 - (a?2 - ^-1)2/3 - a)i?/2 + x^yi].
The result may be written more symmetrically in either of
the forms

(1) A = ± l[(a;i - a^)2/3 + (a^2 - ^^'3)2/1 + (^3 - ^51)2/2],

or

(2) ^=± |[(2/i - 2/2)3^3 + (2/2 - 2/3)2^1 + (2/3 - 2/1)^2],

where in each case that sign is to be chosen which makes the
right-hand side positive.

EXERCISES
Find the area of the triangle whose vertices are in the
points

'^(1,2), (-1,2), (-2,1).

^U5,3),(-3,4), (-2, -1).
3. (1, 2), (2, 1), (0, 0).

Find the area of the triangle whose sides lie along the lines
^^^x-y = 0, x-j-y = 0, 2x + y-3 = 0.
"
15^2 x-{-y- 6 = 0, x-y-hS = 0, x-2y-S = 0.
, 6. Find the area of the convex quadrilateral whose vertices
are in the points (4, 2), (- 1, 4), (- 3, - 2), (5,.- 8).

7. What do formulas (1) and (2) become when one of the

vertices, say (x^, 2/3), is in the origin?
Ans. A = ± ^(xiy2 — x^y^.
10. General Theory of Parallels and Perpendiculars. Iden-
tical Lines.* The line through the point {xi, 2/1) parallel to
the line

(1) Ax + By=C,
has the equation, according to § 6, Ex. 10,
Ax -\- By = Axi -f Byi.
* The discussion in the class-room of the subjects treated in this and the
following paragraph may well be postponed until the need for them arises.
 : :

THE STRAIGHT LINE 45

This equation is of the form

(2) Ax + By= C,
since the constant Ax^ + Byi may be denoted by the single
letter C\
Conversely, equations (1) and (2), for C =^ C, always repre-
sent parallel lines. For, if B=^0, the lines have the same
slope, — A/B ; JB = 0, ^
cannot be zero, and the lines are
if

parallel to the axis of y and hence to each other.

Theorem 1. Tivo Uiies are parallel ivhen and only ivhen their
equations can be ivritten in the forms (1) and (2), luhere C =^ C
The line through the point {x^, y^, perpendicular to the
line (1), has the equation (§ 6, Ex. 11)

Bx — Ay — Bxi — Ayiy
and this equation is of the forra

(3) Bx -Ay= a.
Let the student show, conversely, that equations (1) and (3)
always represent perpendicular lines.

Theorem 2. Tiuo lines are perpendicular ichen and only

when their equations can be ivritten in the forms (1) and (3).
The equations of two parallel lines can always be written
in the forms (1) and (2). But they need not be so written.
Thus the lines,
2x-y=-l,
6aj-32/ = 2,
are parallel, though the equations are not in the forms (1) and
(2). The coefficients of the terms in x and y are not respec-
tively equal. They are, however, proportional : 2 : 6= — 1 — 3.
:

This condition holds in all cases. For the two lines

Ly : A,x^ B,y + Ci = 0,
L2 : A2X 4- Boy -h C2 = 0,

we may state the theorem

46 ANALYTIC GEOMETRY

Theorem 3. Tlie lines L^ and L^ are parallel * if and only if

For, Li and L2 are parallel if and only if the angle <^ be-
tween them, as defined in § 7, is zero but, according to § 7, ;

Ex. 12, or better, tan <^, is zero, when and only when
<f),

A1B2 —
A2B1 = 0. But this equation is equivalent to the pro-
portion Ai A2 = Bi B^.
: :

As a second consequence of § 7, Ex. 12, we obtain the fol-

lowing theorem.
Theorem 4. The lines Li and L2 are perpendicular if and
only if
A1A2 + B1B2 = 0.
Identical Lines. Two equations do not have to be identically
the same in order to represent the same line. Eor example,
the equations,
2x-y-\-l = 0,
6x-3y + S = 0,
represent the same line. The corresponding constants in

them are not equal, but they are proportional. We have,

namely,
2:6 = -l: -3 = 1:3,
or, what amounts to the same thing,

2: -1:1=6: -3:3.
This condition is general. We formulate it as a theorem :

Theorem 5. The lines L^ and L2 are identical if and only if

A^.A2=B^.B2= C\:C2,
or Ai Bx Ci
: : = A2 : B2 : 62.

For, Li and L2 are the same line when and only when they
have the same slope and the same intercept on the axis of y,
that is, when and only when

-^ = -^
Bo
and -§
B,
= -§,
B2
Bx
* Or, in a single case, identical. Cf Th.
. 5.
 THE STRAIGHT LINE 47

or Ai A2 = Bi: B2
: and Bi. B2= Ci : C2,

or, finally, Ai A2
: = Bi : B2 = C\ : CV
This proof assumes that Bi=^0 and B., =^ 0. The proof,
when this is not the case, is left to the student.

EXERCISES
1. Prove Th. 3 directly, without recourse to the results
ofy§ 7.

2. The same for Th. 4.

See also Exs. 15, 16, 17, 18 at the end of the chapter.

Ik^ Second Method of Finding Parallels and Perpendiculars.

[problem 1. To find the equation of a line parallel to the
given line
(1) Ax + By=C,
and satisfying a further condition.
By § 10, Th. 1, the desired equation can be written in the
form
(2) Ax -\-By = V,
where C is to be determined by the further condition.

Example. Consider the first example treated in § 6. In

this case the equation of the desired line can be written in the
form WN
3x-2y = k, ^^
where we have replaced the C" of (2) by k. The "further
condition," by means of which the value of k is to be deter-
mined, is that the line go through the point (1, 2)„ Hence
cc = 1, = 2
2/ must satisfy the equation of the line, or

3-l-2-2 = A:./''
Consequently, k = —l, and the equation of the line is

Sx-2y-hl = 0.
Problem 2. To find the equation of a line perpendicular to
the given line (1) and satisfying a further condition.
48 ANALYTIC GEOMETRY

By § 10, Th. 2, the desired equation can be written in the

form
(3) Bx-Ay= C%
where C
is to be determined by applying the further condition.

This condition does not always have to be that the line

should go through a given point. It may
be any single con-
dition, not affecting the slope of the line, which it seems de-
sirable to apply. We give an example illustrating the method
in such a case.

Example. To find the equation of the line perpendicular to

2a;-?/ -4 =
and cutting from the first quadrant a triangle whose area is 16.
Equation (3) may, in this case, be written as

(4) x-\-2y=: k.

We are to determine k so that the line (4) cuts from the first

quadrant a triangle of area 16. The intercepts of the line (4)

are k and i k, and hence the area of the triangle in question is
^k^. Accordingly, \k^ = 16, and k =± S. But the line cuts
the first quadrant only if k is positive, and so we must have
/-; = 8. The equation of the desired line is, then,

x + 2y-S = 0.
EXERCISES
1. Work Exs. 1-4, 8, 9 of § 6 by this method.
"V 2. There are two lines parallel to the line

X — 2y = % t

and forming with the coordinate axes triangles of area 9.

Find their equations.

3. Find the equatiotis of the lines parallel to the liile of
IEx. 2 and 3 units distant from it.

Suggestion. Write the equation of the required line in the

form (2) and demand that the distance from it of a chosen
point of the given line be 3.
 THE STRAIGHT LINE 49

'4^ Find the equations of the lines parallel to the line

5x-\- 12?/ -3 =
and 2 units distant from the origin.

5. The same as Ex. 2, if the lines are to be perpendicular,

instead of parallel, to the given line.
'he same as Ex. 4, if the lines are to be perpendicular,
istead^of parallel, to the given line.

7. A line is parallel to the line Sx-\- 2y — 6 =0, and forms

'^a triangle in the first quadrant with the lines,

X — 2y = and 2x — y = 0,
whose area is 21. Eind the equation of the line.

Ans. 3x + 2y-2S=0.
EXERCISES ON CHAPTER II

1. Find the equation of the line whose intercepts are twice

those of the line 2x — 3y — 6 = 0.
2. Find the equation of the line having the same intercept

on the axis of x as the line VSx — 2/ — 3 = 0, but making with

that axis half the angle.
3. Find the equation of the line joining the point (7, —2)

with that point of the line 2x — y = S whose ordinate is 2.

-'ir^ perpendicular from the origin meets a line in the point
(5, 2). What is the equation, of the line?
5. The coordinates of the foot of the perpendicular dropped
from the origin on a line are (a, b). Show that the equation
of the line is
^
>v ax-\-by = a'^+ b\
6. The line through the point (5, — 3) perpendicular to a
given line meets it in the point (— 3, 2). Find the equation
"^
^f'the given line.

7. Prove that the line with intercepts 6 and 3 is perpen-

dicular to the line with intercepts 3 and — 6. Is it also per-

pendicular to the line with intercepts 3 — and 6 ?

50 ANALYTIC GEOMETRY

8. Prove that the line with, intercepts a and b is perpen-

dicular to the line with intercepts b and —a.
9. Show that the two points (5, 2) and (6, — 15) subtend a
right angle at the origin.
LO. Prove that the two points, (xi, y^) and (0^2, 2/2)5 subtend a
right angle at the origin when, and only when, X1X2 + 2/i?/2 = 0.
11. Do the points (6, —1) and (—3,4) subtend a right
angle at the point (4, 6) ? At the point (—4, — 2) ?
12. Given the triangle whose sides lie along the lines,

x-2y-{-6 = 0, 2x-y = 3, x-\-y-3 = 0.

Find the coordinates of the vertices and the equations of the
lines through the vertices parallel to the opposite sides.

13. Two sides of a parallelogram lie along the lines,

2x-\-Sy-6 = 0, 4:x-y = 4..

A vertex is at the point (—2, 1). Find the equations of the

other two sides (produced).
14. One side of a rectangle lies along the line,

5£c + 4?/-9 = 0.
A vertex on this side is at the point (1, 1) and a second vertex

is at (2, — 1). Find the equations of the other three sides

(produced).
15. For what value of A will the two lines,

3x-2y-\-6 = 0, Xx-y-{-2 = 0,
(a) be parallel ? (b) be perpendicular ?

.6. For what value, or values, of m will the two lines,

4:X — my + 6 = 0, x-\-my -{-S = 0,

(a) be parallel ? (by be perpendicular ?

L7. For what value of m will the two equations,

mx -{-
y -\- 5 = 0, 4a; + m?/ + 10 = 0,
represent the same line ?
 THE STRAIGHT LINE 51

;. For what pairs of values for k and I will the two

situations,
12x-\-ky + = 0,l lx-5y-{-3 = 0,
represent the same line ?

19. The equations of the sides of a convex quadrilateral are

x = 2, 2/ = 4, y = x, 2y = x. <\
Find the coordinates of the vertices and the equations of the
diagonals. »

20. Find the equation of the line through the point of

intersection of the lines,

3a; -5?/ -11 = 0, 2aj-72/ = ll,

and having the intercept —5 on the axis of y.

21. Find the equation of the line through the point of in-
tersection of the lines,

2x-\-hy = 4, 3x - 4?/ + 17 = 0,
and perpendicular to the first of these two lines.

52. Find the distance between the two parallel lines,

3a; - 4?/ + 1 = 0, 6a; - 82/ + 9 = 0.

Suggestion. Find the distance of a chosen point of the tirst

line from the second.

Let
Ax-\-By^-C=^ and Ax \- By -\- C=^
be any two parallel lines. Show that the distance between
them is
G'-C\ or
.
C'-C
V^2 + jB^' -Ja}^-w-
'24. There are two points on the axis of x which are at the
distance 4 from the line 2a; — S^/ — 4 = 0. What are their
coordinates ?
25. Find the coordinates of the point on the axis of y which
is equidistant from the two points (3, 8), (—2, 5).
 '

^
•AH- r

52 ANALYTIC GEOMETRY

26. There are two lines through the point (1, 1), each
cutting,. ^roni the first quadrant a triangle whose area is 2\.
Find/their slopes. Ans. — i, — 2.
\y27. Find the equation of the line through the point (3, 7)
such that this point bisects the portion of the line between
the axes. Ans. 7x -\-Sy — 42 = 0.
28. The origin lies on a certain line and is the mid-point of
that portion of the line intercepted between the two lines,

y^ Sx — 5y = 6, 4a; + + 6 = 0.
?/

Fin(L4Jne equation of the line. Ans. x -\- 6y = 0.

\J^. The line

(1) Sx-Sy + 5==0

goes through the point (1, 1). Find the equation of the line
(2) through this same point, if the angle from the line (1) to
the line (2) is 45°. Ans. llx-5y-6 = 0.
30.Find the equations of the two lines through the origin
making with the line 2x — Sy = angles of 60°.
 :

o~^
1

CHAPTER III
;, 1

APPLICATIONS

1. Certain General Methods. Lines through a Point. In

many theorems and problems Geometry the question
of Plane
is to show that three lines pass through a jjoint. Plane Geom-
etry affords, however, no general method for dealing with this
question. Each new problem must be discussed as if it were
.the first of this class to be considered.
Analytic Geometry, on the other hand, affords a universal
method, whereby in any given case the question can be settled.
For, from the data of the problem, the equation of each of the
lines can be found. These will all be linear, and can be writ-
ten in the form

Li : A,x-h B,y + Ci = 0,
L2 : A2X+ -B22/ + C2 = 0,
is A^ + B,y + C3 = 0- •

The coordinates of the point of intersection of two of these

lines, asLi and L2, can be found by solving the corresponding
equations, regarded as simultaneous, for the unknown quanti-
ties X and y. Let the solution be written as
x = x', y = y'.
The third line, L-^, ivill pass through this point (x', y'), if and
only if the coordinates of the latter satisfy the equation of L^;
i.e. if and only if
A,x' 4- B,y' + C3 = 0.
Points on a Line. A second question which presents itself
in problems of Plane Geometry is to determine when three

53
 ,

54 ANALYTIC GEOMETRY

points on a straight line. Here, again, the reply of Analytic

lie

Geometry is methodical and universal. From the data of the

problem it will be possible in any given case to obtain the
coordinates of the three points. Call them
{xi, 2/1), (.-^2, 2/2), fe, 2/3).

Now, we know how to write down the equation of a line

through two of them, as (xi, y^ and {X2, y^)- This equation
will always be linear, and can be written in the form

Ax + By-^C=0.
The third point, (xs,ys), ivill lie on this line if and only if its

coordinates satisfy the equation of the line; i.e. if and only if

Ax, -\- By, -{ C= 0.

The student should test his understanding of the foregoing

theory by working Exs. 1-6 at the end of the chapter.

2. The Medians of a Triangle. We recall the proposition

from Plane Geometry, that the medians of a triangle meet in a
point. The proof there
given is simple, provided
one remembers the con-
struction lines it is desir-
able to draw. By means,
however, of Analytic Ge-
ometry we can establish j?;(-i,3)

the proposition, not by

but by the natural
artifices,

and direct application of

the general principle enun-
ciated in the preceding j5.(4,0)
paragraph.
The first step consists in
the choice of the coordinate axes. This choice is wholly in
our hands, and we make it in such a way as to simplify the
coordinates of the given points. Thus, clearly, it will be well
 :

APPLICATIONS 55

to take one of the axes along a side of the triangle. Let this
he the axis of x.
A good choice for the axis of y will be one in which this
axis passes through a vertex. Let this be the vertex not on
the axis of x.

We begin with a numerical case, choosing the vertices A, B,

C at the points indicated in the figure.

The Equations of the Medians. Consider the median AA'.

One point on this line is given, namely ^ : (— 2, 0). A second
point is the mid-point A' of the line-segment BC. By Ch. I,

§ 5, the coordinates of A' are (2, 3).

The student can now solve for himself the problem of finding
the equation of the line Li through ^ : (— 2, 0) and A' : (2, 3).

The answer is, ,

Lii 3a; -4?/ 4- 6 = 0.

In a precisely similar way the coordinates of B' are found
to be(— 1, 3), and the equation of the median BB' is
L^: Sx-\-5y-12 = 0.
Finally, the coordinates of C are (1, 0), and the equation of
the median CC is

L^: Qx-\-y — 6 = 0.
The Point of Intersection of the Medians. The next step con-
the point in which two of the medians, as Li
sists in finding
and X2, intersect. The coordinates of this point will be given
by solving as simultaneous the equations of these lines :

3a- -4?/ -1-6 = 0,

3a; + 5?/ -12=0.

The solution is found to be

a^ =i 2/ = 2.
And now the third median, X3, will go through this point,
(1, 2), if the coordinates of the point satisfy the equation of 2>3,

6a; + y-6 = 0.
 )

56 1 ANALYTIC GEOMETRY

On substituting for x in this equation the value | and for y

the value 2, we are led to the equation

6. 1 + 2-6 = 0.
This is a true equation, and hence the three lines L^, L^, and
1^3 pass through the same point.

Remark. It can be shown by the formulas of Ch. I, § 6,

that the above point (f, 2)

trisects each of the medians AA',
BB', and CC.

EXERCISES
Taking the same triangle as before, choose the axis of x
1.

along the side AB, but take the axis of y through A. The
coordinates of the vertices will then be :

^:(0, 0); 5: (6,0); (7: (2, 6).

Prove the theorem for this triangle.-

C 2^ The vertices of a triangle lie at the points (0, 0), (3, 0),
(tJ^y). Prove that the medians meet in a point.

3. Continuation. The General Case. AYe now proceed to

prove the theorem of the medians for any triangle, ABC. Let

C:(0,c)

B:{ o> 9 A-<l>|)

A:{a,0) B:(b,0)

Fig. 2

the axes be chosen as in the text of § 2. Then the coordinates

of A will be (a, 0), where a may be any number whatever,
 :: :: ;

r
APPLICATIONS 57

positive, negative, or zero. The coordinates of B will be (6, 0),

where h may be any number distinct from a

b=^ a, or a — b =^ 0*
Finally, the coordinates of C can be written as (0, c), where c
is any positive number.
Next, find the coordinates of A', B'y C. They are as sho\vn
in the figure.
The equation of L^ is given by Ch. II, (I), where

X— a y —
=^
It is:
—
I
2
a
2
"'

or
Lii cx-j-(2a — h)y = ac.
The equation of L^ can be worked out in a similar manner.
But it is not necessary to repeat the steps, since interchanging
the letters a and h interchanges the points A and B, and also
A' and B'. Thus L^ passes over into L2. Hence the equation
of L2 is :

X2 ex -\-(2h — d)y = he.

The line L^ is determined by its intercepts, |-(a + 6) and c

by Ch. II, (lY), its equation is found to be

ig 2cx-\-(a-\- h)y = (a 4- b)c.

To find the coordinates of the point in which Li and X2 in-

tersect, solve as simultaneous the equations of Li and L2 :

\cx-{-{2a — b)y = ac,

\cx+(2b — a)y = be.
The result is
a +6 c

* The figure has been drawn for the case in which a is negative and 6

positive.
 :

58 ANALYTIC GEOl^ETRY

finally, to show that this point, (^^4~-' §)' ^^^^ °^ -^3» ^^^

stitute its coordinates in the equation of Z3

2c^+(a + 6)|=(a + &)c.

Since this is a true equation, the point lies on the line, and we
have proved the theorem that the medians of a triangle pass
through a point.
That this point trisects eachmedian can be proved as in the

special case of the preceding paragraph, by means of Ch. I, § 6.

The details are left to the student.

EXERCISE
Prove the theorem of the medians by taking the coordinate
axes as in the first exercise of the preceding paragraph. Here,
the vertices are

^:(0, 0); ^:(a, 0); C'.{h,c),

where a may be any number not 0, h any number whatever,

and c any positive number. Draw the figure, and write in the
coordinates of each point used."

The Altitudes of a Triangle. Another proposition of

4.

Plane Geometry is, that the perpendiculars dropped from the

vertices of a triangle on the opposite sides meet in a point.
The proof of the proposition by Analytic Geometry is direct
and simple. Let us begin with a numerical case, taking the

triangle of Fig. 1. One of the perpendiculars is, then, the

and so all that is
axis of y, necessary to show is that the other
two meet on this axis, or that the cc-coordinate of their point

of intersection is 0.

The equation of the line BC can be written down at once in

terms of its intercepts :

?+^ = l, or 3x-{-2y=:12.
4 6
 :

APPLICATIONS 59

The slope of this line is A =—f

The slope of any line .

perpendicular to it A' = |.
Hence the equation of Xi, the
is

perpendicular which passes through the point A: {—2, 0), is

or
Lii 2x-3y + = 0. 4:

In a similar manner the student can obtain the equation of

the perpendicular L2 from B on the side AC. It is,
X2: 05 + 3?/ — 4 = 0.
On computing the a;-coordinate of the point in which Li and
L2 intersect, it is found that x = 0, and hence the proposition
is established for this triangle.

Remark. For use in a later problem it is necessary to

know the exact point in which the perpendiculars meet. It is
readily shown that this point is (0, ^).

EXERCISES
. 1. Prove the above proposition for the special triangle con-
Cfeidered, choosing the coordinate axes as in Ex. 1 of § 2.

Prove the proposition for the triangle of Ex. 2, § 2.

"sT Prove the proposition for the general case, choosing the

axes as in Fig. 2. First show that the equation of the perpen-

dicular Li from A on BC is
Li hx — cy = ah,
and that the equation of the perpendicular L^ from BoiixiC is

L2 ci^ — c^y = ^^•

Then show that these lines intersect each other on the axis
of y. V ^
4. Show that the point in which the perpendiculars in the

preceding question meet is f 0, V

5. Prove the theorem of the altitudes, when the axes of

coordinates are taken as in the exercise of § 3.
 : :

60 ANALYTIC GEOMETRY

5. The Perpendicular Bisectors of the Sides of a Triangle.

It is shown in Plane Geometry that these lines meet in a
point. Since the student is now in full possession of the
method employed in Analytic Geometry for the proof of this
theorem, he will find it altogether possible to work out that
proof without further suggestion. Let him begin with the
special triangle of Fig. 1. He will find that the equations of
the perpendicular bisectors of the sides are the following

2a;-3?/-f 5 = 0;

a." - 1 = 0.
These lines are then shown to meet in the point (1, |).
He
can work further special examples corresponding to the
exercises at the end of § 2 if this seems desirable.
Finally, let him work out the proof for the general case, tak-
ing the coordinate axes as in Fig. 2. The three lines will be
found to have the equations

ii hx- \{lf- - c%
cy -=

L2,'. ax — = i(a2 — c^),

cy

ig : . X — i(a + 5).
They meet in the point

(a +h ah + cA
V 2 ' 2c /

EXERCISE
Give the proof when the axes of coordinates are taken as in
the exercise of § 3.

6. Three Points on a Line. The foregoing three propositions

about triangles have led to three points, namely, the three
points of intersection of the three lines in the various cases.
In the case of the special triangle of 'Fig. 1, these points are

(1,2); (0,1); (1,1).

 :

.J APPLICATIONS 61

These points lie on a straight line. Let the student try to

prove this theorem by Plane Geometry.
The proof by Analytic Geometry is given immediately as a
direct application of the second of the general principles enun-
ciated in the opening paragraph of the chapter.
Write down the equation of the line through two of these
points, —
say, through the first and third. It is found to be :

A,
The coordinates of the second point, -A ^

x = 0, y = i, \^
are seen to satisfy this equation, and the proposition is proved.

EXERCISES
1. Prove the proposition for the general case (Fig. 2). The
points have been found to be

fa -\-h c\ ff.
ab\ fa + h ab -f c- \
(,"3"' 3/ X' op 2c
[ 2 '
J
2. On plotting the three points obtained in the special case
discussed in the text it is observed that the line-segment de-
termined by the extreme points is divided by the intermedi-
ate point in the ratio of 1 2. Prove this analytically. Is it
:

true in general?

EXERCISES ON CHAPTER III

1. Prove that the three lines,

2aj-32/-5 = 0, 3rr -f 4y - 16 = 0, 4a; - 23i/ -f 7 = 0,

go through a point.
2. Prove that the three lines,

ax + hy = l, hx + ay = l, x — y = 0,
go through a point.
3. Prove that the three points (4, 1), (- 1, - 0), and
(2, — 3)lie on a line.
62 ANALYTIC GEOMETRY

4. Prove that the three points {a, b), {b, a), and
(— a, 2a4- b) lie on a line.

5. Find the condition that the three lines,

bx + ay = 2ab, ax \- by — a? -\- b"^, 3a; — 2i/ = 0,

where a^ is not equal to 6-, meet in a point.

6. Find the condition that the three points {a, b), (b, a),

and (2 a, — b), where a is not equal to b, lie on a line.

Lines through a Point

Q Show that the line drawn through the mid-points

parallel si(i€s of a trapezoid passes through the point of inter-
of the

section/^ the non-parallel sides.

Show that, in a trapezoid, the diagonals and the line
drawn through the mid-points of the parallel sides meet in a
point.

^9. A right triangle has its vertices A, B, and in the points

^
Qo), and (0, 0). The points A' (4, - 4)
(0, 3), and B' (- 3,
: :

3) are marked. Prove that the lines AB', BA', and the per-
pendicular from on the hypothenuse meet in a point.
10. (Generalization of Ex. 9.) Given a right triangle ABO
with the right angle at 0. On the perpendicular to OA in

the point A measure off the distance AA', equal to OA, in the
away from the hypothenuse. In a similar fashion
direction
mark the point B' on the perpendicular to OB in B, so that
BB' = OB. Prove that the lines AB', BA', and the perpen-
dicular from on the hypothenuse meet in a point.
11. Let any point (a, a) of the line x — y = 0, other
P be
than the origin. Through P draw two lines, of arbitrary slopes
Ai and A2, intersecting the ic-axis in A^ and A2 and the ?/-axis
in Bi and B2 respectively. Prove that the lines A1B2 and
A2B1 will, in general, meet on the line x-\-y = 0.
12. on the three sides of a triangle as diagonals paral-
If
lelograms, having their sides parallel to two given lines, are
 ' ' ^
'
Jr I n^^i \ 1^ fi* *v ^

APPLICATIONS 63

described, the other diagonals of the parallelograms meet in a

point.
Prove this theorem, when the given lines are the coordinate
axes,and the triangle has as its vertices the points (1, 6),
(4, 11), (9, 3).

13. Prove the theorem of the preceding exercise, when the

given lines are the axes, and the triangle has its vertices in the
points (0, 0), (a, a), (6, c).

Points on a Line

14. Show- that in the parallelogram ABCD the vertex D,

the mid-^oint of the side AB, and a point of trisection of the
diag^al AC lie on a line.

\/15. Prove that the feet of the perpendiculars from the

point (2, — 1) on the sides of the triangle with vertices in the
points (0, 0), (3, 0), and (0, 1) lie on a line.
16. Prove that the feet of the perpendiculars from the point
(—1,4) on the sides of the triangle with vertices in the points
(2, 0), (- 3, 0), and (0, 4) lie on a line.

17. Show that the feet of the perpendiculars from the point

(0, — ) on the sides of the triangle with vertices in the points

(a, 0), (6, 0), and (0, c) lie on a line.

.--^ Let Mhe

the point of intersection of two opposite sides
of a quadrilateral, and N, the point of intersection of the other
two sides. The mid-point of MJSf and the mid-points of the
diagonals lie on a right line.

Prove this proposition for the special case that the vertices
of the quadrilateral are situated at the points (0, 0), (8, 0), (6, 4),
'

(1,6).
19. Prove the proposition of Ex. 18 for the general case.
Suggestion. Take the axis of x through and N, the M
origin being at the mid-point. The equations of the sides can

then be written in the form

64 ANALYTIC GEOMETRY

y = \^{x - /i), y = \2{y^ - h),

20. Let be the foot of the altitude from the vertex C of

the triangle ABC on the side AB. Then the feet of the per-

pendiculars from on the sides BC and AC and on the other

two altitudes lie on a line.
Prove this theorem for the triangle ABC with vertices in
thepoints(l, 0), (-4, 0), (0, 2).
21. Prove the theorem of the preceding exercise for the
triangle with vertices in the points (a, 0), (h, 0), (0, c). It

will be found that

ac- Oy-C be"' b'^c

/ a'-b -abc \ ( a/>2 - abc\

l^a2 + c'' a2-hcy U- + C-' + C2/
6-

are the coordinates of the four points which are to lie on a

line, and that
c{a-\- b)x -f {ab — c^)y = abc
is the equation of the line.
 : :

CHAPTER IV

THE CIRCLE
1. Equation of the Circle. According to Ch. I, § 7, the
equation of the circle whose center is at the origin, and whose
radius is p, is

(1) .
x''- + f- = p'.

In a precisely similar manner, the equation of a circle with

its center at an arbitrary point
y
C : (a, /3) of the plane, the length
r-(^,y)
of the radius being denoted by p, is
found to be :

(2) (0^
- ay + (2/
- isy- = p\
Example. Find the equation of "q'
the circle whose center is at the fig. i
'

point (— f, 0), and whose radius is |.

Here, « = — |, /? = 0, and p = f. Hence, from (2)

This equation can be simplified as follows

+ ^ + y^ = f,
aj2-f fa^
or, finally,
3a;2 + 3?/2 + + 4 = 0. 8a.'

EXERCISES
Find the equations of the following circles, and reduce the
results to their simplest form. Draw the figure each time.
1. Center at (4, 6) ; radius, 3.

Ans. a;2 + 2/2 - 8a; - 12 ?/ -f 43 = 0.

2. Center at (0, - 2) ;
radius, 2. Ans. x' + y- +4 2/ = 0.
65
 ;

/qQ ANALYTIC GEOMETRY

j

3. Center at (— 3, 0) ; radius, 3.

4. Center at (2, — 4) ;
radius, 8.

5. Center at (0, |) ;
radius, |.

Center at (3, — 4) ;
radius, 5.

Center at (— 5, 12) ; radius, 13,

Center at (i, — |) ;
radius, 2.

Center at (— |, f) ;
radius, ^.
Center at (a, 0) ;
radius, a.

11. Center at (0, a) ;

radius, a.

12. Center at (a, a) ;

radius, a V2.

2. A Second Form of the Equation. Equation (2) of § 1 can

be expanded as follows :

a;-? _|_ ^/-i _ 2ax - 2^?/ + a*^ -h /S" - p^ = 0.

This equation is of the form
(1) aj^ + 2/' + ^^ ^By+C=0.
Let us see whether, conversely, equation (1) always repre-
sents a circle.

Example 1. Determine the curve represented by the

equation

(2) aj2 + 2/2 + 2a; — 62/ + 6 = 0.

We can rewrite this equation as follows :

(a;2-H2a;
)j^{jy'i-Qy ) = - 6.
The first parenthesis becomes a perfect square if 1 is added
the second, if 9 is added. To keep the equation true, these
numbers must be added also to the right-hand side. Thus
\ (a;2 -f. 2iB + 1) + (2/2 - 62/ -f 9)= - 6 -f 1 + 9,
for
(a; +1)2+ (2/ -3)2 = 4.
This equation is precisely of the form (2), § 1, where
a = — 1, yg = 3, p = 2. It therefore represents a circle whose
center is at (— 1, 3), and whose radius is 2.
 ::

"

V ,< \;\

THE CIRCLE 67

Example 2. .
What curve is represented by the equation

(3) ^2 + 2/2+1 = 0?
It is clear that no point exists whose coordinates satisfy this
equation. For, x- and y^ can never be negative. Their least
values are 0, — namely, for the origin, (0, 0), — and even for
this point, the left-hand side of the equation has the value + 1.
Hence, there is no curve corresponding to equation (3).

Example 3. Discuss the equation

(4) x-"^ + 2/- + 2a^ _ 42/ -f 5 = 0.

Evidently, this equation can be written in the form

(5) (x -f l)H (y - 2)^ = 0.

The coordinates of the point (—1, 2) satisfy the equation.
But, for any other point {x, y), at least one of the quantities,
cc -j- 1 and y — 2,is not zero, and the left-hand side of the equa-
tion is positive. Thus the point (— 1, 2) is the only point
whose coordinates satisfy the equation. Hence equation (4)
represents a single point (— 1, 2).

Remark. Equation (5) can be regarded as the limiting case

of the equation
(x-^iy-\-iy-2y = p\
when p approaches the limit 0. This equation represents a
circle of radius p for all positive values of p. When p ap-
proaches 0, the circle shrinks down toward the point (— 1, 2)
as its limit. Accordingly, equation (5) is sometimes spoken
of as representing a circle of zero radius or a null circle.

The General Case. It is now clear how to proceed in the

general case, in order to determine what curve equation (1)
represents. The equation can be written in the form
{x^ + Ax-\- \A^) -f (/ + By + \B') = -C+\A^ + \B\
or
/ .
^\- ,
/ B\ A''JtB--AC
 :

68 ANALYTIC GEOMETRY

If the right-hand side is positive, i.e. if

then equation (1) represents a circle, whose center is at the

point {— ^A, — ^B) and whose radius is

however, A^ -\- B'^ — 4:C =

If, 0, then equation (1) represents
just one point, namely, {— \A, — ^B), — or, if one prefers, a
circle of zero radius or a null circle.
Finally, when A"^ -\- B^ — C<4: 0, there are no points whose
coordinates satisfy (1). To sum up, then

Equation (1) represents a circle, a single j)oint, or there is no

point whose coordinates satisfy (1), according as the expression

is positive, zero, or negative.

Consider, more generall}^, the equation

(6) a{x'^ + /-) -f 6ic + cy + d = 0.

If a = 0, but h and c are not both 0, the equation represents
a straight line.

If, however, a 4^0, the equation can be divided through by

a, and it thus takes on the form :

«^ + 2/^ + -a^ + -2/ + - = 0.

a a a

This is precisely the form of equation (1), and hence the above
discussion is applicable to it.

EXERCISES
Determine what the following equations represent. Apply
each time the method of completing the square and examining
the right-hand side of the new equation. Do not merely sub-
stitute numerical values in the formulas developed in the text.
 ,

THE CIRCLE 69

1. + y^-\-6x-S^j = 0.
x'^

Ans. A circle, radius 5, with center at (— 3, 4).

2. x' + y^ - -f 4?/ + 13 = 0.
6.t' Ans. The point (3, - 2).
3- y" + 2a; + 42/ + 6 = 0. Ans. No point whatever.
^l+
rlc'2 + - 10.T + 24?/ = 0.
2/'

*^x*- + y^ — 7x=5.
.
+ -60^ + 8?/ + 25 = 0.
a;- 2/'

7. 49.r'^ + 492/^-14a; + 28?/ + 5 = 0.

Ans. The point - J|). (^ig-,

.
+ 2/'^H-82/=10.
a;-^

9. x'
,^ y' = 2 ax.
-\-

10?\a;2 -\-f- = 2ay.

+ ^2 _ 6a^. _ 261/ 4_ 9a2 = 0.

'x'^

12. + + 4aa;-86y+ 166- = 0.

a;- ?/-

3. aj-^+^/' + S^O.

^ 14.-'E-^ + _ 2a; + 4?/ + 10 = 0.

?/2

I 0^/7
<^^
15. 3a;2 + 32/2-4a: + 22/ + 7 = 0.

16. 5x'--\-5y'~-6x + Sy = 12.

y^l7. 3a;- + 3?/' — + = 6. /k^

a; 2/

' ii

3. Tangents. Let the circle

(ly x'^ + y'^p'

be given, and let P^ : (x'l, 2/1) be any point of this circle. To
find the equation of the tangent
at Pi.
The tangent at Pi is, by Ele- A-^3;„2/j
mentary Geometry, perpendicular to
the radius, OPi. Hence its slope,
A', is the negative reciprocal of the
slope, yi/xi, of OPi ;
or

X' =- FiQ. 2
2/1
 : : :

70 ANALYTIC GEOMETRY

We wish, therefore, to find the equation of the line which

passes through the point (xi, y^ and has the slope A' =— iCi/2/1.

By Oh. II, § 2, (II), the equation of this line is

(2) 2/-2/i = ~-(-«-a5i).

2/1

This equatj^m can be simplified by multiplying through by

posing

point (it'i, 2/1) is, by hypothesis, on the circle ; hence

coordinates satisfy the equation (1) of the circle

a^i^ + 2/1^ = P^-

The right-hand side of equation (3) can, therefore, be replaced

by the simpler expression, p*^.
We thus obtain, as the final form of the equation of the
tangent, the following

(4) xix-i-y^y = p\
In deducing this equation it was tacitly assumed that 2/1 =^ 0,
since otherwise we could not have divided by it in obtaining
A.'. The final formula, (4), is true, however, even when 2/1 = 0,
as can be directly verified. For, if ?/i
= and Xi = p, then (4)
becomes
px = p^ or x = p,
and this is the equation of the tangent in the point (p, 0).
Similarly, when ?/i
= and Xi = — p.
Any Circle. If the given circle is represented by the
equation

(5) {x-ay-^{y-(3y==p\
precisely the same reasoning can be applied. The equation
of the tangent to (5) at the point Pi : (xi, y^) of that circle is

thus found to be :

(6) (x,-a){x-a) + {y,-li){y-li)=p\

The proof is left to the student as an exercise.
 : , .

7^
THE CIRCLE /
7J

If the equation of the circle is given in the form

(7) x' + if + Ax^-By-^ C= 0,

or in the form (6), § 2, the equation can first be thrown into
the form (5), and then the equation of the tangent is given
by (6).

Example. To find the equation of the tangent to the circle

(8) 3a;2-f-3?/2-f 8x-5?/ =

at the origin.
First, reduce the coefficients of the terms in x- and y- to

unity

Next, complete the squares :

x^ + fa.' +(1)2 + - 12/ + (!)' = ¥ + M = ft

7/2

or (^ + |)^+(2/-|)^ = H.

Now, apply the theorem embodied in formula (6). Since

a^i = 0, = 0, a = - I,
2/1
=f /3

we have %{x + f) - |(2/ - I) = H.

or 8x — 52/ = 0,

as the equation of the tangent to (8) at the origin.

EXERCISES

Find the equation of the tangent to each of the followmg

circles at the given point.

a;2 + 7/2 = 25 at (- 3, 4). Ans. 3a; - 4?/ + 25 = 0.

2. 0^2 _^ 2/2 = a" at (0, a). Ans. y = a.

)(^3. 0^2
+ 2,2 = 49 at (-7,0).
~
^^ {x- 1)2+ (y + 2)2 = 25 at (4, 2). Ans. 3.r + 4.y = 20.
Vs.; (a;4- 5)2+(2/ - 3)2 = 49 at (2, 3).

^6. x2 + - 9a; + ll^/ = at the origin.

^ 2/-
 .

ANALYTIC GEOMETRY
.
2x^ -\-2y^-Sx-y = ll at (- 1, 2).

8. Find the intercepts on the axis of x made by the tangent

,t (- 5, 12) to
a;2 _|. 2/2 3= 169. Ans. - 33f
9. Find the area of the triangle cut from the first quadrant
the tangent at (1, 1) to

3a;2 ^ 3^/2 ^Sx-\- 16y = 30.

(lOJ If the equation
^^ x'^-{-y''- + Ax + By+C=0
represents a circle, and if the point (xi, y^ lies on the circle,

show that the equation of the tangent at this point can be

written in the form

^'^'^'•-^
T f'^-
:
'

(9) x^x + y^-\-^{x + x^) + ^{y + y,)+C =0. 4/

Find the values of a, ^, and p for the

Suggestion. circle,

substitute them in (6), and simplify the result.

11. Do Exs. 6 and 7, using formula (9), Ex. 10.

12. The same for the tangent to the circle in Ex. 9.

13. Show that, if Pi :

(«i, y^) is any point of the circle

'
a;5-|-2/2 + ^a; + 5?/+ C = 0,

at which the tangent is not parallel to the axis of 2/, then the
slope of the tangent at Pi is
2a;i +^
5*
22/1 +

4. Circle through Three Points. It is shown in Elementary

Geometry that a circle can be passed through any three points

not lying in a straight line.

If the points are (a^i, 2/1), fe, .^2), and (x^, 2/3), and if the
equation of the circle through them is written in the form

x'i-\-y'^-\. Ax + By^C = 0,
then clearly the following three equations must hold :
 : : : : :

THE CIRCLE 73

x,^ + 2/1' + ^^i + ^Ih + C= 0,

»2^ + 2/2' + ^^2 + ^2/2 + C'= 0,
2^3'+ ?/3' + ^^z + + C' = 0. jB^^3

We thus have three simultaneous linear equations for de-

termining the three unknown coefficients A, B, C.
Suppose, for example, that* the given points are the
following
(1,1), (1,-1), (-2,1).
The equations can be thrown at once into the form

A-\-B + C = -2,
A-B+C = -2,
-2A + B+C = -5.
Solve two of these equations for two of the unknowns in
terms of the third. Then. substitute the values thus found in
the third equation. Thus the third unknown is completely
determined, and hence the other two unknowns can be found.
Here, it is easy to solve the first two equations for A and B
in terms of C. On subtracting the second equation from the
first, we find
25 = 0; hence B = 0.
Then either of the first two equations gives for A the value

A = -C-2.
Next, set for A and B in the third equation the values just

found
2C+4+0 = -5, (7 = -3.
Hence, finally,
^ = 1, J5 = 0, (7 = -3,
and the equation of the desired circle is

a;2 -f 2/'^
H- a; - 3 = 0.
Check the result by substituting the coordinates of the
given points successively in this last equation. They
are

found each time to satisfy the equation.

 K
:

ANALYTIC GEOMETRY -^ \->^"^

The circle through the three given points has its center in
the point (- |, 0). Its radius is of length V3^= 1.803.
EXERCISES
Find the equations of the circles through the following
les of points. Plot the points and draw the circles.

1. (1, 0), (0, 1), the origin.* Ans. x'^ + y'- - x - y = 0.

|j(l,l),(-l, -1), (1, -1).
(5,10), (6,9), (-2,3).
4. The vertices of the triangle of Ex. 15 at the end of
Ch. Ill, p. 63. Show that the point (2, — 1) of that exercise
lies on the circle.

5. The same question for Ex. 17, p. 63. Show that the

point [0, — I
of that exercise lies on the circle.

6. The vertices of the triangle of Ch. Ill, Eig. 1. Find the

coordinates of the center and check by comparing them with
those of the point of intersection of the perpendicular bisectors
of the sides of the triangle, as determined in Ch. Ill, § 5.

7. The same question for the triangle of Ch. Ill, Fig. 2.

Check.
(8.') The vertices of the triangle formed by the coordinate
ax6s and the line '2x — Sy = 6.
9. The vertices of the triane^le whose sides are

x-y-l = 0, x + + 2 = 0, 2aj -2/4-3 = 0.

2/

Ans. + 32/^4-17a;+162/ + 25 = 0.
3a;'^ X
>L EXERCISES ON CHAPTER IV \^
Find the equation of the circle with the line-segment
i. join-

ing the two points (3, 0) and (5, 2) as a diameter.

2. A circle goes through the origin and has intercepts — 5
and 3 on the axes of x and y respectively. Find its equation.
3. A circle goes through the origin and has intercepts a and
b. Find its equation.
 75

4^jFind the equation of the circle which has its center in

the point (— 3, 4) and is tangent to the line 3x + 8?/ — 6 = 0.
5. A circle has its center on the line 2.t — 3y = and
passes through the points (4, 3), (—2, 5). Find its equation.

6. Find the equation of the circle which passes through the

point (5,—2) and is tangent to the line 3ic — —1= ?/ at
the point (1, 2).

7. There are two circles passing through the points (3, 2),
(— 1, 0) and having 6. as their radius. Find their equations.
There are two circles with their centers on the line,
8.

5x — 3y = S,
and tangent to the coordinate axes. Find their
equajkt^s. |^ \ ^1

/^. Find the equations of the circles tangent to the axes and
passing through the point (1, 2).

10. Find the equations of the circles passing through the

points (3, 1), (1, 0) and tangent to thq line x — y = 0.
Suggestion. Demand that the center (a, /3) be equally
distant from the two points and the line.

11. Find the equations of the circles passing through the

origin, tangent to the line x -{- y — S = 0, and having their
centers on the line x = 2.
12. Find the equations of the circles of the preceding exer-
cise, if their centers lie on the line 2x — y — 2 = 0.
13. Find the equation of the circle inscribed in the triangle
formed by the axes and the line 3x — 4^/ — 12 = 0.
14. Find the equation of an arbitrary circle, referred to two
perpendicular tangents as axes.

15. Do the four points (0, 0), (6, 0), (0, - 4), (5, 1) lie on
a circle ?

16. Find the coordinates of the points of intersection of the

circles
3524. _ x-\-2y = 0,
2/2

a;2^2/2 + 2x— y = 9. 1
 76 ANALYTIC GEOMETRY

17. Find the coordinates of the points of intersection of the

circles „
x^ -t y^
o
-\- ax — 0,
-\-
t
by
/^

x^ + y"^-
-\- hx — ay = 0.
?i

Orthogonality
?18. A circle and a line intersect in a point P. The acute
angle between the line and the tangent to the circle at P is

known as the angle of intersection of the line and the circle at

P. If the line meets the circle in two points, the angles of

intersection at the two points are equal. Determine the angle
in the case of the circle
a;2 + = 25, 2/2

and the line 2ic — — 5 = 0.

t/

19. A circle and a line are said to intersect orthogonally if

their angle of intersection is a right angle. Prove that the
circle, ,

fl;2 4-2/2_4a;4-6?/4-3 = 0,
is intersected by the line, 5x -{-y = 7, orthogonally.
Suggestion. First answer geometrically the question : What
lines cut a given circle orthogonally ?

(20r^how that the circle,

—^ a;^ + 2/- + Ax -{-By-h 0=0,
/ .

inte;?sects the line,

'^
ax -]- by -\- c = Oj
orthg^nally when and only when
aA + bB = 2c.
21. If two circles intersect in a point P, the acute angle
between their tangents at P is known as their angle of inter-
section. If the circles intersect in two points, their angles of
intersection at these points are equal. Find this angle in the
case of the circles,
a;2 + 2/2^25,
a;2_|_2/2_7a; + = 0. 2/
 THE CIRCLE 77

22. Prove geometrically that two circles intersect orthog-

onally, that is, at right angles, when and only when the sum
of the squares of their radii equals the square of the distance
between their centers. Then show that the circles

a;^+ y'^ — 4:X-\-5y- 2=0,

2x^ + 2^/- + 4a; - 62/ - 19 = 0,
intersect orthogonally.

23. Prove that the two circles,

a;2 + + A,x + B,y +

2/' C\ = 0,
a;2 + y2 4. A.X + ^2/ + 6*2 = 0,
intersect orthogonally when and only when

AiA,^ + B^Bo = 2 Ci -f 2 C2.

24. Find the equation of the circle which cuts the circle

aJ' + 2/-H-2a; =
at right angles and passes through the points (1, 0) and (0, 1).

25. There are an infinite number of circles cutting each of

the two circles,
x'^+ f~-4.y-h2 = 0,
aj2-f2/' + 42/H-2 = 0,

orthogonally. Show that they are all given by the equation

aj2 + 2/2 4- «« - 2 = 0,
where a is an arbitrary constant. Where are their centers?
Draw a figure.

26. Find the equation of the circle cutting orthogonally the

three circles,
x2 += 9,2/-

x-2^y'i^Sx-5y + 6 = 0,
a;2 4- 2/2 - 2a; + 3?/ - 19 = 0.
Ans. x- + y''--\-10x-{-9 = 0.
78 ANALYTIC GEOMETRY

Miscellaneous Theokems

27. Prove analytically that every angle inscribed in a semi-

circle is a right angle.

28. Prove analytically that the perpendicular dropped from

is a mean proportional between
a point of a circle on a diameter
the segments in which it divides the diameter.
29. The tangents to a circle at two points P, Q meet in the
point T, The lines joining P and Q to one extremity of the
diameter parallel to PQ meet the perpendicular diameter in
the points R and S. Prove that RT=ST.
30. In a triangle the circle through the mid-points of the
sides passes through the feet of the altitudes and also through
the points halfway between the vertices and the point of inter-
section of the altitudes. This circle is known as the Nine-
Point Circle of the triangle.
Por the triangle with vertices in the points (—4, 0), (2, 0),
(0, 6) construct the circle and mark the nine points through
which it passes.
31. Por the triangle in the preceding exercise find the equa-
tion of the nine-point circle, as the circle through the mid-
points of the sides. Ans. 3a;- + 3?/' + 3 — lli/ = 0.
a;

32. Show that this circle goes through the other six points.
33. Por the triangle with vertices in the points (a, 0), (h, 0),

(0, c) find the equation of the nine-point circle, as the circle

through the mid-points of the sides.

Ans. 2c(x'' -f y'')—(a -\- b)cx -f(a6 - c'^)y = 0.

34. Show that this circle goes through the other six points.
 ;

CHAPTER V
INTRODUCTORY PROBLEMS IN LOCI. SYMMETRY OF
CURVES

1. Locus Problems.* A
is moving under given condi-
point
tions; its locus This type of problem the student
is required.
studied in Plane Geometry. But he found there no general
method, by means of which he could always determine a locus
for each problem he had to devise a method, depending on the
particular conditions of the problem.
Analytic Geometry, however, provides a general method for
the determination of loci. Some simple examples of the
method have already been given. Thus, in finding the equa-
tion of a circle, we determined the locus of a point whose dis-
tance from a fixed point is constant. Again, in deducing the
equation of a line through two points, we found the locus of a
point moving so that the line joining it to a given point has a
given direction.
The method in each of these cases consisted merely in ex-
pressing in analytic terms —
form of an equation
i.e. in the
involving the variable coordinates, x and y, of the moving
point —
the given geometric condition under which the point
moved. We proceed to show how this method applies in less
simple cases.

Example 1. The base of a triangle is fixed, and the dis-

tance from one end of the base to the mid-point of the opposite
side is given. Find the locus of the vertex.
* The locus problems in this chapter may be supplemented, if it is de-
sired, by§§ 6-8 of the second chapter on loci, Ch. XIII, in which the loci
of inequalities and the bisectors of the angles between two lines, together
with related subjects, are considered.
79
 :

80 ANALYTIC GEOMETRY

Let the triangle be OAP, with as the mid-point of AP. M

Let a be the length of the base OA, and let I be the given
distance. It is required to

P:(x.y) find the locus of P, so that

always

M(-|) (1) 0M= I.

It is convenient to take
5 the origin of coordinates in
o A:(a,o)
and the positive axis of x
Fig. 1
along the base. The coordi-
nates of A are then (a, 0). The coordinates of the moving
point P we denote by (x, y). The coordinates of the point
Jtf are
fx + a y\
^ 2 '
2/
The distance OM is

+ r
x-\- a
i 9 1

Thus condition (1), expressed analytically, is

x-\- ay
4 HI =
Squaring both sides of this equation and simplifying, we have

(2) .
(x^ay+y^ = (2iy.
This equation represents the circle whose center is at {—a, 0)
and whose radius is 2 1. We have shown, therefore, that, if

(1) is always satisfied, the coordinates (x, y) of P satisfy (2),

and P lies on the circle. The locus of P appears, then, to be
the circle.
How do we know, though, that P traces the entire circle ?
To prove this, we must show, conversely, that, if the coordi-
nates (x, y) of P satisfy (2), condition (1) is valid. If (ic,
y)
on dividing both sides of (2) by 4 and extract-
satisfy (2), then,
ing the square root of each side, we obtain two equations
 :

INTRODUCTORY PROBLEMS IN LOCI 81

i)
4 '-riHt = ?,

x-\-ay ///V_
ii)
4 ^^
;
^

Equation ii) says that a positive or zero quantity equals a

negative quantity, and is therefore impossible. Thus only
equation i) remains. This equation says that 0M= I. Hence
condition (1) is satisfied by every point of the circle,* and so
the circle is the locus of P.
We have yet to describe the locus, independently of the
coordinate system, with reference merely to the original tri-

angle. Produce the base, in the direction from A to 0, to the

point (7, doubling its length. Then the locus of P is a circle,
whose center is at C and whose radius is twice the given
distance.

Example 2. Determine the locus of a point P which moves

so that the difference of the squares of its distances from two
fixed points Pi, Po is constant,
and equal to c

PPi' - PP22 = c,

(3)
[PP2--PPi- = C.

Take the mid-point of the

segment P1P2 as origin and
the axis of x along P1P2.
Fig. 2
The coordinates of Pi and Pg
can be Avritten as (—a, 0), (a, 0) ; those of P, as (x, y).

By Ch. I, § 3,

PP,2 = (a^ + a)2 4. y2^ pp^2= (x - a)~ + y\

Then the equations (3), expressed analytically, are

* The two points in which the circle cuts the axis of x are exceptions,
since these do not lead to a triangle, OAP.
82 ANALYTIC GEOMETRY

{x+ a)2+ y^-(x- ay - y^ = c,

{x - ay+ If -{x + af - if = c.

These reduce to

(4) 4 aa; = c, 4 aa; =— c.

Hence, if condition (3) is satisfied, P lies on one or the

other of the lines

Conversely, if P lies on one or the other of the lines (5),

then (4) holds, and from (4) we show by retracing the steps
that one or the other of the equations (3) is valid.
Consequently, the locus of P
consists of two straight lines,

perpendicular to the line P^Pi, and symmetrically situated

with reference to the mid-point of PiPn, the distance of either
line from the mid-point being c/4 a. Thus the locus consists
of two entirely unconnected pieces, one corresponding to each
of the equations (3). If c = 0, these equations are the same,

and the two lines forming the locus coincide in the perpendic-
ular bisector of the segment P1P2.

EXERCISES

In solving the following problems, the first step is to find

the equation of a curve, — or the equations of curves, — on

which points of the locus lie. The student must then take
care (a) to show, conversely, that every point lying on
the

curve or curves obtained satisfies the given conditions and ;

the locus, finally, without reference to the

(b) to describe
coordinate system used.

Cl^ A point P moves so that the sum of the squares of its

distences to two fixed points Pi, P2 is a constant, c, greater

than i PiP2^. Show that the locus of P is a circle, with its

center at the mid-point of PiPo.

What is the locus if c = ^ P1P22 ? If c < J PiPo^ ?
 INTRODUCTORY PROBLEMS IN LOCI 83
>^,

(J- ^(^ Find the locus of the mid-point of a line of fixed length
wEich moves so that its end points always lie on two mutually
perpendicular lines.
3. Determine the locus of a point which moves so that the
sum of the squsjees of its distances to the sides, or the sides
produce(J<rtfa given square is constant. Is there any restric-
tion^H^essary on the value of the constant ?

4. Determine the locus of a point which moves so that the

square of its distance to the origin equals the sum of its
coordinates. Ans. A circle, center at (|,
i), radius = |V2.
5. Show that the locus of a point which moves so that the
sum of its distances to two mutually perpendicular lines
equals the square of its distance to their point of intersection
consists of the arcs of four circles, forming a continuous curve.
Where are the circles, and which of their arcs belong to the
locus ?
6. The base of a triangle is fixed, and the trigonometric

tangent of one base angle is a constant multiple, not — 1,

of the trigonometric tangent of the other. Find the locus of
the vertex.

2. Symmetry. In the problems of the preceding paragraph,

the equations of the loci were familiar and the curves they
tjepresented were easily identified. In subsequent chapters,
Z' however, we shall have locus problems to consider in which
the resulting equations will be new
to us. In drawing the curves which
^-^•^^' (*,!/)
these equations represent, it will be I

useful to have at hand the salient

facts concerning the symmetry of
^

curves. i
(x.-v)

Symmetry in a Line. Two points,

P and F', are said to be symmetric
in a line L, if L is the perpendicular bisector of PP\
If L is the axis of x and (x, y) are the coordinates of P, then
it is clear that {x, — y) are the coordinates of P'.

I
84 ANALYTIC GEOMETRY

Similarly, if L the axis of y and

is P lias the coordinates
(a;, y), then P' has the coordinates (— x, y).

Example 1. Given the curve

(1) 2/' = ^^
Let P : (xi, 2/i) he any point on it, i.e. let

(2) yi^ = x,

be a true equation. Then the point P' symmetric

: (a^i, — 2/1),

to P in the axis of x, also lieson the curve.

For, if we substitute the coordinates of P^ into
(1), the result is {— y^"^ = x^, or (2), and (2) we
know is a true equation. We say, then, that
the curve (1) is symmetric in the axis of x.

The test for symmetry in the axis of x,

employed in this example, is general in appli-
cation. We state it, and the corresponding
test for symmetry in the axis of y, in the form
Fig. 4
of theorems.
Theorem 1. A curve is symmetric in the axis of x if the sub-
stitution of —y for y in its equation leaves the equation unchanged.
Theorem 2. A curve is symmetric in the axis of y if the sub-

stitution of — X for X in its equation leaves the equation unchanged.

Symmetry in a Point. Two points, Pand P', are symmetric

in a given point, if the given point is the mid-point of PP'.
If the given point is the origin of V
coordinates and P has the coordinates yOix,y)

then the coordinates of P' are X

(x, y), O
cr
evidently {— x, —y). (-«, -V)

Fig, 5
Example 2. Consider the curve

(3) 2/ = ^•
If P : is any point on this curve, then the point
{xi, 2/1)
p' : ( _ Xi, — symmetric to P in the origin, is also on the
?/i),

curve. For, the condition that P' lies on the curve, namely.

-yi={-^iy or 2/1
= x^^,
 ;

INTRODUCTORY PROBLEMS IN LOCI 85

is equivalent to the coudition :

y^ = Xi', that P lie on the curve.
We say, then, that the curve (3) is symmetric in the origin.
This test, too, is general in application
we formulate it as a theorem.
Theorem 3. A curve is symmetric in the
origin of coordinates, if the substitution of
— X for X, and of — y for y, in its equation
leaves the equation essentially unchanged.
A case in which the test leaves the
equation wholly unchanged is that of the
circle, x^ -{- y"^ = p^, or the curve xy = a- Fig. 6
(Fig- 7).
Now the circle in question is symmetric in both axes. It
follows then, without further investigation, that it is sym-
metric in the origin, the point of intersection of the axes.
This conclusion holds always in fact, ;

y we may state the theorem.

Theorem 4. If a curve is symmetric
in both axes of coordinates, it is symmetric
in the origin.
o * The details of the proof are left to
the student as an exercise. It is to be
noted tha t the converse of th e theorem,
na mely, that if a curve is s ymmetric in
Fig. 7 t he^rigin, it is symmetric in the axes, is
not true. For, the curve of Example 2
is symmetric in the origin, but not symmetric in either axis ;

this is true also of the curve xy = a^ of Fig. 7.

EXERCISES
1. Prove Theorem 4.

2. Test, for symmetry in each axis and in the origin, the

curves given in the following exercises of Ch. I, § 7 :

(a) Exercise 2 ;
-
(c) Exercise 7 ;

(6) Exercise 6 ;
(d) Exercise 8.
86 ANALYTIC GEOMETRY

In each of the following exercises test the given curve for

synanetry in each axis and in the oi^gm. Plot the curve.
+ 1=0.
"^Zyxy 6. f -\- 4 cc = 0.
y^yA^y = x\ "l. .r2_2/2 3=4.

^ 20a; =yK ^8. a;2 + 2 = 16.

?/2

EXERCISES ON CHAPTER V
The base of a triangle is fixed and the ratio of the
lengths of the two sides is constant. Eind the locus of the
vertex. Ans. A circle, except for one value of the constant.
2. A point P moves so that its distance from a given line
L is proportional to the square of its distance to a given point
/i, not on L. If P remains always on the same side of L as
K, show that its locus is a circle.

3. Find the locus of P in the preceding exercise, if it re-

mains always on the opposite side of L from K. Does your

answer cover all cases ?
4. If, in Ex. 2, K lies on L and P may be on either side of
X, what is the locus of P?
5. Three vertices of a quadrilateral are fixed. Eind the
locus of the fourth, if the area of the quadrilateral is constant.

6. Find the locus of a point moving so that the sum of

the squares of its distances from the sides of an equilateral

triangle is constant. Discuss all cases.

Ans. A circle, center at the point of intersection of the me-

dians ; this point ; or no locus.

7. The from the point P (X, Y)

feet of the perpendiculars :

on the sides of the triangle with vertices in the points (0, 0),
(3, 0), (0, 1) lie on a line. Find the locus of P.
Alls. The circle circumscribing the triangle.

8. The preceding problem, if the triangle has the points

(2, 0), (- 3, 0), (0, 4) as vertices.

9. Problem 7, for the general triangle, with vertices at

(a,0), (5,0), (0, c).
 INTRODUCTORY PROBLEMS IX LOCI 87

10. Show that the equation of the circle described on the

line-segment joining the points (xi, y^), (x^, 2/2) 3.S a diameter
may be written in the form

{x - Xi)(a; - x<^) + (2/

- yi){y - 2/2) = 0.
Suggestion. Find the locus of a point P moving so that the
two given points always subtend at P a right angle.
11. The two points, P and P\ are symmetric in the line,
X — y = 0, bisecting the angle between the positive axes of x
and y. Show that, if (x, y) are the coordinates of P, then
(2/, x) are the coordinates of P'.

12. Prove that a curve is symmetric in the line x — y =

if the interchange of x and y in its equation leaves the equation

unchanged.
13. If P and P' are symmetric in the line x + y = and
P has the coordinates (x, y), show that the coordinates of P'
are {— y, — x).

14. Give a test for the symmetry of a curve in the line

x-\-y =0.
Test each of the following curves for symmetry in the
15.
lines X — y = and x -{-y = 0.
(a) xy = a^; (c) x'--y^ = a?',

(b) xy==-a- ;
(d) (x - ijY--2x -2y = 0.
16. Plot the curve of Ex. 15, (d).

In each of the following exercises find the equation of the

locus of the^'oint P. Plot the locus from the equation, mak-
ing^bifthe use possible of the theory of symmetry.
/l7. The distance of P from the line a; +2 = equals its

distance from the point (2, 0).

^ The sum of the distances

18. of P from the points (3, 0)

and (- 3, 0) is 10.
*19. The difference of the distances of P from the points
(5,0) and (-5,0) is 8.
 \j'

V
]

CHAPTER VI
THE PARABOLA

1. Definition. A parabola is defined as the locus of a point

P, whose distance from a fixed line is always equal to its D
distance from a fixed point F, not on
the line. It is understood, of course,
that P is restricted to the plane deter-
mined by D and F.
One point of the locus is the mid-point
A, Fig. 2, of the perpendicular FE
dropped from on D. Through A
F
draw T parallel to D. Then no other
point on T, or to the left of T, can belong
to the locus, for all such points are
Fig. 1
clearly nearer to D than they are to F.
Further points of the locus can be
obtained as follows. To the right of T
draw L parallel to D, cutting AF, pro- D T >N
duced if necessary, in S. With ES as
radius and F as center describe a circle,
cutting i in P and Q. Then P and Q
lie on the locus.
E A F S

A large number
of points having been
obtained in way, a smooth curve
this
can be passed through them. The curve Q.

is symmetric in the line AF, and evi-

Fig. 2
dently has r as a tangent.
The line D is called the directrix, and the point F, the focus,
of the parabola ; A is the vertex, and the indefinite line AF,
the axis ; FP is a focal radius.
88
 THE PARABOLA 89

The studentis familiar with the fact that all circles are

similar have the same shape, and differ only in size. A

; i.e.

like relation holds for any two parabolas. Think of them as

lying in different planes, and choose in each plane as the unit
length the distance be-
tween the focus and the
directrix. Then the one
parabola, in its plane, is

the replica of the other,

in its plane. Conse-
quently, the two parab-
olas differ only in the
which they are
scale to
drawn, and are, there-
fore, similar. Fig
The details of the
proof just outlined can be supplied at once by showing that
the triangles FPM and MFE are similar, respectively, to
F'P'M' and M'F'E\ the angles i/a in Fig. 3 being equal by
construction. Hence
FP ^ EF
F'P~ E'F'^
i.e.focal radii, FP
and F'P', which make the same angle with
the axes always bear to each other the same fixed ratio.

EXERCISES

01. Take a sheet of squared paper and mark D along one of

the vertical rulings near the edge of the paper. Choose F
at a distance of 1 cm. from D. Then the points of the locus

on the vertical rulings — or on as many of them as one desires

— can be marked off rapidly with the compasses. Make a
clean, neat figure.

Place a card under the curve of Ex. 1 and, with a needle,

2.

prick numerous points of the curve through on the card, and

mark, also, the focus and axis in this way. Cut the card along
 z :;

90 ANALYTIC GEOMETRY

the curve with sharp scissors. The piece whose edge is con-
vex forms a convenient parabolic ruler, or templet, to be used
whenever an accurate drawing is desired.
A small hole at the focus and a second hole farther along
the axis make it possible, in using the templet, to mark the
focus and draw the axis.
A second templet, to twice the above scale, will also be

found useful.
3. The focus of a parabola is distant 5 units from the
directrix. In a second parabola, this distance is 2 units. How
much larger is the first parabola than the second, i.e., how do
their scales compare with each other ?

2. Equation of the Parabola. The first step is to choose

the axes of coordinates in a convenient manner. Evidently,
one good choice would be to take
the axis of x perpendicular to D
and passing through F. Let us do
this, choosing the positive sense
from A toward F.
Eor the axis of y three simple
— choices present themselves, namely

(a) through A ;

(b) along D ;

(c) through F.

Perhaps (p) seems most natural

Fig. 4 but (a) has the advantage that the
curve then passes through the
origin, and this choice turns out in practice to be the most
useful one. We will begin with it.

Let P : {x, y)be any point on the curve. Denote the dis-

tance of F from D
by m. Then

m
AF = ~, ^^=?.
EF=m, and
 :

V THE PARABOLA 91

By Ch. I, § 3,

^^=\/(--^T+2/'
On the other hand, the distance of P from D is
X -\

2
By definition, these two distances are equal, or

Square each side of the equation, so as to remove the radical,

and expand the binomials :

(2) x^ - mx -j- —4 + y^ = x^ + mx + —
4
'

The result can be reduced at once to the form

= 2ma;, ^-^
rtx-Toco ^\i
(3) 2/2

and this is the equation of the parabola, referred to its vertex

as origin and to its axis as the axis of x.

The proof of this last statement not yet, however, com-is

plete ;
for it remains to show conversely that, if {x, y) be any
point whose coordinates satisfy (3), it is a point of the parab-
ola. From (3) we can pass to (2). On extracting the square
root of each side of (2), we have two equations :

•^ If '^nV ,
, m
2'

\+y'=-(='+fj'
one of which must be true, and both of which may conceivably
be true. ISTow, cc is a positive quantity or zero for, by ;

hypothesis, the coordinates of the point (x, y) satisfy equation

(3). Hence ii) is impossible, for it says that a positive or
zero quantity is equal to a negative quantity. Thus only i)
 92 ANALYTIC GEOMETRY

remains, and this equation is precisely the condition that the

distance of {x, y) from D be equal to its distance from F.
Hence the point (a;, y) lies on the parabola, q. e. d.

EXERCISES

^
. .

y^l! Show that the choice (6) leads to the equation

(4)
y'^ = 2mx — m^.
This is the equation of the parabola referred to its directrix
and axis as the axes of y and x respectively, with the positive
axis of X in the direction in which the curve opens. .

2. Show that the choice (c) leads to the equation

(5) y^ = 2mx + m^.

This is the equation of the parabola when the focus is the
origin and the positive axis' of x is along the axis of the curve
in the direction in which the curve
opens.
^'^
|^\3. Taking the axes as indicated in
^ Fig. 5, show that the equation of the
-^ parabola is

V Fig. 5 x^ = 2my.
P 4. Choosing the axis of y as in the foregoing question, show
that the equation of the parabola is

x^ = 2my — m^,
in case the axis of x is along D, and is

x^ = 2 my -\- m%
in case F is taken as the origin.
5. If the axis of x taken along the axis of the parabola,
is

but positively in the direction from toward D, and if the F

origin is taken at the vertex, show that the equation of the
curve is
2/2 = — 2 mx.
 THE PARABOLA \ \
-
93

6. If the axis of y is taken along the axis of the parabola,

but positively in the direction from toward /), and if the F

origin is take»-' at^i^e vertex, show that the equation of the
iurve^
a;2 = — 2 my.
7. Determine the focus and directrix of each of the follow-
ing parabolas :

(a) f^ = ^x. ' Ans. (1,0); = 0.

.^-{-1

(6) y = x\ 4^/ + 1 = 0.
'
4ns. (0, J)
;
"

^)~-32/- -5x = 0. +
(d) = 0.
32/2 22 .t

y = - 2x\ ij,.^/Xf)5a? + 12y =

^(g) y'"=px. ^^1^ x^ = Aay.
8. It appears from the foregoing that any equation of the
form
?/^ =± Ax, or «- = ± Ay,
where A is any positive constant, represents a parabola with
its Formulate a general rule for ascer-
vertex at the origin.
taining the distance of the focus of such a parabola from the
vertex.

f
9. Find the Equations of the following parabolas :
^ ,

<^ ^i
\^ (a) A^ertex at (0, 0) and focus at (2, 0).

It (b) Vertex at (0, 0) and 2 a; +5= as directrix.

^""^ ^jT
I (c) Vertex at (0, 0) and focus at (0, — f).
i
f(d) Vertex at (0, 0) and 2 1/ —1= as directrix.

j
^(e) Focus at (0, 0) and vertex at (—3, 0).

I
^(/) Focus at (0, 0) and 3?/ -h 4 = as directrix.

I 0(g) Focus at (6, 0) and axis of y as directrix.

j ^ Qi) Focus at (0, — 7) and axis of x as directrix.

3. Tangents. The student will next turn to Chapter IX

and study §§ 1, 2. It is there shown that the slope of the
parabola '
-
\

(1)
y'^ = 2mx *
94 ANALYTIC GEOMETRY

at any one of its points {x^, 2/1) is, in general, given by the
formula

(2) X = 711

2/1

and that the equation of the tangent line at any point (xi, y^
can, without exception, be written in the form

(3) yiy = m{x-\-x,).

Latus Rectum. The phord, PP', of a parabola which passes
throiigh the focus and is perpendicular to the axis is called the
latus rectum (plural, latera recta).
y
Its half-length is found by setting x = m/2
in the equation of the parabola, and solving
i\(Y,»i) fQp i\^Q positive y :

"X y^m.
E F \2J
V Thus the length, PP', of the latus rectum
\ P'
is 2 m.
The tangent at either or P' makes an P
Fig. 6
angle of 45° with the axis of x. For, the
slope of the tangent at P is, from (2) :

m _m _ ^

2/1 ''^

Let E
be the point in which the tangent at meets the P
axis of x. Since FP = m, and 45°, ZFEP= and so EF=m
E lies on the directrix. Consequently, the tangents at and P
P cut the axis of a; at the point of intersection of the directrix
with that axis.
This theorem can also be proved by writing down the equa-
tion of the tangent at P,

and finding the intercept, of this line on the axis of x.

 THE PARABOLA 0^o< 95

EXERCISES
^ (lj_ Find the equation of the tangent to the parabola y- =3 a;

at the point (12, 6). . Ans. a; — 4?/ + 12 = 0.

p L^^ Find the equation of the normal to the same parabola
at the given point.
*
An^. 4 a; + y = 54.
As./ Find the length of the latus rectum of the parabola of
ExTI.
\ Show that the tangents to any parabola at the extremi-
the latus rectum are perpendicular to each other.
'5. Show that the tangent to the parabola ?/- = 4a; at the
piQiny(36, 12) cuts the negative axis of a; at a point whose dis-

tance from the origin is 36.

6. At what point of the parabola of Ex. 5 is the tangent

perpendicular to the tangent mentioned in that exercise ?
^"s. (3V, - i).
7. Show that the two tangents mentioned in Exs. 5 and

6 intersect on the directrix, and that the chord of contact of

these tangents, i.e. the right line drawn through the two points
of tangency, passes through the focus.

_Show that the tangent to the parabola (1) at any point

its the negative axis of a; at a point M whose distance
from the origin is the same as the distance of P from the axis
of y.

/ 9. Prove that the two parabolas,

2/2 = 4a; + 4 and y''' = — 6x-^9,
intersect at right angles. Assume that the slope of the parab-
ola of Ex. 2, § 2, at the point (a;i, yi) is m/yi.

I
IQ^ two parabolas have a common focus and their axes
If
lie^along the same straight line, their vertices, however, being
on opposite sides of the focus, show that the curves cut each
other at right angles.

Optical Property of the Parabola.

4. If a polished reflector,

like the reflector of the headlight of a locomotive or a search-

96 ANALYTIC GEOMETRY

light, be made in the form of a paraboloid of revolution, i.e.

the surface generated by a parabola which is revolved about

its axis, and if a source of light be placed at the focus, the

reflected rays will all be parallel.
This phenomenon is duetto the fact that the focal radius FP
drawn to any point P of the parabola makes the same angle
with the tangent at P as does the
line through P parallel to the axis.
The proof of this property can
be given as follows. Let the
tangent at P:{xi,yi) cut the axis
of X in 31. Then the length of
OM is equal to Xi, by § 3, Ex. 8.

Furthermore, OF = m/2. Hence

the distance from i^f to i<^ is

Fig. 7
MF= Xi + m
But this is precisely the distance of P from D, § 2, and
hence, by the definition of the parabola, it is also equal to FP.
We have, then, that MF=FP. Consequently, the triangle
MFP is isosceles, and
^ FMP = ^ MPF.
But ^F3rP=^SPT,
and the proposition is proved.
The result can be restated in the following

Theorem. The focal 7'adius FP of a x>ardbola at any point

P of the curve and the parallel to the axis at Pmake equal angles
with the tangent at P.

Heat. If such a parabolic reflector as the one described

above were turned toward the sun, the latter's rays, being
practically parallel to each other and to the axis of the reflector,
would, after impinging on the polished surface, proceed along
lines, all of which would pass through F. Thus, in particu-
lar, the heat rays would be collected at F, and if a minute
 THE PARABOLA 97

charge of gunpowder were placed at F, it might easily be

fired.

It is to this property that the focus (German, Brennpunkt)

owes its name. The Latin word means hearth, or fireplace.
The term was introduced into the science by the astronomer
Kepler in 1604.

^^_^^ EXERCISES ON CHAPTER VI

/T 1.h. parabola opens out along the positive axis of y as axis,

ife^ocus is in the point (0, 3) and the length of its latus
rectum is 12. Find its equation. Ans. x^ = VI y.
/^2. A parabola has its vertex in the origin and its. axis along •

kl3jg^xis of X. If it goes through the point — 3), what is its

(2,
equation? Ans. 2?/- — 9a; = 0.
,.-"3. Show that the equation of a parabola with the line x — c
^A as di'rectrix and with the point (c + m, 0) or (c — m, 0) as

locus is

2/2 = 2m{x — c) — m^, or y'*- := — 2m(x — c) — iii^.

Hence prove that every parabola with the axis of x as axis

has an equation of the form : x = ay"- -(- h, where a and h are .

constants, a ^ 0.
^)xr 4. Find the equation of the parabola which has its axis .
q^
along the axis of x and goes through the two points (3, 2),

(_2, -1). Ans. 3a; = 5?/'- -11.

5.Prove that every parabola with an axis parallel to the
axis of y has an equation of the form

y = ax^ -\-bx -\- c,

where a, b, c are constants, a^O.

Suggestion. Find the equation of the parabola which has
the line ?/ = ^' as directrix and the point {I, k + ?7i) or (I, k — m)
as focus.

6. Find the equation of the parabola which has a vertical

axis and goes through the points (0, 0), (1, 0), and (3, G).
Ans. y = x'- — x.
98 ANALYTIC GEOMETRY

^. A circle is tangent to the parabola i/- = a; at the point

(4, 2) and goes through the vertex of the parabola. Find its

equation.
8. What is the equation of the circle which is tangent to
the parabola 2/" = 2 mx at both extremities of the latus rectum ?
Ans. 4ic2 -}- 42/- — 12ma3 + m^ = 0.
Find the coordinates of the points of tangency of the
9.

tangents to the parabola y^- = 2 mx which make the angles 60°,

45°, and 30° with the axis of the parabola. Show that the
abscissae of the three points are in geometric progression, and
that this is true also of the ordinates.

10. Show that the common chord of a parabola, and the

circle whose center is in the vertex of the parabola and whose
radius is equal to three halves the distance from the vertex to
the focus, bisects the line-segment joining the vertex with the
focus.

11. Let N be the point in which the normal to a parabola

at a point P, not the vertex, meets the axis. Prove that the
projection on the axis of the line-segment PN is equal to one
half the length of the latus rectum.

12. On a parabola, P is any point other than the vertex,

and N is the point in which the normal at P meets the axis.
Show that P and N are equally distant from the focus.

13. The tangent to a parabola at a point P, not the vertex,

meets the directrix in the point L. Prove that the segment
LP subtends a right angle at the focus.
14. Show that the length of a focal chord of the parabola
2/- = 2 mx is equal to Xi-\- X2-\- m, where oji, X2 are the abscissae
of the end-points of the chord. Hence show that the mid-
point of a focal chord is at the same distance from the direc-
trix as it is from the end-points of the chord.

Exercises 15-26. The following exercises express properties

of the parabola which involve an arbitrary point on the parab-
ola. In order to prove these properties, it will, in general, be
 THE PARABOLA 99

necessary to make actual use of the equation which expresses

analytically the fact that the point lies on the parabola.
15. An arbitrary point P
of a parabola, not the vertex, is
joined with the vertex A, and a second line is drawn through
P, perpendicular to AP, meeting the axis in Q. Prove that
the projection on the axis of PQ is equal to the length of the
latus rectum.

16. The tangent to a parabola at a point P, not the vertex,

meets the tangent at the vertex in the point K. Show that
the line joining K to the focus is perpendicular to the tangent
at P. .

17. The tangent to a parabola at a point P, not the vertex,

meets the directrix and the latus rectum produced in points
which are equally distant from the focus. Prove this
theorem.
18. Prove that the coordinates of the point of intersection
of the tangents to the parabola y^=2 mx at the points (a^i, 2/1),

(^) 2/2) iiia^y be put in the form

(Vifh Vx + y2 \
V2m' 2 )
Suggestion. To reduce the coordinates to the desired form,
use the equations which express analytically the fact that the
two points lie on the parabola.
19. Show that the intercept on the axis of x of the line join-
ing the points (a^j, Vi), (a^, 2/2) of the parabola ^^ = 2 mx may be
expressed as
_Wl.
2m
By means of the results of the two preceding exercises prove
the following theorems.
20. The point of intersection of two tangents to a parabola
and the point of intersection with the axis of the line joining
their points of contact are equally distant from the tangent at
the vertex, and are either on it or on opposite sides of it.
100 ANALYTIC GEOMETRY

21. Tangents to a parabola at the end-points of a focal

chord meet at right angles on the directrix.
22. If the points of contact of two tangents to a parabola
are on the same side of the axis and at distances from the axis
whose product is the square of half the length of the latus
rectum, the tangents intersect on the latus rectum produced.
23o The end-points of a chord of a parabola, which sub-
tends a right angle at the vertex, are on opposite sides of the
axis and at distances from the axis, whose product is the
square of the length of the latus rectum.
24. The chords of a parabola, which subtend a right angle
at the vertex, pass through a common point on the axis this ;

point is at a distance from the vertex equal to the latus rectum.

25. The distance from the focus of a parabola to the point

of intersection of two tangents is a mean proportional between
the focal radii to the points of tangency.
26. to a parabola at the points P and Q inter-
The tangents
and the normals at P and Q meet in N. Then the
sect in T,
segment TM, where M
is the mid-point of TN, subtends a

right angle at the focus. ,

'fM '-^

^, ... ,

Locus Problems ' '

"''^S^^

27. Show that the locus of a point which moves so that

the difference of the slopes of the lines joining it to two fixed
points is constant is a parabola through the two fixed points.
What are its axis and vertex ?

28. Determine the locus of a point which moves so that its

distance from a fixed circle equals its distance from a fixed

line passing through the center of the circle.

Ans. Two equal parabolas, with foci at the center of the
circle and axes perpendicular to the fixed line.

29. The base of a triangle is fixed and the sum of the trigo-
nometric tangents of the base angles is constant. Find the
locus of the vertex.
 .

CHAPTER VII

THE ELLIPSE

1. Definition. An ellipse is defined as the locus of a point

P, the sum of whose distances from two given points, F and
F', is constant. It is found con- ^ P
venient to denote this constant
by 2 a. Then
(1) FP^F'P = 2a.
It is understood, of course, that P
always lies in a fixed plane pass-
ing through and F' F Fig. 1
The points F and F' are called
the foci of the ellipse. It is clear that 2 a must be greater
than the distance between them.

Mechanical Construction. From the definition of the ellipse

a simple mechanical construction readily presents itself. Let
a string, of length 2 a, have its ends fastened at and F'j and F
let the string be kept taut by a pencil point at P. As the
pencil moves, its point obviously traces out on the paper the
ellipse.
The student it convenient to use two thumb tacks
will find
partially inserted at F and F'. A silk thread can be tied to
one of the thumb tacks and wound round the other so that
itwill not slip. Thus a variety of ellipses with different foci
and different values of a can be drawn.
Let the student make finally one ellipse in this manner, and
draw it neatly.
101
 : ;

102 ANALYTIC GEOMETRY

Center, Vertices, Axes. It is obvious from the definition, —

and the fact becomes more striking from the mechanical con-
struction, —
that the ellipse is symmetric in the line through
the foci. It is also symmetric in the perpendicular bisector
of FF'. Hence it is symmetric, furthermore, in the mid-point,
0, of the line FF'.
The indefinite line through
the foci, F and F', is called the
transverse axis of the ellipse
the perpendicular bisector of
FF', the conjugate axis. The
point is called the center of
the ellipse ; the points A, A',
its vertices.

The line-segments AA' and

BB', which measure the length and breadth of the ellipse,
are known respectively as the major axis and the minor
axis of the ellipse. The word " axes " refers sometimes to
the transverse and conjugate axes, and sometimes to the major
and minor axes, or their lengths, the context making clear in
any case the meaning.
When P is at A, equation (1) becomes
FA -^ F' A = 2a.
But FA = A'F'.
Hence AA' = 2a and OA = a.
Thus
it appears that the length of the semi-axis major, OA,

is a. Let the length of the minor axis be denoted by 2b, and

the distance between the foci by 2 c. Then, from the triangle
FOB, we have
(2) a*^ = 6^ + c\
Note that, of the three quantities a, h, and c, the quantity a
is always the largest.

Eccentricity, All circles have the same shape, i.e. are simi.
lar ; and the same is true of parabolas. But it is not true of
 : :

THE ELLIPSE 103

ellipses. As a measure of the roundness or flatness of an

number, called the eccentricity has been chosen; this
ellipse a ,

number is defined as the radio c/a and is denoted by e

(3) e = ^.
a
Since c is always less than a, it is seen that the eccentricity
of an ellipse is always less than unity :

e < 1.

In terms of a and 5, e has the value

(4) e=y:^^'
a
All ellipses with the same eccentricity are similar, and con-
versely. For the shape of an ellipse depends only on 6/a, the
ratio of its breadth to its length, and since from (4)

=xM!T-
all ellipses for which the ratio h/a is the same have the same
eccentricity, and conversely.
A circle is the limiting case of an ellipse whose foci ap-
proach each other, the length 2 a remaining constant. The
eccentricity approaches 0, and a circle is often spoken of as an
ellipse of eccentricity 0.

EXERCISES
1. The semi-axes of an ellipse are of lengths 3 cm. and 5 cm.
Find the distance between the foci, and the eccentricity.
Ans. 8; 4.

2. The eccentricity 6f an ellipse is | and the semi-axis

minor is 4 in. long. How long is the major axis ?
3. The major axis of an ellipse is twice as great as the
^inor axis. What is the eccentricity of the ellipse ?
4. The major axis of an ellipse is 39 yards, and the eccen-
tricity, y\. Find the minor axis.
 —
104 ANALYTIC GEOMETRY

5. Express the eccentricity of an ellipse in terms of h and c.

Show, from Fig.

6. 2, that the eccentricity is given by the
formula
e^GOsOFB.
Give a proof, based on similar triangles, that two
7.
ellipses

having the same eccentricity are similar.

2. Geometrical Construction.
.
Points on the ellipse may be
obtained with speed and accuracy by a simple geometrical
construction. Draw the major
"^ "^ axis and mark the points A, F,
F', A' on it. Mark an arbitrary

^'
, ,

^'
2
^
—
i

^
e_ point Q between F and F\
With F as center and AQ a,s
.
,
radius describe a circle, and
^ ^ with F' as center and A'Q as
Fig. 3 i •
i
radius describe a second circle.
The points of intersection of these two circles will lie on the
ellipse, since the sum of the radii is

AQ + A'Q=2a.
It is, of course, not necessary to draw the complete circles,

but only so much of as to determine their points of in-

them
tersection. Moreover, four points, instead of tivo, can be ob-
tained from each pair of settings of the compasses by simply
reversing the roles of F and F\

EXERCISES

X 1- Construct the ellipse for which c — 2\ cm., a = 4 cm.

2. Erom the ellipse just constructed make a templet,

with
holes at the foci and with the axes properly drawn.
"^
. 3. Construct the ellipse whose axes are 4 cm. and 6 cm.

3. Equation of the Ellipse. It is natural to choose the axes

of the ellipse as the coordinate axes (Eig. 4). Let the foci lie
 :

THE ELLIPSE 105

on the axis of x, and let F : {x, y) be any point of the ellipse.

Then, from (1), § 1, .

y _

Fig. 4

(1) V(a;-c)2+2/2 + V(x' + c)2+ y^ = 2a.

Transpose one of the radicals and square

(x — cy-\- 2/2= (a; + 0)2+ 2/^ — 4 a^{x -j- c)2 + 2/2 + 4 a'.

Hence
(2) a V(aj + c)2 + 2/2 = a^ -f ca;.

To remove this radical, square again :

(3) a2a;2 + 2 a'ca; + a2c2 ^ ^2^2 = ^4 2 ^2^0; -f _|_ c2a;*^,

or (a2 — c2)a;2 a^y"^ = a2(a2 — c2).

-\-

But, by (2), § 1,
106 ANALYTIC GEOMETRY

however, we extract a square root we obtain two equations

each time, and so we are led, finally, to the four equations

± V(a;-c)2+2/2± V(a: + c)2+2/2 = 2 a,

the ambiguous signs being chosen in all possible ways. The
four equations can be characterized as follows :

i) + +; ii) + -;
iii) - H- ;
iv) - -.
We wish to show that i) is the only possible one of the four
equations. This is done as follows.
Equation iv) is satisfied by no pair of values for x and y,
since the left-hand side is always negative and so can never be
equal to the positive quantity 2 a.
Equations ii) and iii) say that the difference of the distances
of from F and F' is equal to 2 a, and hence greater than
{x, y)

the lineFF'=: 2 c. Thus, in the triangle FPF' the difference

of two sides is greater than the third side, and this is absurd.*
Hence equations ii) and iii) are impossible and equation, i)
alone remains, q. e. d.
Consequently, if we start with equation (5) as given and
require that a 6, (5) > then represents an ellipse with
semi-axes a and h and foci in the points (± c, 0), where
c = Va2^^^.
The Focal Radii. Erom equation (2) we obtain a simple
expression for the length of the focal radius, F'P. Dividing
(2) by a and remembering that c/a = e, we have :

V(x H- c)2 -f 2/- = a -{-ex.

But the value of the left-hand side of this equation is precisely
F'P. Hence
(6) F'P = a + ex.

* If, in particular, the point (x, y) lay on FF', we should not, it is

true, have a triangle. But it is at once obvious that in this case, too,
equations ii) and iii) are impossible.
 : ^

THE ELLIPSE 107

If, in transforming (1),the other radical had been transposed

to the right-hand side and we had then proceeded as before,
we should have found the equation

a^{x — cy + y^ = a2 _ ex.
From this we infer that

V(a; — c)2 -f 2/2 = a — ex^

or
(7)
'\
,
FP=a- ex.
r '
EXERCISES
^ What is the equation of the ellipse whose axes are of

lengths 6 cm. and 10 cm. ? ,

^- ^ *.^. Ans. — -f ^ = 1.
^ O vX - 25 9
2. Find the coordinates of the foci of the ellipse of Ex. 1.
'

^C_§,^ The foci of an ellipse are at the points (1, 0) and (—1, 0),
X'land the minor axis is of length 2. Find the equation of the ^-

ellipse. Ans. x^ -{2rf = 2.

(4. Find the lengths of the axes, the coordinates of the foci,
and the eccentricity of the ellipse ^
2oa;2 + 169^/2 = 4225.
5y An ellipse, whose axes are of lengths 8 and 10, has its
center at the origin and its foci on the axis of y. Obtain its

equation.

, ^. Show that, ii B> A, the equation f

^ + ^=1 -^
A^ B^
still represents an ellipse with its axes lying along the axes of
coordinates; but the foci lie on the axis of y at the points
(0, C) and (0, - Q, where
B'~ = A--{- C\
The eccentricity is

e —C
B

.. \
 : :

^^ ANALYTIC GEOMETRY

7. Find the lengths of the axes, the coordinates of the foci,

and the value of the eccentricity for each of the following

ellipses: ^ ^/^:,^, A.. ,^,,.
(a) 9a;2 4-4^2^36; (d) 53^24.3^2^45.
(6) 3a;2 + 22/2 = 12; "
(e) 2x^ + lif = 10',
(c) ^a;2 + 22/2= 4; '

(/) lla;2+ 2/'= 3.

4. Tangents. The ellipse has the remarkable property that

the tangent to the curve at any point makes equal angles with
the focal radii draivn to that point

i) Mechanical Proof. The simplest

proof of this theorem is a mechanical
one. Think of a flexible, inelastic string
of length 2 a with its ends fastened at
the foci, F and F'. Suppose a small,
smooth bead to be threaded on this string. Let a cord be
fastened to the bead and then pulled taut, so that the cord
and the two portions of the string will be under tension.
Evidently, the bead can be held in this manner at any point.
(No force of gravity is supposed to act. The strings and bead
may be thought of as resting on a smooth horizontal table.)
The forces that act on the bead are
(a) the tension S in the cord ;

two equal tensions, i?, in the string,

(6)
directed respectively toward the foci.*
Draw the parallelogram of forces for the
forces R. It will be a rhombus, and so
the resultant of these forces will bisect Fig. 6
the angle between the focal radii.
On the other hand, the force S, equal and opposite to this
resultant, is perpendicular to the tangent at P. In fact, if

* Since the bead is smooth, the tension in the string is the same at all
its points, and so, in particular, is the same on the two sides of the bead.
 THE ELLIPSE liS

instead of the flexible string we had a smooth rigid wire, in the

form of the ellipse, for the bead to slide on, the bead would be
held at P by the cord exactly as before. But the reaction of
a smooth wire is at right angles to its tangent. This is the
very conception of a smooth wire. For otherwise, if S were
oblique, it could be resolved into a normal and a tangential
component. But the smooth wire could not yield a reaction,
part of which is along the tangent.
It follows, then, that the normal at P bisects the angle be-
tween the focal radii, and hence these make equal angles with
the tangent at P, q. e. d.

ii) Proof by Means of Minimum Distances. A Lemma. A

barnyard is bounded on one side by a straight river. The
cows, as they come from the pasture,
enter the barnyard by a gate at A,
go to the river to drink, and then
keep on to the door of the barn at B. —
What point, P, of the river should a
^
cow select, in order to save her steps p^^ ^
so far as possible ?
It easy to answer this question by means of a simple con-
is

struction. From B drop a perpendicular BM

on the line of
the river bank, L, and produce it to B', making MB' = BM.
Join A with B',and let AB' cut L at C. Then C is the posi-
tion of P, for which the distance under consideration,
AP-\-PB,
is least.

For, the straight line AB' is shorter than any broken line
APB':
AB'<APB\
But PB = PB' and CB=CB'.
Hence
AB' =zAC+CB and APB' = AP + PB.
It follows, then, that

AC -\- CB < AP + PB,

110 ANALYTIC GEOMETRY

if P is any point of L distinct from C. Hence C is the point

for which APB is a minimum.
The point C is evidently characterized by the fact that

We can state the result, then, by saying that the point P, for
ivhich the distance APB is least, is the point for ivhich
^ APN = ^ BPM.
Optical Interpretation. We have used a homely example of
cows and a barnyard. The problem we have solved is, however,
identical with the optical problem of finding the point at
which a ray of light, emanating from A^ will strike a plane
mirror L, if the reflected ray is to pass through B. For, the
law of light is, that it will travel the distance in the shortest
possible time, and hence it will choose the shortest path.

Application to the Ellipse. The application of this result

to the ellipse is as follows. The tangent to any smooth, closed,
convex curve evidently is characterized
by the fact that it meets the curve in
one, and only one, point.
Let P be any point of the ellipse.
Draw the tangent, T, at P. Let Q be
any point of T distinct from P. Now
F'P + FP= F'R 4- FR,
since the sum of the focal radii is the same for all points of an
ellipse. But
FP<RQ-\-FQ
and so
F'R + FR< F'R ^RQ-^FQ = F'Q + FQ.
Therefore
F'P-\-FP<F'Q + FQ.
Hence P is that point of T for which the distance F'QF is
least,and consequently the lines F'P and FP make equal
angles with T, q. e. d.
 THE ELLIPSE 111

EXERCISE
Show that the normal of an ellipse at any point distinct from
the vertices A, A' cuts the major axis at a point which lies
between the foci.

5. Optical and Acoustical Meaning of the Foci. Let a thin

strip of metal, — say, a strip of brass a yard long and a quarter
of an inch wide, — be bent into the form of an and ellipse
polished on the concave side. Let a light be placed at one of
the foci. Then the rays, after impinging on the metal, will
be reflected and will come together again at the other focus,
which will, therefore, be brilliantly illuminated.*
The same is true of heat, since heat rays are reflected from
a polished surface by the same law as that of light rays. If,
then, a candle is placed at one focus and some gunpowder at the
other, the powder can be ignited by the heat from the candle.
Sound waves behave in a similar manner. The story is told
of the Eatskeller in Bremen, the walls of which are shaped
somewhat like an ellipse, that the city
fathers were remarkably well informed
concerning the feelings and views of the
populace. For, the former drank their Fi^ 9
wine at a table which was situated at a
focus, and thus could hear distinctly the conversation at a dis-
tant table, which stood at the other focus and about which the
Burger congregated.

6. Slope and Equation of the Tangent. The student will

next turn to Ch. IX, § 2, where the slope of the ellipse

* The statement is, of course, strictly true only for such rays as travel
in the plane through the foci, which is pei-pendicular to the elements of
the cylinder formed by the polished band. Since, however, only a nar-
row strip of this cylinder is used, other rays will pass very near to the
second focus and contribute to the illumination there.
 112 ANALYTIC GEOMETRY

at the point {xi, ?/i) is found to be

(1)

The equation of the tangent line at this point is

(2)
^a2
,
M=i.
62

j Latus Rectum. The latus rectum of an ellipse is defined as

a chord perpendicular to the major axis and passing through

a focus. The term is also used to mean
the length of such a chord.
Thus, in the ellij)se

X"

25 16
Fig. 10
one focus is at the point (3, 0). The length
of the latus rectum is twice that of the positive ordinate
corresponding to this point. Setting, then, x = 3 in the equa-
tion of the curve and solving for that ordinate, we have

9
1:=!--^ = 16
16 25 25'

Hence the length of the latus rectum is 6|.

EXERCISES
1. Find the equation of the tangent to the ellipse

oi'-
+.4=1
225 25
at the point (9, 4). Ans. aj + 4?/ = 25.
2. Find the equation of the normal 'to the ellipse of Ex. 1
at the same point. Ans. 4:X— y= 32.
3. At what point does the tangent to the ellipse
2x'-\-3y''' =U
at the point (—1, 2) cut the axis of y?
 THE ELLIPSE 113

At what angle does the straight line through the origin,

rhich bisects the angle between the positive axes of coordinates,
cut J;he ellipse Sx^ -\- 4:y^- = 7? Ans. 81° 53'.
/^ 5. Find the area of the triangle cut off from the first quad-
rant by the tangent to the ellipse of Ex. 3 at the point (1, 2).
Alls. 8Jl.

6. Find the length of the latus rectum of the ellipse of

Ex. 1. Ans. 3^.

7. The same for the ellipse of Ex. 3.

8. Show that the length of the latus rectum of the ellipse

b <a,
a'

is given by any one of the expressions

2&2
2&Vl d2. 2 a(l - e2).
a
Find its value in terms of c and e.

9. Find the length of the latus rectum of the ellipse

25 x^ + 16 ?/2 = 400. Ans. 6|.

10. Prove that the minor axis of an ellipse is a mean pro-

portional between the major axis and the latus rectum.

7. A New Locus Problem. Given a line D and a point F

distant m from D. To find the locus of a point P such that
the ratio of its distance FP from F to
its distance from MP D is always

equal to a given number, e :

FP = e, or FP=eMP.
(1)
MP
It is understood that P shall be re-
stricted to the plane determined by F
and D.
If, in particular, e = 1, the locus Fig. 11
 :

114 ANALYTIC GEOMETRY

is a parabola with D as directrix and F as focus ; Ch.

VI, § 1.

To treat the general case, let D be taken as the axis of y

and let the positive axis of x pass through F. Then
FF = V{x — m)2 -I- i/, MP =±x,
the lower sign holding only when x is negative, and (1) be-
comes
(2) ^(x-my-\-y^ = ±ex.
On squaring and transposing we obtain the equation

(3) (1 - e2)a;2 -2mx-\-y^-{- m^ = 0.

This is the equation of the proposed locus.
The student will now turn to Ch. XI and study carefully
§1.
EXERCISES
/IT) Take = i and m = 3, the unit of length being 1 cm.
e

With and compasses construct a generous number of

ruler
points of the locus,* and then draw in the locus with a clean,
firm line.

\2\ Work out the equation of the locus of Ex. 1 directly,

using the method of the foregoing text, but not looking at the
formulas. Ans. 3 x'^ 4- 4 ?/2 _ 24 x -f 36 = 0.
3. Take c =| and m= 4:, the unit of length being 1 cm.
Draw the locus accurately, as in Ex. 1.

4. Work out directly the equation of the locus of Ex. 3.

Ans. 16 a;2 -\-2oy^- 200 x =- 400.

5. By means of a transformation to parallel axes show that

the curve of Ex. 2 is an ellipse whose center is at the point

(4, 0) and whose axes are of lengths 4 and 2V3. What is its
eccentricity ?

* The details of the construction are an obvious modification of the

corresponding construction for the parabola in Ch. YI, § 1. A circle of

arbitrary radius is drawn with its center at F, and this circle is cut by a
parallel to D, whose distance from D is twice the radius of the circle.
 ::

THE ELLIPSE 115

6. Show that the curve of Ex. 4 is an ellipse whose axes

are 7^ and 6. What is its eccentricity ?

;
I
J. Discussion of the Case e < 1. The Directrices. From
A^ equation (3) of § 7 follows

— 2 m y' m'
(1) if2
1-e-
:X +
l-€2
The first two terms on the left-hand side are also the first two
in the expansion of

X — ^ 2m
^^=^^2__2i/L^4. ^'
1-^ (1 - £2)2
If, then, we add the third term of the last expression to both
sides of (1), we shall have
9 ^t^
+
m^ y2 _ m^ rri'
a;2—
l-e2
X +
(1- €2)2^1-^2 (l-€2)2 1 _ '
e2
or

(2) X — m + f_ _ e2m2
1 - e2y 1 - e2 (^1 _ ^2)2
This equation reminds us strongly of the equation of an
ellipse. In fact, if we transform to parallel axes with the
new origin, 0', at the point

Xn
m = 0,
2/0
1-^2'
the equations of transformation are

(3) x' =X— m

and (2) then takes on the form

y
n £2m2
(4) a;'2 + l_e2 (l-e2)2'
or
./•I

(5)
a' 62
where Fig. 12

(6) a = -
:j 6 = £7?l

1 -e2 vr
 :

116 ANALYTIC GEOMETRY

Thus the locus is seen to be an ellipse with its center, 0\ at

the point
m ,0
(7) ,
l-€2
the semi-axes being given by (6).

The value of c is given by the equation c^ = a? — W. Hence

€-m
(8)
l-e2
The eccentricity, e = c/a, is now seen to be precisely e

e = e;

i.e. the given constant, e, turns out to he the eccentricity of the

ellipse.

Finally^ F is one of the foci. For, the distance from F to 0'

is

00'-0F = m — m = e^m
l-£2 l-e2'
and this, by (8), is precisely c.

The line D is called a directrix of the ellipse. Its distance

from the center is

00' = m _ me 1 _a
l-€2~l-e2 e~7'
The Directrices. From the symmetry of the ellipse it is

clear that there is a second directrix, D', on the other side of

the conjugate axis, parallel to that
axis, and at the same distance
from it as D. This line D' and
X the focus F' stand in the same re-
lation to the ellipse as the first

line, D, and the focus F. Thus

Fig. 13 the ellipse is the locus of a point
so moving that its distance from a
focus always bears to its distance from the corresponding
directrix the same ratio, e, the eccentricity.
Since the distance of D from the center of the ellipse is a/e,
the equations of the directrices of the ellipse
 : e

THE ELLIPSE 117^,

x"^

are

e e

EXERCISES
P.'
t> 1. Show that the distances of the vertices, A and A' from
^

J' O are

1 H-e 1 —
/'^2. Collect the foregoing results in
a syllabus, arranged in
tabular form, giving each of the quantities a, 6, c, 00', 0^,
0^', OF, OF' in terms of 7?i and e.
3. Work out each of the quantities of Ex. 2 directly for the
ellipse of § 7, Ex. 4, and verify the result by substituting the
values e = =
I, 771 4 in the formulas of the syllabus. - '

4. Between the jive constants of the ellipse, a, 6, c, e, 7?i,

there exist three relations, which may be written in a variety

of ways ; as, for example,

i)a^ = 52 + c^ ii) e = ^; iii) m = i-:^a.

a e

By means of these relations, any three of the five quantities

can be expressed in terms of the other two. Thus, in Ex. 2,
m and e are chosen as the quantities in terms of which all
others shall be expressed.
Taking the semi^axes, a and 6, (a > 6), as the preferred pair,
the other quantities in terms of them.
^how that the tangent to the ellipse

25 16

at an extremity of a latus rectum cuts the transverse axis in

the san^e point in which this axis is cut by a directrix.
 ANALYTIC GEOMETRY

©The same for any ellipse.

Prove directly that, if P is any point of the ellipse

X 2/^_
-.+?.=;^>
6^
b < a,
a^

the ratio of its distance from a focus to its distance from the
corresponding difectrix is equal to the eccentricity.

( S^Showthat in an ellipse the major axis is a mean propor-

tionalbetween the distance between the foci and the distance
between the directrices.
9. Show that the distances from the center and a focus of
an ellipse to the directrix corresponding to the fqcus are in
the same ratio as the squares of the semi-axis major and the
semi-axis minor.

9. The Parabola as the Limit of Ellipses. We have proved

that, when c < 1, equation (3), § 7, represents an ellipse with
eccentricity e = e. We
know that , if e = 1, the
equation represents a
—
. parabola. If, then, in the
equation we allow e to
approach 1 through values
< 1, the ellipse which the
equation defines ap-
proaches a parabola as its

limit.
We can visualize the
ellipse, going over into
a. parabola, by drawing
a number of ellipses
Fig. 14
having the same value of
m, but having values for which are increasing toward 1
c

as their limit, viz. e = |^, —.

e = -J-,
e =f
The directrix D, ,

along the axis of y, and the focus F: (m, 0) are the same for
all the ellipses. But the center 0' and the right-hand vertex
.
 THE ELLIPSE 119

A' of each successive ellipse are farther away from' O, and

their distances from 0, namely,

00' = m OA' = m
1- 1^
increase without limit. Thus, as e approaches 1, the ellipse
approaches as its limit the parabola whose directrix is and D
whose focus is F.

10. New
Geometrical Construction for the Ellipse. Para-
metric Representation. Let it be required to draw an ellipse
when its axes, AA' and BB', are given.
Describe circles of radii a = OA and
b =OB, with the origin as the
common center. Draw any ray from
0, making an angle <^ with the posi-
tive axis of X, as shown in the Z
figure. Through the points Q and B
draw the parallels indicated. Their
point of intersection, P, will lie on
the ellipse. For, if the coordinates of Fig. 15
Pbe denoted by (x, y), it is clear that
\
\
(1) x = a cos (f), y =h sin <^.

From these equations <j> can be eliminated by means of the

trigonometric identity
sin2 (ji -f cos2 (^ = 1. y
Hence
(2) ^ + ^ = 1.
a2 62

Conversely, any point {x, y) on the ellipse (2) has corre-

sponding to an angle <^, for which equations (1) are true.

it

Equations (1) afford what is known as a parametric repre-

sentation of the coordinates of a variable point {x, y) of the
terms of the jyarameter <^. When b
ellipse in = a, the ellipse
becomes a circle, and the equations (1) become

(3) x = a cos- (f>, y = a sin <^.

 120 ANALYTIC GEOMETRY

These parametric representations, though little used in Ana-

lyticGeometry, are an important aid in the Calculus.
The larger of the two circles in Fig.
15 is commonly called the auxiliary
^:{z,ij) circle of the ellipse, and the points R
and P are
known as corresponding points.
The angle ^ is called the eccentric angle.

EXERCISE
Fig. 16
By means of the foregoing method,
draw on squared paper an ellipse whose
axes are of length 4 cm. and 6 cm.

EXERCISES ON CHAPTER VII

rlj)The earth moves about the sun in an elliptic orbit.* The

shortest and longest distances from it to the sun are in the
ratio 29 : 30. What is the eccentricity of the orbit ?
2y^Show that the slopes of the tangents to an ellipse at the
esoemities of the latera recta are ± e.

^ 3. The axes of an ellipse which goes through the points

(4, 1), (2, 2) are the axes of coordinates. Find its equation.
4. The center of an ellipse is in the origin and the foci are

on the axis of x. The ellipse has an eccentricity of |- and goes

through the point (12, 4). What is its equation ?

Ans.
. x"
25
,

h --
1/2

16
—
= 169
25
•

\h. >Bolve the preceding problem if the foci may lie on either
coordinates. \j ^
Find the equations of the ellipses which have the axes
'6>r

of coordinates as axes, go through the point (3, 4), and have

tlieir major and minor axes in the ratio 3 2. :

S^o"^ ^2it the ellipses represented by the equation

^/
'^'
I

2a;^ + 3?/2=c\
* The planets describe ellipses about the sun as a focus, and the comets
usually describe parabolas with the sun as the focus.
 THE ELLIPSE 121

where c>is an arbitrary positive constant, are similar. What

is the^ommon value of the eccentricity ?
8. How many ellipses are there with eccentricity i, having
their centers in the origin and their foci on the axis of a:?
Deduce an equation which represents them all.

Ans. 3x2-f-42/-=-c2.
9. The foci of an ellipse lie midway between the center
and the vertices. What is the eccentricity ? How many such
ellipses are there, with centers in the origin and foci on the
axis of X? Write an equation which represents them all.
V40. The line joining the left-hand vertex of an ellipse with
the upper extremity of the minor axis is parallel to the line
joining th^center with the upper extremity of the right-
hand h^s rectum. Answer the questions of the preceding
exer/glse.

yih The foci of an ellipse subtend a right angle at either

I
extremity of the minor axis. What is the eccentricity?
Find the equation of all such ellipses with centers in the
origin and foci on the axis of y.

(12. Prove that the ratio of the distance from a focus of an

ellipse to the intersection with the transverse axis of the
normal at a point P, and the distance from this focus to P
equals the eccentricity of the ellipse.

13. The projections of a point P of an ellipse on the trans-

verse and conjugate axes are P^ and Pg- The tangent at P
meets these axes in Ti and T2. Prove that OPj OT^ = a? and
•

OP2 OTz — b"^, where

' is the center and a and b are the
semi-axes of the ellipse.

14. Prove that the segment of a tangent to an ellipse be-

tween the point of contact and a directrix subtends a right
angle at the corresponding focus.

15. Determine the points of an ellipse at which the tangents

have intercepts on the axes whose absolute values are propor-
tional to the lengths of the axes.
 122 ANALYTIC GEOMETRY

16.Through a point Mot the major axis of an ellipse a line

is drawn parallel to the conjugate axis, meeting the ellipse in
P and the tangent at an extremity of the latus rectum in Q.
Show that the distance MQ equals the distance of P from the
focus corresponding to the latus rectum taken.
17. Prove that the line joining a point P of an ellipse with
the center and the line through a focus perpendicular to the
tangent at P meet on a directrix.
Prove that the distance from a focus P to a point P of
18.
an ellipse equals the distance from F to the tangent to the
auxiliary circle at the point corresponding to P.

19. Find the equation of a circle which is tangent to the

ellipse

^ + •^ = 1
at both ends of a latus rectum.

[ 20.In an ellipse whose major axis is twice the minor axis,

a' line of length equal to the minor axis has one end on the
The two ends are
V
ly ellipse, the other on the conjugate axis.
always on opposite sides of the transverse axis. Prove that
the mid-point of the line lies always on the transverse axis.
21. A
number of ellipses have the same major axis both in
length and position. A tangent is drawn to each ellipse at
the upper extremity of the right-hand latus rectum. Prove
that these tangents all pass through a point.

Exercises 22-28. In these exercises, in which properties in-

volving an arbitrary point P of an ellipse are to be proved, it
will, in general, be necessary to make actual use of the equa-
tion expressing the fact that the point P lies on the ellipse.

22. The tangent to an ellipse at a point P meets the tan-

gent at one vertex in Q. Prove that the line joining the other
vertex to P is parallel to the line joining the center to Q.

23. The lines joining the extremities of the minor axis with
a point P of an ellipse meet the transverse axis in the points
 THE ELLIPSE 123

Jf and N. Prove that the semi-axis major is a mean propor-

tional between the distances from the center to 3f and N.
24. Prove the theorem of the preceding exercise when the
major and minor axes, and the transverse and conjugate axes,
are interchanged.

25. Show that the segment of a directrix, between the

points of intersection of the lines joining the vertices with a
point on an ellipse, subtends a right angle at the correspond-
ing focus.
26. Prove that the product of the distances of the foci of
an ellipse from a tangent is a constant, independent of the
choice of the tangent.
Let F' and F be the foci of an ellipse and
27. P any point
on it.Prove that b"^ FK^ = F'P FP, where
: : FK is the dis-
tance from F to the tangent at P.
28. The normal to an ellipse at a point P meets the axes in
^1 and ^2- Show that PJSfi • PN.2 is equal to the product of
the focal radii to P.
Loci
29. A point moves so that the product of the slopes of the
two lines joining it to two fixed points is a negative constant.
What is its locus ?

30. A circle whose diameter is 10 cm. is drawn, center at 0.

On a radius OA sl point B
marked distant 4 cm. from 0.
is

If OQ is any second radius, show how to construct, with ruler-

and compasses, a point P on OQ, whose distance from the
circle equals its distance from B. In this way plot a number
of points on the locus of P.

31. Find the equation of the locus of the point P of the

preceding exercise.Take the origin of coordinates at the
mid-point of OB.
32. The base of a triangle is fixed and the product of the
tangents of the base angles is a positive constant. Find the
locus of the vertex.
 ;

CHAPTER VIII

THE HYPERBOLA
J)efinition. A hyperbola is defined as the locus of a point
P, the difference of whose distances from two given points, F
and F', is constant. It is found
convenient to denote this constant
by 2 a. Then
FP-F'P^2a,
or F'P-FP=2a.
It is understood, of course, that P
Pjq j
is restricted to a particular plane
through F and F'.
The points F
and F' are called the foci of the hyperbola.
It is clear that 2 a must be less than the distance between
them. Denote this distance by 2 c.

Geometrical Construction. Draw the indefinite line FF\

mark the mid-point, 0, of the segment FF'^ and the points A
and A' each at a distance a
from O:
0^ = 0^' =a ; 0F= OJ^' = c.

The point ^ lies on the locus

for,

FA = c-a, F'A = c-\-a, ^^^ 2

and hence F'A— FA = 2 a.
Likewise, A' lies on the curve.
Mark any point, N, to the right of F. With radius AN and
center F, describe a circle. Next, with radius A'N and center
124
 THE HYPERBOLA 125

F', describe a second circle. The points P and Q in which

these circles intersect are points of the locus. For,

F'P-FP= A'N- AN^ A' A = 2 a.

Two more points, P' and Q', can be obtained from the same
pair of settings by interchanging the centers, F and F', of the
circles.

By repeating the construction a number of times, a goodly

array of points of the hyperbola can be obtained. These
points will lie on two distinct arcs,
symmetric to each other in the
perpendicular bisector BOB' of
FF'. Thus it will be seen that
the hyperbola consists of two F'A O A\F
parts, or branches, as they are
called. These branches, besides
being the images of each other in Fig. 3
BB', are each the image of itself
in FF'. It is natural to speak of the indefinite straight lines
FF' and BB' as the axes of the hyperbola. FF' is called the
transverse, BB' the conjugate axis O is the center, and A, A'
;

are the vertices.

EXERCISES
1. Taking c =3 cm. and a =2 cm., make a clean drawing of
the corresponding hyperbola.
2. Reproduce the drawing on a rec-
tangular card and, with a sharp knife
or a small pair of scissors, cut out the
center of the card along the hyperbola
and two parallels to the transverse axis.
On the templet which remains make
Fig. 4 holes at the foci and draw the two axes.

2. Jlquation of the Hyperbola. The treatment here is paral-

lel to that of the ellipse,"Ch.^¥"II, 8 3r Let the transverse axis
 :: : :

126 ANALYTIC GEOMETRY

be chosen as the axis of x the conjugate axis, as the axis of y.

;

Then the equation of the right-hand branch of the hyperbola

can be written in the form
P:(x,v)
(1) V(a;+c)2-l-y2_V(a;-c)2 + ?/2 = 2a.
F:(-c,o) F:(c,0) Transpose the first radical and square
(x — c)2-f- y^=(x-\- c)2+ ?/2

— 4 aV(x + c)2+ + 4 a\ 2/2

FiQ. 5
Hence

(2) a^/(x + c)2+ 2/2 = a2 4- ex.

Square again
a^x"^ + 2 a^cx + a^c^ -f ^22/2 = ^t* + 2 a2ca; + c2a;2,

or
(3) (a2 - c^)a;'^ + aY'= a'^(a''- - c^).
This is precisely the same equation that presented itself in

the case of the ellipse ; but the locus is a curve of wholly dif-
ferent nature. The reason
that a and c have different
is,

relative values. a was greater than c, and hence

In the ellipse,
q2. _ (.2 ^as positive. It could be denoted by If-. Here, a is
less than c a2 — c- is negative, and it cannot be set equal to &2.
;

It can, however, be set equal to — Ir. This we will do

(4) a2-c2 = -&2^ or c2 = a2 + 62^

thus defining the quantity & in the case of the hyperbola by
the equation
?)=Vc2-a2.
The final equation between x and y can now be written in
the form

(5)
a2 62
~
This equation by the coordinates of all points on
is satisfied

the right-hand branch, as seen from the way in which it

is

was deduced. It is, however, also satisfied by the coordinates

of all points on the left-hand branch. For such a point, the
 :

THE HYPERBOLA 127

r

signs of both radicals in (1) will be reversed. Starting, now,

with the new equation and proceeding as before, we find the
same equation (3), which we may again write in the form (5),
and thus the truth of the statement is established.
Is (5) satisfied by the coordinates of still other points ? To
answer this question, let (a;, y) be any point whose coordinates
satisfy (5). Then, starting from (5), we retrace our steps,
admitting, each time that we extract a square root, both signs
of the radical as conceivably possible. Thus we can be sure
that (ic, y) will satisfy one of the four equations

± V(x + c)'^+ 2/2± V(a; -c)^-f 2/2 = 2a,

corresponding to the four conceivable choices of the signs of

the radicals
i) - -f ;
iii) - - ;

ii) + -; iv) + +.
If {x, y) satisfies i) or ii), the point lies on the hyperbola.
The other two cases are impossible. For, case iii) says that a
negative quantity is equal to a positive quantity, and case iv)
says that F'P + FP=2a. Now F'P+ FP, being the sum of
two sides of the triangle FPF', is greater than the third side,
FF', or 2 c. But 2 a is actually less than 2 c. Hence we have
a contradiction, and this case cannot arise.
We have shown then, finally, that (5) is the equation of the
hyperbola.
EXERCISE
Plot the hyperbola

25 16
\
directly from its equation, taking 1 cm. as the unit of length.

\
Sy^j^ev Eccentricity, Focal Eadii. The transverse and the
^conjugate axis have already been defined in § 1. The segment
AA' of the transverse axis is called the major axis, and this
term is also applied to its length, 2 a. The segment BB' of
the conjugate axis, whoge center is at and whose length is
 : •

128 ANALYTIC GEOMETRY

2 b, is called the minor axis, and this term is also applied to

its length, 2 b.

The major axis of an ellipse is always longer than the minor

axis. In the case of the hyperbola, however, this is not al-
ways true. For example, if 2 c and 2 a
are taken as 10 and 6 respectively, then
2 6 = 8. Thus the major axis of the
hyperbola is to be understood as the
principal axis, but not necessarily as the
longer axis.
The eccentricity of the hyperbola is defined as the number
c
e =—
a
Since c is greater than a, the eccentricity of a hyperbola is

always greater than unity.

The eccentricity characterizes the shape of the hyperbola.
All hyperbolas having the same eccentricity are similar, differ-
ing only in the scale to which they are drawn, and conversely ;

cf. Exercise 8. f

The focal radii FP, F'P can be represented by simple ex-

pressions, similar to those which presented themselves in the
case of the ellipse. On -dividing equation (2), § 2, through by
a, we have : ^

-"
/#^K V(a^ -^ cf -{-y' = a + ex.
Hence, when P is a point of the right-hand branch,

(1) ^^ F'P = eo>-{-a.

The evaluation,
(2) FP = ex-a,
is obtained in a similar manner.*
If P is a point of the left-hand branch, these formulas
become
(3)^ F'P = -{ex^a)\ FP=-(ex-a).
* P being a point of the right-hand branch, x is positive and greater
than or equal to a ; also, e> 1. Hence ex>a, and ex — a is positive,

as it should be.
/ («)
130 ANALYTIC GEOMETRY

Let a point P
(x, y) move off :

P:(x,y) along a branch of the hyperbola

(1)
^ _ 2/^ _ 1
a

and let this take place, for def-

initeness, in the tirst quadrant.

Fig. 7 The slope of the line OP is

MP ^y
OM X
Since the coordinates (a;, y) of P satisfy (1), it follows that

h
(2) y =- Va;2 — a^,
a
and hence

(3) -=-\l---
^
a; a a?"

When P x increases without liniit, and

recedes indefinitely,
the right-hand side of this equation approaches the liniit h/a.
Thus we see that the slope of OP approaches that of the line OQ,

(4) y = h-x,
a
as its limit, always remaining, however, less than the latter
slope, so that P is always below OQ.
It seems likely that P will come indefinitely near to this
line but this fact does not follow from the
;

foregoing, since P might approach a line

parallel to (4) and lying below it. In that
case, all that has been said would still be true. -X

That P does, however, actually approach

(4) can be shown by proving that the dis-
tance PQ approaches as its limit. Now, Fig. 8

PQ=MQ-MP,
and, from (4),

MQ = -x.
a
 :

THE HYPERBOLA 131

Furthermore, MP is the ^/-coordinate of the point P on the

hyperbola

Hence PQ =z-[x — Va;- — aH.

a
To find the limit approached by the square bracket, we re-
sort to an algebraic device. The value of the bracket will
clearly not be changed if we multiply and divide it by the
expression x + Va^- — a^:
X ^/^^ ^^ = ~ ^'^~ - ^') ^^ + ^^' ~ ^')
^"^

X — a^ -f- '\^x'^

But the numerator of the last expression reduces at onca to a~.

Hence
a2
X — VaJ- a- =
X -h Vaj2 — d^
From this form it is evident that the bracket approaches ^
when X increases indefinitely; and hence the limit of PQ is
zero,* q. e. d.

Similar reasoning, or considerations of symmetry, applied in

the other quadrants, show that in the second and fourth
quadrants P approaches the line

while in the third quadrant, as in the first, P approaches (4).

The equations (4) and (5), of the asymptotes, can also be
written in the form

a 6 a b

* The limit approached by the variable x — v x2 — a^ can be found

geometrically as follows. Construct a variable
right triangle, one leg of which is fixed and of |

length a, the hypothenuse being variable and of

length X. Then the above variable, x — \/x2 — a^,

is equal to the difference in length between the

hypothenuse and the variable leg. This difference obviously approaches
as X increases indefinitely.
132 ANALYTIC GEOMETRY

It is easy to remember these equations, since they can be

written down by replacing the right-hand side of (1) by 0, fac-

toring the left-hand side :

(M)(M)=^'
and putting the individual factors equal to zero.
The slopes of the asymptotes are h/a and — b/a. Conse-
quently, the asymptotes make equal angles with the transverse
axis.
Since the ratio of 5 to a is unrestricted, the asymptotes can
make any arbitrarily assigned angle with each other. If, in

particular, 6 = a, this angle is a right angle, and the curve is

called a rectangular, or equilateral, hyperbola. Its equation can

be written in the form :

(6) x''^ -y^ = a\

Its eccentricity is e = V2.
Construction of the Asymptotes. Mark with heavy lines the
major and minor axes, and through the extremities of each
draw lines parallel to the other,
thus obtaining a rectangle. The
diagonals of this rectangle, pro-
duced, are the asymptotes, since
their slopes are clearly ± b/a.
The diagonals of the rectangle
have lengths equal to the distance
2 c between the foci, for, c^ = a^ _|_ 52
and the lengths of the sides of the rectangle are 2 a and 2 b.
If the acute angle between an asymptote and the transverse
axis is denoted by a, then

e = sec a.

EXERCISES
1. Find the equations and slopes of the asymptotes of the
hyperbolas of Exercise 1, § 3. Draw the hyperbolas.
 THE HYPERBOLA 133

2. Show that the asymptotes of the hyperbola

Ax'- - Bif- = C,

where A, B, and C are any three positive quantities, are given

by the equations

Find the equation of the hyperbola whose asymptotes

3.

make angles of 60° with the axis of x and whose vertices are
situated at the points (1, 0), and (— 1, 0). Aiis. Sx- —y^ S. =
4. Show that the slopes of the asymptotes are given by
the expression ± Ve^ — 1. ^ x.

5. The slope of one asymptote of- ariy.perbola is |. Find

the eccentricity. .^ Ans. e — 1\.
6. The distance of a focus of a certain hyperbola from the
center is10 cm., and the distance of a vertex from the focus is
2 cm. What angle do the asymptotes make with the conju-
gate axis ^^°^'-
?
--—
V--TL"*<u ^i, .^
-^f^'
f v»-^\^ Show that the circle circumscribed about the rectangle
i ^(T the text passes through the foci.

>^. A perpendicular dropped from a focus F on an asym-

yptote meets the latter at E. Show that OE = a, and EF = h.
Find the equation of the equilateral hyperbola whose
~

foci are at unit distance from the center. i - *- ^^

LO. Find the equation of the equilateral hyperbola which

passes through the point (—5, 4).

5. Tangents. The method of finding the slope of an ellipse,

Ch. IX, § 2, can be applied to the hyperbola, and it is thus

shown that the slope of this curve,

£C2 _ 1/2 _ .

at the point {xi, y-^ is

 134 ANALYTIC GEOMETRY

The equation of the tangent of the hyperbola at this point is

Q a^ 62

Theorem. TJie tangent of a hyperbola at any point bisects the

^
angle between the focal radii.

\t^ To prove this proposition we recall the theorem of Plane

w Geometry which says that the bisector of an angle of a triangle

\f divides the opposite side into seg-

ments which are proportional to the
f \ ^^^i'-' (a"i, yi) adjacent sides. seen It is easily
that the converse * of this proposi-
tion is also true, and hence it is
I
sufficient for our proof to show that

^ ^
FM Fm'
Pjq We already have simple ex-
pressions for the numerators. If
-P- (^1) 2/i) t)e a point of the right-hand branch of the curve,
then, by § 3,
FP = exi — a ; F'P = ex^ + a.

To compute the denominators, find where the tangent at P,

whose equation is given by (1), cuts the axis of x. Denoting
the abscissa of 31 by x', we have :

x' = —'
Xi

Now, FM= OF- OM=c- x\

and c — = a^
x'— — a^
= cXi
,
c
a-l X,
But c = ae, and so
cxi — a^ = a{exi— a).

Thus c — x' = ~ (exi — a),

Xi

* Let the student prove this proposition as an exercise.

 THE HYPERBOLA 135

and we arrive finally at the desired expression for FM:

FM=^{ex,-a).
In a similar manner it is shown that

.Tj

From these evaluations it appears that

FP -^ x. F'P X
FM —a
,
and
F'M" a
Hence and the proof is complete for the
(2) is a true equation,
case that P
on the right-hand branch. Since, however,
lies
the curve is symmetric in the conjugate axis, the theorem is
true for the left-hand branch also.

jj Latus Rectum. The latus rectum of a hyperbola is defined

' ^
as a chord passing through a focus and perpendicular to the
\Ay transverse axis. The term is also applied to the length of
such a chord.
1

EXERCISES
1. Find the slope of the hyperbola ^x^ — y^^W at the
point (2, — 1). Arts. — 8.
2. Find the equation of the tangent of the hyperbola of

Ex. 1 at the point there mentioned. Ans. 8 a; -f- ?/ = 15.

3. Find the angle at which the line through the origin bi-
secting the angle between the positive axes of coordinates cuts
the hyperbola of Ex. 1. Ans. 30° 58'.

4. Find the length of the latus rectum of the hyperbola

— — ^ = 1.
r Ans. 4i.
^
16 9 .

5. Find the length of the latus rectum of the hyperbola of

Ex. 1. Ans. 15.49.
6. Find the equation of the normal of the hyperbola

^_f_^ 1
25 144
 ^

136 ANALYTIC GEOMETRY

at the extremity of the latus rectum which lies in the first

quadrant. Ans. 25 x -\- 65 y = 2197.
7. Show that the length of the latus rectum of the hyperbola

^_^=1
252 CL' 62
IS
a
8. Prove that the tangents at the extremities of the latera
recta have slopes ± e.

In an ellipse, the focal radii make equal angles with the

9.

tangent. Prove this theorem by the method employed in this

paragraph to prove the corresponding theorem relating to the
hyperbola.
-
f f « .

\ /^.^New Definition. The Directrices. The locus defined

%, m Ch. VII, § 7, can now be shown to be a hyperbola when
e > 1. The analytic treatment given there and in § 8 down
to equation (2) and the transformation (3) holds unaltered
for the present case.

\^ I
When, however, e > 1, the new origin, 0', lies to the left of

0, in the point (
— m , ), and it is more natural to write

(3) in the form

(1) x' = x-\- m 1' y =y>

and likewise (4) as

i'2.
^ 2/yM2
Xn y___—.
(2)
£2 _ 1
"^2
(g2 _ 1)2
«-s,x'
This equation passes over into the form

(3)
^-yl — \
on setting

(4) a = em b = em
Ve^^
 :

^
THE HYPERBOLA 137

Thus the locus is seen to be a hyperbola with its center, 0',

at the point /
—- — -, j, the semi-axes being given by (4).

The value of c is given by the equation c- = a^ -f 62, Hence

j>

(5) c = -i!^^.
The eccentricity, e =c/a, is seen to be precisely e

e = e,

and thus the given constant, e, turns out to be the eccentricity of the
hyperbola.
Finally, F is one of the foci. For, the distance from 0' to F
is

and this, by (5), is precisely c.

The line D is called a directrix of the hyperbola. Its dis-
tance from the center is

0'0 = m em 1 a
- ±1
— :2
e- €2
€- _1
— € e

Tlie Directrices. There is a second directrix, namely, the

line D' symmetric to D in the conjugate axis. It is clear
from the symmetry of the figure that what is true of the hy-
perbola with respect to the focus F and the corresponding
directrix D is equally true with respect to the focus F' and
the directrix D'. Accordingly, the hyperbola is the locus of a
point whose distance from a focus bears to its distance from
the corresponding directrix a fixed ratio, the eccentricity.
The equations of the directrices of the hyperbola,

a2 62

are x = a- and
J
x = a

*^n /A

n
 :

138 ANALYTIC GEOMETRY

EXERCISES
1. Take c =2
and m=S, the unit of length being 1 cm.
With and compasses construct a generous number of
ruler
points of the locus, and then draw in the locus with a clean,
firm line.*
2. Work out the equation of the locus of Ex. 1 directly,
using the method of Ch. VII, § 7, but not looking at the
formulas. Ans. Sx''- — y''- -{- 6x = 9.
3. By means of a transformation to parallel axes show
that the curve of Ex. 2 is a hyperbola whose center is at the

point (— 1, 0) and whose axes are of lengths 4 and 4V3.

4. Show that in the general case the distances of the
vertices, A and A', from are

0^ = -^^, A'0 = -^'

£-1-1 e-1
5. Collect the results of this paragraph in a syllabus,
arranged in tabular form, giving each of the quantities, a, b, c,

O'O, OA, A'O, OF, and F'O, in terms of m and c.

6. Work out each of the quantities of Ex. 5 directly for
the curve of Ex. 2 and verify the result by substituting the
values c = 2, ?/i =3 in the formulas of the syllabus.

7. Show that the tangent to the hyperbola

^^_^—
16~9~ 1

at an extremity of a latus rectum cuts the transverse axis in

the same point in which this axis is cut by a directrix.

8. The same for any hyperbola.

* The footnote of p. 114 applies in the present case with the obvious
modification that the distance of the parallel from D must now be half
the radius of the circle. Moreover, tioo parallels to D must now be drawn,
the second one, as soon as the radius has increased sufficiently, giving
points on the left-hand branch. » >
 THE HYPERBOLA 139

9. Prove directly that, if P is any point of the hyperbola

^^
_^- .0
1
'
a2 b'-

the ratio of its distance from a focus to its distance from the
corresponding directrix equals the eccentricity.
10. Prove that the ratio of the distance between the foci of
a hyperbola to the distance between the directrices equals the
square of the eccentricity.

7. The Parabola as the Limit of Hyperbolas. Summary.

Equation (3) of Ch. VII, § 7, namely,

(1) (1 - €2)a;2 + 2/^-2 mx + m- = 0,

represents a hyperbola when e > 1 and a parabola when e = 1.

If, then, we let e approach 1 through values greater than 1,

the hyperbola which (1) represents will approach a parabola
as its limiting position.
Suppose, for example, that we take m=2 and let e take on
successively the values 2, l^, IJ, li, •••. Drawing the corre-
sponding hyperbolas, we find that, whereas the directrix D
and the right-hand focus F are always fixed, the center and the
left-hand vertex keep receding to the left, and that their
distances from 0, namely,

'>
£' 1' -^ --,_!'
increase without limit. Thus, when e approaches 1, the left-
hand branch of the hyperbola recedes indefinitely to the left
and disappears in the limit, whereas, meanwhile, the right-
hand branch gradually changes shape and in the limit becomes
the parabola whose directrix is D and whose focus is F.

Summary. Let us now combine the results of § 6 with those

of § 8, Ch. VII. We have proved that equation (1) repre-
sents an ellipse, a parabola, or a hyperbola, according as e< 1,

c = 1, or e > 1. In case of the ellipse and the hyperbola the

140 ANALYTIC GEOMETRY

constant e turned out to be the eccentricity e. We are led

then to give to the parabola an eccentricity, namely,
e = e = 1.
Theorem. The locus of a point ivhich moves so that its dis-
tancefrom a Jixed point bears to its distance from a fixed line,
not passing through the fixed point, a given ratio e is an ellipse,
a parabola, or a hyperbola, according as c is less than, equal to,
or greater than unity. In every case the constant e equals the
eccentricity.

Since always c = e, we may suppress e in future work, and

use e exclusively. Thus equation (1) becomes

(2) (1 - e'')x''' + 2/- - 2 ma; + m2 = 0.

The theorem furnishes a blanket definition for the ellipse,
parabola, and hyperbola, which might have been used instead
of the separate definitions which we have given. It should
be noted, however, that this blanket definition does not include
the circle. For, if we set e = in (2), the equation reduces to

(x — m)-+ ?/2= 0,
which represents merely the focus F : (m, 0).
The fact that the blanket definition does not yield a circle
as- a special case in no way discredits the circle as the limiting
form of an ellipse when the eccentricity approaches zero,
Ch. VII, The reason that a circle cannot be defined in
§ 1.
the new manner is because it has no directrices. When the
eccentricityof an ellipse approaches zero, the major axis
remaining constant, the distance a/e of the directrices from
the center increases indefinitely, so that in the limit, when the
ellipse becomes a circle, the directrices have disappeared."*

* It is, of course, possible to obtain the circle as a limiting curve ap-

proached by ellipses defined in the new way. If the points F and A of
Fig. 12, Ch. VII, are held fast and m is allowed to increase indefinitely,
then it can be shown that e approaches zero and that a and h both approach
the fixed distance AF. Thus the variable ellipse approaches a circle as
its hmit.
 THE HYPERBOLA 141

^^ Hyperbolas with
H^erbolas. Let
Foci
the student
on the Axis
show that the equation
of y. Conjugate
of the
hyperbola whose foci are at the points (0, ± C) on the axis
of y and the difference of whose focal radii is 2 ^ is

where
02 = A' + B\
The transverse axis of this hyperbola is the axis of y ; the
conjugate axis, the axis of x. The length of the major axis is
25 ; that of the minor axis, 2 A. The eccentricity is C/B and
the asymptotes have the equations,

and ^ + 1 = 0.
A B A B
Conjugate Hyperbolas. The two hyperbolas,

^_^ = 1
&2
and
X"'
r_
-^-yL^-A '
a' a^ b'-

have the same asymptotes. The transverse axis of each is the

conjugate axis of the other, and the major axis of each is the
minor axis of the other.
Taken together, the two
hyperbolas form what is

called a pair of conjugate

hyperbolas. The relation-
ship between them is per-
fect in its duality. We
say, then, that each is the
conjugate of the other.
The two hyperbolas to-

gether are tangent exter-

nally at their vertices to the rectangle of § 4 at the mid-points
of its sides. Moreover, all straight lines through the common
center 0, except two, meet one hyperbola or the other m two
points, and the segment thus terminated is bisected at 0.
 142 ANALYTIC GEOMETRY

The student should compare these facts with the correspond-

ing ones concerning a single ellipse and the circumscribed
rectangle.

EXERCISES
1. Find the coordinates of the foci, the lengths of the axes,
the slopes of the asymptotes, and the value of the eccentricity
for each of the hyperbolas :

^'> /».'>

(a) ^_ll = _l; (c) f--x''-=4.',

^
^ ^
9/ 16 '
^ ^

(b) 5a/^ - 4=2/+ 20 = ; (d) Sx'- -2y^-\-6 = 0.

Draw an accurate figure in each case.

2. What are the equations of the hyperbolas conjugate to

the hyperbolas of Ex. 1 ?

3. Find the equation of the hyperbola whose vertices are in

the points (0, ± 4) and whose eccentricity is |.
Ans. 4a^2_5 2,2_|_80 = 0.

4. Find the equation of the hyperbola the extremities of

whose minor axis are in the points ( ± 3, 0) and whose eccen-

tricity is 1^.

5. Prove that the sum of the squares of the reciprocals of

the eccentricities of the two conjugate hyperbolas

'
9 16 ' . 9 16
is equal to unity.

6. Prove the theorem of Ex. 5 for the general pair of conju-

gate hyperbolas.

•V 7. Show that the foci of a pair of conjugate hyperbolas

lie on a circle.

9. Parametric Representation. It is possible to construct a

hyperbola, given its axes, AA' and BB', by a method much
like that of Ch. VII, § 10, for the ellipse

KyP-
 THE HYPERBOLA
rp
143

Let the two circles, C and C, and the ray from 0, be drawn
as before. At the point L
draw the tangent to C", and
mark the point Q where the
iP^ix.y)
ray cuts this line. At R draw
the tangent to C and mark the
point >S' where this tangent
cuts the axis of x.

The locus of the point

P: (x, y), in which the paral-
lel to the axis of x through
Q and the parallel to the
axis of y through S intersect, is the hyperbola.

For,
144 ANALYTIC GEOMETRY

(b) a =3 cm., 6=3 cm.

(c) a = 2 cm., b = 3 cm.

2. Obtain a parametric representation of the hyperbola

^_^=_1
10. Conic Sections. The ellipse (inclusive of the circle), the
hyperbola, and the parabola are often called conic sections,
because they are the curves
in which a cone of revolution
is by planes.
cut
Suppose a plane cuts only M
one nappe of the cone, as is
shown in the accompanying
drawing. Let a small sphere
be placed in the cone near O,
tangent to this nappe along a
circle. It will not be large
enough to reach to the plane
31. Now let the sphere grow,
always remaining tangent to
the cone along a circle. It
will finally just reach the
plane. Mark the point of
tangency, F, of the plane M
with the sphere, and also the
FiQ. 15 circle of contact, C, of the
sphere with the cone.
As the sphere grows still larger, it cuts the plane M, but
finally passes beyond on the other side. In its last position, in
which it still meets M, it will be tangent to M. Let the point
of tangency be denoted by F', and the circle of contact of the
sphere with the cone by C".
Through an arbitrary point P of the curve of intersection of
31 with the cone passes a generator OP of the cone let it cut ;

C in M and C" in H'. Then RB', being the slant height of

 THE HYPERBOLA 145

the frustum * cut from the cone by the planes of C and C', is
of thesame length, 2 a, for all points P.
P with F. Then PF and PE, being tangents from P
Join
to the same sphere, are equal. Similarly, PF' and PR' are
equal. Hence
FP + F'P =RP+ R'P = i^i?',
or FP-\- F'P = 2a.

But this locus is by definition an ellipse with its foci

at F and F', and hence the proposition is proved for the case
that M cuts only one nappe, the intersection being a closed
curve.
If the plane M cuts both nappes, but does not pass through /
0, it is aharder to draw the figure, one sphere being
little

inscribed in the one nappe, the other, in the other nappe.

A similar study shows that here the difference between
P
FP and i^' is equal to RR\ and hence the locus is a
hyperbola.
The parabola corresponds to the case that M meets only one
nappe, but does not cut it in a closed curve. This case is
realized when M does not pass through and is parallel to a
generator of the cone.
Let i be a line which is perpendicular to the axis of the
cone in a point of the axis distinct from the vertex. As a
plane, M, rotates about L, it will cut from the cone all three
kinds of conies. This will still be true if we take, as L, any
line of space which does not pass through the vertex and is
not parallel to a generator.

11. Confocal Conies. Two conies are said to be confocal if

they have the same foci.; in the case of two parabolas, we de-
mand, further, that they have the same axis.

* No technical knowledge of Solid Geometry beyond the definitions of

the terms used (which can be found in any dictionary) is here needed.
On visualizing the figure, the truth of the statements regarding the space
relations becomes evident.
 .

146 ANALYTIC GEOMETRY

Consider an ellipse and a hyperbola which are confocal.

They evidently intersect in four points.*
Let P be one of these points. Join with P
and F'. F
Then FP and F^P are focal radii both of the ellipse and of the
hyperbola. Now, the tangent to a hyper-
bola at any point not a vertex bisects the
angle between the focal radii drawn to
that point, § 5 and the normal to an
;

ellipse atany point not on the transverse

axis bisects the angle between the focal
radii drawn to that point, Ch. VII, § 4.
Fig. 16
It follows, then, that the tangent to
the hyperbola at P
and the normal to the ellipse at this
point coincide. Hence the two curves intersect at right
angles, or orthogonally, as we say. We have thus proved the
following

Theorem. A pair of confocal conies, one of which is an el-

lipse and the other a hyperbola, cut each other orthogonally.

Confocal Parabolas. Consider two parabolas having the

same focus and the same axis. If both open out in the same
direction, they have no point in common. If, however, they

open out in opposite directions, they intersect in

two points which are symmetrically situated with
respect to the axis.
In the latter case, the parabolas intersect orthogo-
nally, as has already been proved analytically ; cf
Ch. VI, § 3, Ex. 10.
This result could have been forecast, as a conse- Fig. 17
quence of the relations established in § 7. Eor, if
one focus, F, and the two corresponding directrices of a pair
of confocal conies, consisting of an ellipse and a hyperbola,
are held fast, and if the other focus is made to recede in-
definitely, each of the conies approaches a parabola. But the

* Let the student satisfy himself that two confocal ellipses do not in-
tersect, and that the same is true of two confocal hyperbolas.
 ;

THE HYPERBOLA 147

conies always intersect orthogonally, and so the same will be

true of the limiting curves, the parabolas.
To obtain a prescribed pair of parabolas, like those described
above, as limiting curves, it is necessary merely to choose the

two confocal conies so that the directrices corresponding to F

are at the proper distances from F.

Mechanical Constructioyis. It is possible to draw with ease a

large number of confocal ellipses by the method set forth in Ch.
VII, § 1. Let thumb tacks be inserted at F
and F', but not pushed clear down. Let a
thread be tied to the tack at F, passed round
the tack at F% and held fast at M. Then an
ellipse can be drawn with F and F as foci. '

Now let the thr.ead be unwound at F' Fig. 18

and drawn in or paid out slightly, so that
the length of the free thread between and F' is changed.F
On repeating the above construction, a second ellipse with
its foci at F
and F' is obtained and so on, ;

, Ther^ an analogous construction for a hyperbola, which

is

has not yet been mentioned. Tie a thread to a pencil point,*

pass the thread round the pegs at F and
F' as shown, hold the free ends firmly
together at M, and, keeping the thread
taut by pressing on the pencil, allow M to
move. The pencil then obviously traces
out a hyperbola.
Fig. 19 By pulling one end of the
thread in slightly at M, or by
paying it out, and then repeating the construction,
a new hyperbola with the same foci is obtained
and so on.
Parabolas. The accompanying figure suggests
a means for drawing a parabola mechanically. Fig. 20

* To keep the thread from slipping off, cut a groove in the lead, such
as would be obtained if the pencil were turned about its axis in a lathe
and the point of a chisel were held against the lead close to the wood.
148 ANALYTIC GEOMETRY

A ruler, D, is held fast and a triangle, T, is allowed to slide

along the ruler. A thread is tied at F and Q, and a pencil
point, P, keeps the thread taut and pressed against the
triangle.

EXERCISES
1. Show that the conies,

^
24
+ ^'=1
8
and ^-^=1,
4 12 '

are confocal.

2. Prove that the equation,

9
flj2

+ A'+ 5 4- A
= ^ 1,

represents an ellipse for eacli value of A greater than —5 and

represents a hyperbola for each value of A between —9 and
— 5. Show that all these ellipses and hyperbolas are confocal,
with the points (±2, 0) as foci.

3. For what values of A does the equation

^2
a?^

a2 + A + t/^=1,
62 _^ A

where a and h are given positive constants such that a > 6,

represent i) ellipses ? ii) hyperbolas ? Show that all these

conies are confocal.

4. Draw a set of confocal ellipses and hyperbolas.

5. Draw a set of confocal parabolas, all having the same
transverse axis, some opening in one direction, some in the
other. \J^ i;'^' v "
>-- t >

EXERCISES ON CHAPTER VIII

^'5^ The axes of a hyperbola which goes through the points

(1, 4), (— 2, 7) are the axes of coordinates. Find the equation
of the hyperbola. Ans. y^ — 11 ic^ = 5.
(J) Show that the hyperbolas defined by the equation
 THE HYPERBOLA 149

where c is an arbitrary constant, not zero, all have the same

asymptotes.

(^ How many hyperbolas are there with the lines ^.

Sx^ — 16y- = yj .--y-^j^

as asymptotes ? Find an equation which represents them all.

Ans. Sx"-— IGy"' = c, c=^0.

/^4^ What is the equation of all the rectangular hyperbolas

with the axes of coordinates as axes ?

^^ Ahyperbola with the lines 4,x'' — y'' = as asymptotes

goes through the point (1, 1). What is its equation ?
_. 3^^ l."^ Ans. 4:X'-y'^ = S.
6.( The asymptotes of a hyperbola go through the origin
and have slopes ± 2. The hyperbola goes through the point
(1, 3). Find its equation. Ans. 4 — y^ = — 5. a;''

^'^7i The two hyperbolas of Exs. 5 and 6 have the same

asymbtotes, but lie in the opposite pairs of regions into which
the I lane is divided by the asymptotes. Show that the sum
of the squares of the reciprocals of. ^heir eccentricities equals
unity .\ r- ^?^
Frove that of the hyperbolas of Ex. 2 those for which
8.

c is positive are all similar, and that this is true also of those
for which c is negative. If e is the common value of the ec-
centricity of the hyperbolas of the first set and e' is that of
the hyperbolas of the second set, show that

(1) 1+1=1.

9. Prove that the relation (1) is valid for the eccentricities

of any two hyperbolas which have the same asymptotes but

lie in the opposite regions between the asymptotes.

10. Show that two hyperbolas which are related as those

described in the previous exercise have the same eccentricity
if and only if they are rectangular hyperbolas.
 0^

150. \
^ ANALYTIC GEOMETRY

l\ A hyperbola with its center in the origin has the eccen*

tricity 2. Find the equations of the asymptotes, (a) if the
foci lie on the axis of x ih) if the foci lie on the axis of y.
;

Ans. (a) 3x'-y'' = 0; (6) a^'^-3/ = 0.

dgj What is the equation representing all the hyperbolas

^ which have their centers in the origin -and eccentricity 2,
(a) if the foci lie on the axis of x ? (b) if the foci lie on the
axis of y? Show that in either case the vertices lie midway
between the center and the foci.

13. Prove that the vertices of the hyperbola

subtend a right angle at each of the points (0, ± b) when and

only when the hyperbola is rectangular. What is the corre-
sponding theorem in the case of the ellipse ?
14. The projections of a point P of a hyperbola on the
transverse and conjugate axes are Pj and P2. The tangent at
P meets these axes in Ti and Tg. Show that OPi OTi= • a^
and OP2 0X2=— • where b""', is the center of the hyperbola
and a and b are the semi-axes.

Prove that the segment of a tangent to a hyperbola be-

15.
tween the point of contact and a directrix subtends a right
angle at the corresponding focus.
16. The projection of a point P of a hyperbola on the
transverse axis is P^ and the normal at P meets this axis at
Ni. Show that the ratio of the distances of the center from
Ni and P^ equals the square of the eccentricity.

17. Prove that the line joining a point P of a hyperbola

with the center and the line through a focus perpendicular to
the tangent at P
meet on a directrix.
Pind the equation of the circle which
18. is tangent to a
hyperbola at the upper ends of the two latera recta.

19. Let be the center, A a vertex, and F the adjacent

focus of a hyperbola. The tangent at a point P meets the
 THE HYPERBO]
transverse axis at T and the tangent at A meets
Show that TV is parallel to AP.
20. Show that an asymptote, a directrix, and the line through
the corresponding focus perpendicular to the asymptote go
through a point.
21. A line through a focus F parallel to an asymptote meets
the hyperbola at P. Show that the tangent at P, the other
asymptote, and the line of the latus rectum through F meet in
a point.
22. Let P be a focus and D the corresponding directrix of
a hyperbola. A line through a point P of the hyperbola parallel
to an asymptote meets D in the point K. Prove that the tri-

angle FPK is isosceles.

Exercises 23-33. In proving the theorems in these exercises

it will, in general, be necessary to make actual use of the
equation expressing the fact that a certain point lies on the
hyperbola.
23. The tangent to a hyperbola at a point P meets the tan-
gent at one vertex in Q. Prove that the line joining the other
vertex to P
is parallel to the line joining the center to Q.

<C34l Let P be
a focus and D the corresponding directrix of a
hyperbola. Prove that the segment cut from D by the lines
joining the vertices with an arbitrary point on the hyperbola
subtends a right angle at P. 'j ^ ^
Prove that the product of the distances of the foci of a
25.
hyperbola from a tangent is constant, i.e. independent of the
choice of the tangent.
Let A and
A' be the vertices of a rectangular hyperbola
and let P and P'
be two points of the hyperbola symmetric in
the transverse axis. Prove that AP is perpendicular to A'P'
and that AP
is perpendicular to A'P.

27. Show that the product of the focal radii to a point on a

rectangular hyperbola is equal to the square of the distance of

the point from the center.

152 ANALYTIC GEOMETRY

/ 28.\Prove that the angles subtended at the vertices of a

hyperbola by a chord parallel to the conjugate axis
retjtarigular
are supplementary.
29. Prove that the product of the distances of an arbitrary
point on a hyperbola from the asymptotes is constant, i.e.

the same for every choice of the point.

30. A line on a hyperbola
through an arbitrary point P
parallel to the conjugate axis meets the asymptotes in and M
N. Show that the product of the segments in which P divides
MN is constant.

Prove that the segment of a tangent to a hyperbola cut

31.
out by the asymptotes is bisected by the point of contact of
the tangent.
32. Show that the tangent to a hyperbola at an arbitrary point
forms with the asymptotes a triangle which has a constant area.
33. The tangent to a hyperbola at a point P meets the tan-
gents at the vertices in M and N. Prove that the circle on
MN as a diameter passes through the foci.

Loci

Pind the locus of a point whose distance from a given

34.
circle always equals its distance from a given point without
the circle. Pirstgive a geometric construction, with ruler and
compass, for points on the locus. Then lind the equation of
the locus.
35. The base of a triangle is fixed and the product of the
tangents of the base angles is a negative constant. What is

the locus of the vertex ?

,^ A line moves so that the area of the triangle which it

forms with two given perpendicular lines is constant. Find
the locus of the mid-point of the segment cut from it by these
lines.
Ans. Two conjugate rectangular hyperbolas, with the given
lines as asymptotes.
 THE HYPERBOLA 153

37. Given a fixed line L and a fixed point A, not on L. A

point Pmoves so that its distance from L always equals the
distance AQ, where Q is the foot of the perpendicular dropped
from P on L. What is the locus of P ?
38. What is the locus of the point P of the preceding exer-
cise, if the ratio of its distance from L to the distance -4Q is
constant ?

t ^"-^

- 1 r
bU .x_>-

O!^
 CHAPTER IX
«

CERTAIN GENERAL METHODS

1. Tangents. Let it be required to find tlie tangent line to

a given curve at an arbitrary point.
In the case of the circle the tangent is perpendicular to the
radius drawn to the point of tangency. But this solution is
of so special a nature that it suggests no general method of
attack. A general method must be
based on a general property of tan-
gents, irrespective of the special curve
considered. Such a method is the
following. Let P be an arbitrary
point of a given curve, O, at which it
is desired to draw the tangent, T. Let
a second point, P\ be chosen on (7,
and draw the secant, PP'. As P'
moves along C and approaches the
fixed point P as its limit, the secant rotates about P as a pivot
and approaches the tangent, T, as its limiting position. Thus
the tangent appears as the limit of the secant.
If, now, in a given case we can find an expression for the
slope of the secant, the limit approached by this expression
will give us the slope of the tangent. The slope of the tangent
to the curve at P we shall call, for the sake of brevity, the
slope of the curve at P.

Example 1. Tind the slope of the curve

(1) y = x^
at a given point, P.
154
 : ,

CERTAIN GENERAL METHODS 155

Let the coordinates of P be

(a^i, ?/i) ; those of P', {x', y'),
or (x^ -h h, ?/i
4- k). Then
PQ = h, QP = k,
and we have, for the slope of
the secant PP', the expression

(2) tanr' =
h

where r' =^ QPP'. The slope of the tangent line, T, at P is,

then,

(3) tan T = lim tan t' = lim -

P'=P h=^o Ji

where t^'^QPT. The sign = is used to mean " approaches

as its limit," and the expression: lim tanr', is read: "the

r', as P' approaches P."

limit of tan
Suppose, for example, that is the point (1, 1). P Let us
compute k and tan r' for a few values of li. Here, x^ = 1 and
yi — 1. If A = .1, then
x' = Xi h = 1.1,
-{-

y' = y, + k = (l.iy=1.21,
k = .21,
.21
and hence tan t' = 2.1.
Next, let P' be the point for which
a.-' =1.01.
Then ?/'= 1.0201,

h = .01, k = .0201,
0*^01
and hence tanr' = ^^^^^ = 2.01.
.01

Let the student work out one more case, taking a;' = 1.001.
He will find that here k = .002001 and

tanr' = 2.001.
 :

156 ANALYTIC GEOMETRY

These results can be presented conveniently in the form of

a table

h
 CERTAIN GENERAL METHODS 157

We can say", then, that the slope of the curve (1), at an arbi-
trary point P : {xi, y-^ on it, is

If, in particular, P is the point (1, 1), the slope of the tan-
gent there is X = 2 1 = 2, and thus
• the indication given by
the above table is seen to be borne out.

Example 2. Find the slope of the curve

(6) y =
|
at an arbitrary point P : {xi, y^ of the curve.
Denote, as before, the coordinates of a second point, P', by
x' =Xi-\- h, y' = y^-\- k.

Then, since P and P' lie on the curve,

2/1 =-
x^

and yi-\-k =
Xi -j- h
(^ a^
Hence A; =
Xi-\- h Xi

Nothing is more natural than to reduce the right-hand side

of this equation to a couimon denominator. Thus
,
k = — aVi
^i(^i 4- ^)
Consequently,
= -k = — d^
tan T f
,

h x^{xi + K)
We are now ready to let P' approach P:

lim tan t' = lim

The limitapproached by the right-hand side is obviously
— a^/x^, and so
a2
tanr =
x^
158 ANALYTIC GEOMETRY

We have, then, as the final result : The slope of the curve (6),
at an arhitary point {xi, 2/1) on it, is

x^

Equation of the Tangent. Since the tangent to the curve (1),

y=A

at the point (1, 1) has the slope 2, its equation is

y-l=2{x-l), or 2x-y-l=0.
Similarly, the equation of the tangent to the curve (1) at an
arbitrary point P : (xi, y^) is

y-Vi = 2xi(x-Xi),
or
y — y^ = 2 XiX — 2 x^.
This equation may be simplified by use of the equality,

2/1 = ah^
which says that the point P lies on the curve. For, if we re-
place the term 2 x^ by its equal, 2 y^, and then combine the
terms in 2/1, the equation becomes

2/ + 2/1 = 2 x^x.
This equation of the tangent is of the first degree in x and y,
as it should be. The quantities x^ and 2/1 are the arbitrary,
but in any given case fixed, coordinates of P and are not
variables.

Equation of the Normal. The line through a point of a P

curve perpendicular to the tangent at is known as the P
normal to the curve at P.
Since the tangent to the curve y = x^ at the point (1, 1)
has the slope 2, the normal at this point has the slope — ^.
Consequently, the equation of the normal is

2/-l=-|(a;-l), or x-}-2y-3 = 0.
 CERTAIN GENERAL METHODS 159

EXERCISES
1. Determine the slope of the curve y = x- — x Sit the point

(3, 6). First make out a table like that under Example 1, and
hence infer the probable slope. Then take an arbitrary point
(^1) 2/i) on the curve and determine the actual slope at this
point by finding

h^ h
2. The same for the curve Sy = 3x^Sit the point (2, 3).
3. The same for the curve y = 2x^ — 3x-\-l at the point
(1, 0).

Find the slope of each of the following curves at an arbitrary

point F : (xi, y^). No preliminary study of a numerical case,
like that which gave rise to the table under Example 1, is

here required.

4. y = x" — Sx-\-l. Ans. \ = 2x^ — S-

5. y = 2x^-x-4.. 7. y = 4:X^ - 2x''- -\- 5.

e. y = x^ — X. S. y = x^ -\- x^- -\- X -{- 1.

9. y = x^-\-px + q. Ajis. = S xi^ -{-p.

X
10. y = x^ — a\ Ans. A = 4:Xi\
1 2
11- y=-,' .
Ans. X =
x^ Xi^

XT

13. y=- •
Ans. X = .

1-x (1 - x,y
14. y =- -• Ans. X =—
3aj-4 (3a;i-4)2
15. y = ax'' -^ bx -\~ c. Ayis. X = 2 axi -f b.
16. y = ax^ + ex d.
bx''- -\- -\-

17. y = x"",a positive

(ii, integer) Ans. X. = i}x^''~^.

18. y= ex"".
160 ANALYTIC GEOMETRY

Find the equations of the tangents to the following curves

at the points specified. In each case reduce the equation ob-
tained to the simplest form.

19. The curve of Ex. 1 at the points (3, 6) ;

(cci, y^.
Ans. 5 a; — —9=?/ ; {2xi—l)x — y — x^ = 0.
20. The curve of Ex. 3 at the points (a.'i, 2/1) ; iX, 0).

21. The curve of Ex. 4 at the points {xi, ?/i) ; (— 1, 5). •

22. The curve of Ex. 11 at the points (1, 1) ;

(iCi, 3/1).

23. The curve of Ex. 17 at the point {xi, y^.

Ans. nxi^'^x —y— (71 — l)yi = 0.
24. The curve of Ex. 13 at the point whose abscissa is 2.

25. The curve of Ex. 14 at the point whose abscissa is 4.

26. Find the equations of the normals to the curves of Exs.

21, 22 at the designated points.

2. Continuation. Implicit Equations. We have applied the

general method to curves whose equations are given in the
form :
2/ = a simple expression in x. More precisely, this
" simple expression " has each time been a polynomial (or even
a monomial), or the ratio of two such expressions.
But even the simplest forms of the equations of the conies
are, as a rule, such that, if the equation be solved for y, radi-
cals will appear. In such cases, the following method of treat-
ment can be used with advantage.

TJie Parabola. Let it be required to find the slope of the

parabola
(1) 2/^ = 2 mx
at any point P : (xi, y^) on the curve.
We will treat first a numerical case, setting m=2 :

(2) 7f = ix.
Since P is on the curve, we have

(3) yi' = 4.x,.

 :

CERTAIN GENERAL METHODS 161

Since P' : (a^i + /i, 2/1 + ^) is also on the curve, we have ;

or
(4) 2/r + 22/ifc + A;- = 4:X, -h 4/i.
Subtract (3) from (4)
2yjc + r- = 4.h,
Divide this equation through by h, to obtain an equation for
tan t' = k/h :

2ij^--{-l€- = 4:, or 2 2/1 tan t' + A: tan/ = 4.

h h

Solve the latter equation for tan t' :

A
tan t' =

We are now ready to let P' approach P as its limit. This

means that h and k both approach 0. We have, then,

lim tan t' = lim -,

F'=p A=H) 2 2/1 + A;

or
.

tanr = --
4
—=—
2
22/1 2/1

It has been tacitly assumed that 2/1 ¥= 0. If 2/1 = ^> ^^^^

tanr' increases indefinitely as h, and with it k, approaches

zero. Thus the tangent line is seen to be perpendicular to the

axis of X Sit this point, as obviously is, in fact, the case, since
the point is the vertex of the parabola.
The student will now carry through by himself the corre-
sponding solution in the general case of equation (1). He
will arrive at the result The slope \ of the parabola :

2/2 = 2 mx
at an arbitrary point (xi, 2/1) of the curve is

(5) X = --
 : : . '

162 \1„. ANALYTIC GEOMETRY

The Ellipse. The treatment in the case of the ellipse,

is precisely similar. Writing (6), for convenience, in the

form
(7) b^x' + ay- = a"b\
we are led to the following equations * :

(8) b^Xi^ + a^yi^ = a-^b^ ',

. .

or
(9) b^x^^ 4- a22/i2 + 2 b^x^h + 2 a22/iA; + bVi^ + «2A;2 = a262.
Subtract (8) from (9)

2 b^x^h + 2 a^yjk + 52/^2 _|_ a2A:2 = 0.

Divide hy h:
2 b^x, + 2 a22/i- + bVi + a2A;^^h = 0,
h
or ' 2 62iCi + 2 a22/i tan t' + &^/i + «^^ tan r' = 0.
Solve this equation for tan t' :

tan T =—
2 a22/i
^—
4-
'—a% ~r •

Now let P' approach P as its limit

lim tan r = lim — — •

P =F A=M) 2a2|/i 4- a^k

Hence tan t = - -—-^ = - ^

We have thus obtained the result : The slope X of the ellipse

a2 52

ai an arbitrary one of its points {xi, y^) is

(10) X = -*^:
* The student will do well to paraphrase the text at this point with a
numerical case, — say, 4 x2 -f 9 7/2 — 36.
 CERTAIN GENERAL METHOPa. .^ 163

The Hyperbola. The treatment is left to the student. The

result is as follows.
The slope X of the hyperbola
x^ __y^ __^

at an arbitrary one of its points (xi, y{) is

(11) A=^-
Equation of the Tangent. Since the slope of the ellipse at
the point (xi, y^) is — b-Xyjc^y^^ the equation of the tangent at
(^1, yi) is

or, after clearing of fractions and rearranging terms,

b'^x^x -f a^y^y = b'^x-^ -f a^y^.
If we divide both sides of this equation by ^252^ ^^ have

But, since the point (a^i, y-^ lies on the ellipse, it follows that

"^
a2 52 '

and the equation of the tangent becomes

a2 62 '

The equation of the tangent to the ellipse

^-4-^ = 1
a"^ b^
at the point (xi, y-^ is

(12) •
^_^.M=l.
^ / a^ 62

In a similar manner let the student establish the equations

of the tangents to the hyperbola and the parabola.
 1 .

164 ANALYTIC GEOMETRY

The tangent to the hyperbola

x^ _y^ _ ^
a^~'b^~
at the point (xi, yi) has the equation

The tangent to the parabola

2/2 = 2 mx
at the point (xi, yi) has the equation

(14) 2/i2/ = m(a; + a^i).

EXERCISES
Find the slope of each of the following six curves at an
arbitrary one of its points, applying each time the method
set forth in the text.

1. 2i»2-|-3?/2= 12. 3. y^ = 12x.

2. a;2— 4?/2 = 4. 4. x^ — = a\ y^-

5. Ax'^ + By^ = C, where A, B, C are all positive.

6. y^ = Ax-\- B, where A =^ 0.

7. Find the slope of the parabola 2?/ = 6 at the point

?/2 -|- a;

3
^ ^ Vi +
8. What is the slope of the parabola of Ex. 7 at the origin?
Ans. 3.
9. Find the slope of the curve
ic^ — 2/2 —3 +4 =
a; ?/

at the origin. Ans. A= f

Suggestion. First find the slope at an arbitrary point {xi, y-^.

Then substitute in the result the coordinates of the origin.

10. What angle does the curve

2iK3-3?/2 + a;-2/4-l =0 •

make with a parallel to the axis of x at the point (1, 1) ?

Ans. 45°.
 CERTAIN GENERAL METHODS 165

11. Find the slope of the curve xy = a^ at any point (xi, 2/1)
by the method of the present paragraph, and show that your
result agrees with that of § 1, Example 2.

Find the equation of the tangent to each of the following

curves at the point designated, applying each time the method
of the text. Reduce the equation to its simplest form.

12. The curve of Ex. 1 at the point (x^, y-^).

Ans. 2 XiX -f 3 yiy = 12.

13. The curve of Ex. 3 at the points (x^, 2/1) ; (3, — 6).

14. The curve of Ex. 5 at the point (a^i, 1/1).

Ans. AxiX + Byiy = C.

15. The curve of Ex. 6 at the point {xi, y^).

16. The curve of Ex. 7 at the points (a;i, 2/1) ; (i, 1)-

17. The curve of Ex. 9 at the origin.

18. Find the equations of the normals to the curves of Exs.

12, 13 at the points specified.

3. The Equation u + kv = 0. Consider the following ex-

ample.
The equations \
(1) a; + .v-2 = a,
(2) x-y = 0,
represent two straight lines intersecting
in the point (1, 1), as shown in Fig. 3.
What can we say concerning the curve *
(3) {x + y-2)+h{x-y)=0,
where k denotes a constant number ?
This curve is a straight line, since (3) is an equation of the
first degree in x and y. Suppose, now, that various different
values are given to k. Then (3) represents various straight
lines in turn. What do all these lines have in common ?
* The word "curve" is used here in the sense common in analytic
geometry, to denote merely the " locus of the equation." Consequently
a curve in this sense is not necessarily crooked it may be a straight line.
;
 :

166 ANALYTIC GEOMETRY

Since the lines (1) and (2) intersect in tlie point (1, 1), the
coordinates of this point make the left-hand sides of equations
(1) and (2), namely, the expressions,

X -\- y —2 and x — y,
vanish. Consequently, they always make the left-hand side
of equation (3) vanish. In other words, equation (3) is satisfied
by the coordinates of the point of intersection of the lines (1) and
(2), NO MATTER WHAT VALUE k HAS. This mcaus that all the
straight lines represented by (3) go through the point of inter-
section of the lines (1)and (2).
The result can be restated in the following form. Let the
single letter u stand for the whole expression x -{- y — 2:

u = x-\-y-2,
the sign = meaning identically equal, i.e. equal, no matter what
values X and y have. Similarly, let v stand for x — y:
V =:X —y.
Then (3) takes on the form
(4) w + kv = 0.
We now restate our result.

If uz=0 and v = are the equations of tico intersecting straight

lines, theyi the equation

u + T^^' =
represents a straight line which goes through the j^oint of inter-
section of the two given lines.

By giving to A: a suitable value, u -f kv = can be made to

represent any desired line through the point of intersection
(^ij 2/i) of the given lines, with the sole exception of the line
V = 0. For, let L line, and let {x2, 2/2) t)e a
be the desired
point of L from {x^, y^. Then, on substituting for x
distinct
and y the values X2 and y^ in the equation u -\-kv = 0, we ob-
tain an equation, in which k is the unknown. This equation
can be solved for k, since v does not vanish for the point
(^2) 2/2)'
 : .

CERTAIN GENERAL METHODS 167

Example. Find the equation of the line L which goes

through the point of intersection of the lines (1) and (2) and
cuts the axis of y in tke point (0, — 4).
The required line, L, is one of the lines (3) ; i.e. for a suit-
able value of ky (3) will represent L. To find this value of A:,

we demand that (3) contain the given point (0, —4) of L.

We have, then, setting a; and ?/ =
4 in (3) =—
(0_4-2)4-A:(0 + 4)=0 or A: =f
Consequently, the equation of the line L is
x-\-y — 2-^^{x — y)=0 or ox— — 4 = 0.
?/

That the line represented by the latter equation does actually

go through the points (1, 1) and (0, — 4) can be verified

directly.
The principle which has been set forth for two straight lines

evidently applies to any two intersecting curves whatever, so

that we are now in a position to state the -following general
theorem.

Theorem 1. Let u = and v = be the equations of any

tico intersecting c^irves. Then the equation

u-{-kv = 0, k^O,
represents^ in general,* a curve ichich j^ci'^^ses through all the

points of intersection of the two given curves, and has no other

point in common ivith either of them.

The last statement in the theorem is new. To prove it,

we have but to note that, if the coordinates of a point P satisfy

the equation u -j-kv = and also, for example, v = 0, they
must satisfy the equation u = -, that is, if P is a point on the
curve ic -\-kv = 0, which lies on one of the given curves, it
lies also on the other and so is a point of intersection of the
two.

* It may happen in special cases that the locus u + kv = reduces to

a point, as when, for example,
u = 2x2 ^. 2y2_ X, i3 = x2 + 2/2-x, k = -l.
168 ANALYTIC GEOMETRY

Suppose, now, that the equations u = and v = represent

two curves which have no point of intersection. It follows,
then, from the argument just given, that the curve

has no point in common with either of the given curves. But

it may
happen, in this case, that there are no points at all
whose coordinates satisfy the equation u -\- kv = 0. Thus, if
u = x^ -\- y^ — 1,
V = x^ -{- y^ — 4:y

and A; = — 1, we have
u + kv = 3,
and there are no points whose coordinates satisfy the equation
3 = 0.
The general result can be stated as

Theorem 2. Let u = and v = be the equations of two

non-intersecting curves. Then the equation
u -\-kv = 0, k =it 0,

represents, in general, a curve not meeting either of the tivo given

curves. In x:)articular, it may happen that the equation has no
locus.*

In the special case that u and v are linear expressions in

X and y, it is possible to say more.
Ifu = 0a7idv = are the equations of two parallel straight
lines, the equation
u -\-kv = 0, k^ 0,
represents, in general, a straight line parallel to the given lines.
For a single value of k, the equation has no locus.

Thus, if the parallel lines are

?t = + = 0,
a; 2/ 1? = + + 1 = 0,
ic 2/

the equation
(5) u-\-kv={l-\-k)x+{l-\-k)y +k=
* It may happen, also, that the equation represents just one point, as
when, for example,
M = x2 + 1/2 — 2, u = x2 + y2 _ 1. lc = —2,
 CERTAIN GENERAL METHODS 169

has no locus when A: = — 1, but otherwise it represents a line,

of slope — 1, parallel to the given lines. In fact, it yields all

the lines of slope — 1, except the line i; = 0, since, if we re-

write it in the form,

the quantity fc/(l -f- 7c) may be made to take on any value, ex-
cept 1, by suitably choosing k.

Pencils of Curves. All the lines through a point, or all the

parallel lines with a given slope, form what is called a pencil
of lines. Equation (5) represents, when h is considered as an
arbitrary constant, all the lines of slope — 1, except the line

in this case, then, u -\-'kv = and v = together represent all

the lines 'of slope — 1, that a 2yencil of parallel lines.
is,

Similarly, it -}- kv = 0, when = and = are the lines

2/ -u

(1) and (2), yields all the lines through the point (1, 1),
except the line (2); hence u-^-kv^O and -u together =
represent all the lines through the point (1, 1), — a pencil of
intersecting lines.
Thus, if 2^ = =
and v are any two lines, the equations

(6) u + kv = and v =Q
together represent a pencil of lines.
If we set k = in /I in u -\-kv = and multiply by Ij the re-
sulting equation
(7) lu + mv =
is equivalent to the equation u -\-kv = when Z 9^ 0, and when
1 = (m becomes v = 0. Consequently, the two equa-
=5fc 0), it
tions (6) may be replaced by the single equation (7).
The pencil of lines through the point (1, 1), for example,
may now be given by the single equation
l{x + y-2)^m{x-y)=0,
where I and m have arbitrary values, not both zero.
170 ANALYTIC GEOMETRY

In general, if u = and v = are any two curves, all the

curves represented by the equation

lu -f- w-^J = 0,
where I and m have arbitrary values, not both zero, form what
is called a pencil of curves.

Applications. Example 1. Let

u = X- -\- if' -\- ax -\- hy -\-c = Oj

'y = + aj^ ?/2 -f a'x + h'y + = 0,

c'

be the equations of any two circles which cut each other.

Then the equation
u-v={a-a')x+{h-h')y+{c-c')=0
represents a curve which passes through the two points of
intersection of the circles. But this equation, being linear,
represents a straight line, and is, therefore, the equation of
thecommon chord of the circles.
The foregoing proof is open to the criticism that conceivably
we might have
a-a' = 0, b-b' = 0,
and then the equation u —v = would not represent a straight
line. But in that case the circles would be concentric, and we
have demanded that they cut each
Wi=0 other.

Example 2. We can now prove

the following theorem : Given three
circles, each pair of which intersect.
«i=0
Then their three common chords
pass through a point, or are parallel.
Let two of the three given circles
be those of Example 1, and let the

equation of the third circle be

w = x'^-\-y^ + a''x + b"y + c" — 0.

 :

CERTAIN GENERAL METHODS 171

Then the equations of the three common chords can be written

in the form
u — v = 0, V— ^o = 0, w— u = 0.
Let

We observe that the equation,

(8) Ui-\-Vi-\- ivi = or — Wi = %i + Vi,

holds identically for all values of x and y. Consequently, the

line Wi = is the same line as

Ui + Vi = 0,
and therefore it passes through the point of intersection of
Ui = and Vi = 0, or, if these lines are parallel, is parallel to
them. Hence the theorem is proved.
The above proof is a striking example of a powerful method
of Modern Geometry known as the Method of Abridged Nota-
tion.* By means of this method many theorems, the proofs of
which would otherwise be intricate, or for whose proof no
method of attack is readily discerned, can be established with
great ease.

EXERCISES
Find the equation of the straight line which passes
1.

through the origin and the point of intersection of the lines

2x-3y~2 = 0, 5x-\-2y + l = 0.
Ans. 12x -\- y = 0.
Find the equation of the straight line which passes
2.

through the point (—1, 2) and meets the lines

x-\-y = 0, x + y-^3 =
at their point of intersection.

* The first general development of this method was given by the

geometer, Julius Pliicker, in his Analytisch-geometrische Entwicklungen
of 1828 and 1831.
 •

172 ANALYTIC GEOMETRY

3. Find the equation of the straight line which passes

through the point of intersection of the lines

5x-2y-3 = 0, 4a; + 72/ -11 =
and is parallel to the axis of y.

4. Find the equation of the straight line which passes

through the point of intersection of the lines given in Ex. 3

and makes an angle of 45° with the axis of x.
5. Find the equation of the straight line which passes

through the point of intersection of the lines of Ex. 1 and is

perpendicular to the first of the lines given in Ex. 3.
Ans. 38a; + 95^ + 58 = 0.

6. The same, if the line is to be parallel instead of per-

pendicular.

7. Find the equation of the common chord of the parabolas

y2_2y-\.x=z0, y'^-\-2x—y = 0.
Ans. x-\- y = 0.
8. The same for the "parabolas

2a;2_5a; + 2?/ = 3,
3a;2-l-7a;-92/ = 4.
9. Write the equation of the pencil of curves determined

by the two curves (a) of Ex. 1 (b) of Ex. 3 (c) of Ex. 7. ; ;

10. What is the equation of the pencil of circles determined

by the two circles
a;2 + 2/2 — 2a;— 1 = 0,
a;2-|-2/^ + 4a; — 1 = 0?
Draw a figure showing the pencil. Find the equation of that
circle of the pencil which goes through the point (2, 4).
11. Find the equation of the pencil of parallel lines (a) of

slope 1 ; (6) of slope —3 ;

(c) of slope Xq.

Ans. (a) y = x-\-k.

12. Find the equation of the pencil of lines through (a)

the point (0, 0) ;

(b) the point (3, 2) ;
(c) the point (0, b) ;

(d) the point (a;o, 2/0) ^^s- (^) lx-\-my = 0.

 CERTAIN GENERAL METHODS 173

4. The Equation uv = 0. Consider, for example, the equa-

tion
(1) x^ - / = 0.
Since aj^ — 2/^ = (x — y)(x + y),
it is clear that equation (1) will be satisfied

(a) if {x, y) lies on the line

(2) x-y = 0',

(b) if {x, y) lies on the line

(3) ^'
+^= 0;

and in no other case. Equation (1), therefore, is equivalent to

the two equations (2) and (3) taken together, and it represents,
therefore, the two right lines (2) and (3).
It is clear from this example that we can generalize and
say:
Theorem. The equation
nv =
represents those points (x, y) ichich lie on each of the tivo curves^

w = 0, v = 0,
and no others.

It follows as an immediate consequence of the theorem that

the equation
uvw..^=0,
whose left-hand member is the product of any number of
factors, represents the totality of curves corresponding to the
individual factors, when these are successively set equal to zero.

Example. Consider the equation,

a;4 - 2/4 ^ 0.
Here,* >

x^ - y" ={x'^ - y''){x'^ + y'^)=={x - y){x + y){x'' -\-y'').

* It is is an identity, and so the sign
true that the following equation
= instead might be expected. The use of the sign = for an identical
of =
equation is not, however, considered obligatory, the sign = being used
when it is clear that the equation is an identity, so that the fact does
not require special emphasis.
174 ANALYTIC GEOMETRY

The given equation is, therefore, equivalent to the three equa-

tions ^ ^ •

—y= = 0.^
:

x {), x-\-y = 0, ic- -h 2/2

The first two of these equations represent right lines. The

third is satisfied by the coordinates of a single point, the
origin. Since this point lies on the right lines, the third
equation contributes nothing new to the locus.

EXERCISES
What are the loci of the following equations ?

3. 2x'''-^Sxy-2if = 0. 4. xy -\- x i- 2y +2 =0.

5. x^' -\- — 2 X — 2y = 0.
xy 6. x^ + xy^ = x.
7. Sx'y-2xy = 0. 8. x' - y^ -2x'- -\-2y^ = 0.
9. {x-{-y — l){x''' -\-y^~)=0. Ayis. The line whose inter-
cepts on the axes are both 1, and the origin.

10. (x-\-y){x'-{-f' + l)=:0.

11. (x y)[{x
-j- - ly + f^2= 0.
12. x^ + x'^y — xy^ — y^ z= 0.

Find, in each of the following exercises, a single equation

whose locus is the same as that of the given systems of equa-

tions.

13. a;-2 = 0, ^Z
- 4 = 0.
14. x = 2, 2/ = 4.
15. x-{-y-2 = 0, x-y-{-2 = 0.

16. x-3y = 5, 4a' + 3 = 0.

x_y x_
a ah y

5. Tangents with a Given Slope. Discriminant of a Quad-

ratic Equation. From elementary algebra we know that the
roots of the quadratic equation
 CERTAIN GENERAL METHODS 175

(1) Ax^-i-Bx -\- C = 0, A^O,

are

2A 2A
ic,
'
=- —-—
2A2A
-VB^-4.Aa

From these formulas the truth of the following theorem at

once becomes apparent.

Theorem 1. TJie roots of the quadratic equation (1) are

equal if and only if
^2-4^0=0.
The quantity ^- — ^AC is known as the discriminant of the
quadratic equation (1).
By means of the theorem we shall solve the following prob-
lem.

Problem. Let it be required to find the equation of the

tangent to the parabola

(2) f = Qx,
which is of slope \.
Let i be a line of slope \
which meets the parabola in
two points, Pj and Pg. If we
allow L to move parallel to ^
itself toward the tangent, T,
the points P^ and P^ will move
along the curve toward P, the Fig. 5
point of tangency of T\ and
if L approach T as its limit, the points P^ and P2 will approach
the one point P as their limit.
It is clear that these considerations are valid for any conic.
Accordingly, we may state the following theorem.

Theorem 2. A line which meets a conic intersects it in

general in two points. If these two points approach coincidence
176 ANALYTIC GEOMETRY

in a single point, the limiting position of the line is a tangent to

the conic*
In applying Theorem 2 to the problem in hand, let us denote
the intercept of the tangent T on the axis of y by ft. The
equation of T is, then,

(3) 2/ = i-a^ + )S.

The coordinates which T is tangent to the

of the point P, in
parabola, are obtained by solving equations (2) and (3) simul-
taneously. Substituting in (2) the value of y given by (3),
we have

or
(4) a;2 + 4(^-6)aj + 4/32 = 0.
The roots of equation (4) are equal, since they are both the
abscissa of P. Accordingly, by Theorem 1, the discriminant
of (4) is zero. Hence
16 (/3 - 6)2- 16^2 ^0, or -^ 12y8 + 36 = 0.
Thus /8 = 3, and the tangent to the parabola (2) whose slope
is -i-
has the equation
(5) x-22/ + 6 = 0.
If in (4) we set /? = 3, the resulting equation,
a;2-12a; + 36 = 0,
has equal roots, as it should. The common value is cc = 6, and
the corresponding value of y, from (2), is 1/ = 6. The coordi-
nates of the point of tangency, P, are, then, (6, 6).

Second Method. We proceed now to give a second method

of solution for the type of problem just discussed. Let the
conic be the ellipse

(6) 4iB2 + ^2 = 5^
and let the given slope be 4.

A tangent to a conic might then be defined as the hmiting position

*
of a line having two points of intersection with the conic, when these
points approach coincidence in a single point this is a generahzation of ;
 .

CERTAIN GENERAL METHODS 177

It is evident from the figure that there are two tangents of

slope 4 to the ellipse. Let the intercept on the axis of y of
one of the tangents be /8. The equation of
this tangent is then

(7) 2/ = 4a;+y8.
Our problem now is to determine the value
of p. To this end, let the coordinates of the
point of contact of the tangent be {xiy y^.
Then a second equation of the tangent is,

by (12), § 2,
(8) 4a;ia; + 2/i2/
= 5.
Since equations (7) and (8), which we FiQ. 6
rewrite as
4a;-i/ + = 0,
y8

4a;ia;-f 2/i?/-5 = 0,
represent the same line, it follows, from Ch. II, § 10, Th. 5,
that

4 -1 ^ *

From the equality of the first and third ratios we have

(9) X\ —— —
/3

Since the second and third ratios are equal.

Furthermore, the point (xi, y^) lies on the ellipse and so the
values of Xi and y^, given by (9) and (10), satisfy equation (6).
Accordingly,
100 —-=o,
,
25 .
or —=
25 1
1.

Hence p has the value 5 or — 5.

the definition of § 1. A
tangent cannot be defined as a line meeting the
conic in a single point, for there are lines of this character which are not
tangents, viz., a line parallel to the axis of a parabola, or to an asymptote
of a hyperbola.
178 ANALYTIC GEOMETRY

Substituting these values of ft

in turn in (7), we obtain

4a; — 1/ -f 5 = 0, 4x — — 5 = 0,
2/

as the equations of the two tangents of slope 4 to the ellipse

(6). From equations (9) and (10) it follows that the points of
contact of these tangents are, respectively, (—1, 1) and
(1,-1)-
Both the methods described in this paragraph are general
in application. For the usual type of problem met with in a
first course in Analytic Geometry either method may be used

with facility. It is, however, to be noted that the second

method presupposes that the equation of the tangent to the
curve at an arbitrary point on the curve is known, whereas
the first does not. Accordingly, in case a curve is given, for
which the general equation of the tangent is not known, — for
example, the parabola, y ^Zx^ — 2x-^lj — the first method
will be shorter to apply.

EXERCISES
Determine in each of the following cases how many tan-
gents there are to the given conic with the given slope. Find
the equations of the tangents and the coordinates of the points
of tangency. Use both methods in Exs. 1, 2, 3, checking the
results of one by those of the other.
 ;

CERTAIN GENERAL METHODS 179

8. What are the equations of the tangents to the circle

a;2 = 10,
+ 2/2

which are parallel to the line Sx — y-{-5 = 0?

9. What is the equation of the tangent to the ellipse

which is perpendicular to the line x-\-Sy —3= and has a

positive intercept on the axis oi y?
10. Find the equation of the tangent to the parabola
y = Sx''-2x-{-l,
which is perpendicular to the line x -\- 4:y -\- S = 0.

Ans. 4:X — y — 2 = 0.
11. Make clear geometrically that, no matter what direction
is chosen, there are always two tangents to a given ellipse,
which have that direction.

12. How many tangents are there to the parabola y"- = 2mx,
which have the slope 0? theorem relating
State a general
to the number of tangents to a parabola which have a given
slope.

13. Are there any tangents of slope 3 to the hyperbola

If so, what are their equations ?

14. The preceding exercise, if the given slope is (a) 1

(6) 2. Give reasons for your answers.

6. General Formulas for Tangents with a Given Slope.

Consider first the hyperbola

(1) ^:-i;=i.

Before attempting to find a general formula for the equations

of the tangents to the hyperbola, which have a given slope, A.,
we shall do well to ask if such tangents exist. In answer to
this question we state the following theorem.
 :

180 ANALYTIC GEOMETRY

Theorem. All the tangents to the hyperbola (1) are steeper

than the asymptotes. Tlieir slopes A all satisfy the inequality

(2) or
'
a a^

Conversely, if \ satisfies (2), there are two tangents of slope \

to (1). If, however, \^<¥/a^y there are no tangents of
slope \

to (1).
the theorem, let a point P, starting from the vertex
To prove
A, trace the upper half of the right-hand branch of (1). Then
the tangent, T, at P, starting from
the vertical position at A, turns
continuously in one direction, and,
as Precedes indefinitely, approaches
the asymptote S as its limit. In
other words, the slope, X, of T
decreases continuously through all
Fig. 7 positive values greater than the
slope, h/a, of S, and approaches h/a
as its limit.* Consequently, A is always greater than h/a

a
* The geometrical evidence of this is convincing, but not conclusive.
To clinch it, we give the following analytical proof : If the coordinates
of P are (x, y), the slope X of T is, by (11), § 2,

62x
X =
a22/

X_a
According to Ch. YIQ, § 4, eq.' (3), -=
y~b
\ x2

Hence X =
x^

When P traces the upper half of the right-hand branch of (1) and re-
cedes indefinitely, x increases continuously from the value a through all
values greater than a. Then a^/x^ decreases continuousl y from 1 and
approaches as its limit ; and 1 — a^/x"^, and hence Vl — a^/x\ in-
 CERTAIN GENERAL METHODS 181

If P now traces the lower half of the right-hand branch, A

is negative, and always :

These two inequalities can be combined into the single in-

equality (2). Thus (2) is satisfied by the slope A of every tan-
gent to the right-hand branch of (1), and hence also, because
of the symmetry of the curve, by the slope X of every tangent
to the left-hand branch.
From the reasoning given in the first case, when P traces
the upper half of the right-hand branch of (1), it follows, not
only that \ >
6/a, but also that X takes on every value greater
than h/a. Hence, if a value of A, greater than h/a, is arbi-
trarily chosen, there is surely at least one tangent of this
slope A to (1), and consequently, because of the symmetry of
the curve, there are actually two. Similarly, if a value of A
less than — h/a is given.
To find the equations of the two tangents of slope A to (1),
in the case that A does satisfy (2), we apply the first of the
two methods of § 5. Let the equation of one of the tan-
gents be

(3) y = \xJrp,
where /8 is to be determined. Proceeding to solve (1) and (3)
simultaneously, we substitute for y in (1) its value as given by
(3) and obtain the equation,

h'^x' - o? {Xx -\- PY = o?h\

or

(4) (62 _ a?X^)x^ - 2 cO^Xx - a\b'- -f ^'-) = 0.

The roots of equation (4) are both equal to the abscissa of the

creases continuously from and approaches 1 as its limit. Conse-

quently, the reciprocal, 1/Vl — a^/x"^, of Vl — a^/x"^ decreases continu-
ously through all positive values greater than 1 and approaches 1 as its

limit. Hence, finally, X decreases continuously through all positive values

greater than b/a and approaches b/a as its limit, q. e. d.
182 ANALYTIC GEOMETRY

point of contact of the tangent (3), and hence the discriminant

of (4) must vanish. have, then, We

or, simplifying,

(5) P'^
= o?X^ - b\
Hence /S has either of the values

± Va2A2 - b\

and the equations of the two tangents, written together, are

(6) y = \x± Va'X.'^ - b\

Since X satisfies (2), or the equivalent inequality a^X" — 6^ > 0,
the quantity under the radical is positive and so has a square
root."* We have thus obtained the following result.
TJie equations of the tangents to the hyperbola (1), lohich have
the given slope A, where \ satisfies the inequality (2), are given
by (6).

Let the student deduce the following results, using either of

the two methods of § 5.

The equations of the tayigents to the ellipse

a^ 52

ivhich have an arbitrarily given slope X, are

(8) ?/ = A^±Va-A2 + 5'.

The equation of the tangent to the parabola

(9) /- = 2maj,
which has a given slope X, not 0, is

(10)
y=^+f-x
* If we take a value of X, for which X2 < h'^/a'^, then a2X2 — 52 is

negative and has no square root. Consequently, there are no tangents

with this slope, as the theorem states. Finally, if X = ± 6/a, then
a2X2 _ 52 — 0, and (4) is not a quadratic equation.
 :

CERTAIN GENERAL METHODS 183

Condition that a Line he Tangent to a Conic. The two

methods used to find the tangent to a conic with a given slope
apply equally well to the problem of determining the condi-
tion that an arbitrary line be tangent to a given conic. In
fact, in finding the equations of the tangents of slope \ to the
hyperbola (1), we have at the same time shown that the con-
dition that the line

(11) y = Xx + (3,

where ive now consider Xand /? both arbitrary, he tangent to the

hyperhola (1), is that X and (3 satisfy the equation (5)

^2 = a-X' - h\

Similarly, the work of deriving formula (8) or (10) involves

finding the condition that the line (11) be tangent to the ellipse
(7) or the parabola (9).

Example. Is the line ^x — 2y -\-o =^0 tangent to the hyper-

bola x- - 4 ?/-= 4 ?
It is, if, when we write the equations of the line and the
hyperbola in the forms (11) and (1), the values which we
obtain for A, /3, a^, and 6-, namely, f, #, 4,
and 1, satisfy (5).

It is seen that they do not, and hence the line is not tangent
to the hyperbola.

EXERCISES
1. Derive formula and at the same time show that the
(8)
condition that the line (11), where now X and (3 are both arbi-
trary, be tangent to the ellipse (7) is that X and (3 satisfy the
equation

(12) ^2 = ^2^2 ^ 52_

2. Show that the line (11) is tangent to the parabola (9) if

and only if

(13) 2 A/3 = m.

Hence prove the validity of formula (10).

184 ANALYTIC GEOMETRY

3. By direct application of the methods of the text, show

that the condition that the line (11) be tangent to the circle

(14) «^ + 2/' = «'

is that

(15) ^2 = aHl + A2).

Using formulas (6), (8), and (10), find the equations of
4.

the tangents which are required in Exs. 1, 2, 3, 5, and 7 of § 5.

5. Has the hyperbola 9 a?- — 4^/2 = 36 any tangents whose

inclination to the axis of x is 60° ? Whose inclination is 45° ?

If so, find their equations.

6. Eind the equations of the tangents to the parabola

2/2 = 8.T, one of which is parallel to and the other perpendicu-
lar to the line 3a^ — 2?/ + 5 = 0. Show that these tangents
intersect on the directrix.

7. Prove that any two perpendicular tangents to a parabola

intersect on the directrix.

In each of the following exercises determine whether the

given line is tangent to the given conic. If it is, find the
coordinates of the point of contact.

Conic Line
8. 2a;- + 3?/^ = 5, 2.x- 3^/ - 5 = 0.
9. = 2a.^
?/'' a;+ 4?/ + 8 = 0.
10. 3a;- -5?/'^ = 7, 6a;- 5?/- 8 = 0.

In each of the following cases the equation of the given line

contains an arbitrar}' constant. Eind the value or values of
this constant, if any exist, for which the line is tangent to the
given conic.
Conic Line
11. a;2+ 3= 4,
T/*^ a;— 3 + c = 0. Ans.
?/ c = ± 4.
12. a;2-2/2 = 3, 2a; + - 3 = 0.
f/?/

13. 5?/2 = 3a;, - IOt/ + 15 = 0.

A:a!

14. 4a;^- 32/^ = 1, a;+ 22/ + = 0. A;

 :

CERTAIN GENERAL METHODS 185

15. Is the line x -^y =l tangent to the parabola y = x^^'^?

16. Show that the lines 3 a: ± y + 10 = are common tan-
gents of the circle x' -\-y~ = 10 and the parabola y- = 120 a;.

17. Find the equations of the common tangents of the

parabola y~ = 4V2ic and the ellipse «- + 2?/2 = 4.

Ans. ^2x±2y-\-4: = 0.
J7. Tangents to a Conic from an External Point. Given a
J)oint P external to a conic, that is, lying on the convex side of
[^the curve. From P it is possible, in general, to draw two
tangents to the conic. It is required
to find the equations of these tan-
gents. • t

Let the conic be the ellipse ~sL^

(1) 05^ + 22/2 =3
and let P be the point (- 1, 2). We
find the equations of the two tangents
drawn from P to the ellipse by find-
ing first the coordinates of the Fig. 8
points of tangency. Let P^ be the
point of tangency of one of the tangents, and let the coordi-
nates of Pi, which are as yet unknown, be {xi, 2/1)- The
equation of this tangent is then, by (12), § 2,

(2) x^x-\-2y,y = Z.
There are two conditions on the point P^, to serve as a means
of determining the values of x^and y^. In the first place, the
tangent (2) at P^ must go through the point (— 1, 2) hence ;

(3) -0^1 + 42/1 = 3.

Secondly, the point P^ lies on the ellipse (1) ; that is,

(4) a:i2 +2 2/1^ = 3.

Equations (3) and (4) are two simultaneous equations in the
unknowns Xi, 2/1- If we solve (3) for oji

(5) 0^1 = 42/1-3,

186 ANALYTIC GEOMETRY

and substitute its value in (4), we obtain, on simplification, the

following equation for yi :

(6) + 1 = 0.
32/i2- 42/1

The roots of this equation are = 1 and = i the corre- 2/1 i/j ;

sponding values of Xi are, from (5), 1 and — |. Hence {xi, 2/1)

= (1, 1) and = (— i) are the solutions of (3) and (4).

(iCi, 7/i) 1^,

The coordinates of the points of tangency are, therefore,

(1, 1) and (— f, ^). Substituting the coordinates of each point
in turn for x^ 1/1 in (2) and simplifying the results, we obtain,
as the equations of the two tangents,

(7) « + 22/-3 = and 5a;- 2^ + 9 = 0.

The method used in this example is universal in its applica-
tion, not only to conies, but to other curves as well. It should
be noted, however, that the equation corresponding to (6) does
not, in general, have rational, that is, fractional or integral,
roots. Usually its roots involve radicals and hence so do the
final equations of the tangents. If one were dealing with an
arbitrary point P external to an arbitrary conic, for example,
the ellipse
^'
+ ^' = 1,
these radicals would be complicated. Accordingly, we make no
attempt to set up general formulas for the tangents to a given
conic from an external point. We have expounded a method
which is applicable in all cases, and this is the purpose we set
out to achieve.

Second Method. We give briefly an alternative method of

finding the equations of the tangents from the point (— 1, 2) to
the ellipse (1).
Suppose one of the tangents is the line

(8) y = Xx + l3.

Since it is a tangent to (1), we have, according to § 6, Ex. 1,

2^^ = 6A2 + 3.
 CERTAIN GENERAL METHODS 187

Since it contains the point (—1, 2),

If we and (3 simultaneously, we find

solve these equations in A
that A =—^ and that ^ = |- or f
or I Substituting these pairs
.

of values for X and (3 in turn in (8) and simplifying the results,

we obtain the equations (7).

EXERCISES
1. Make clear geometrically that from a point external to
an ellipse or a parabola there can always be drawn just two
tangents to the curve.
2. How many tangents can be drawn to a hyperbola from
its center ? From a point on an asymptote, not the center ?
From any other external point ? Summarize your answers in
the form of a theorem.
3. Let P be a point external to a hyperbola from which two
tangents can be drawn to the curve. How must the position of
P be restricted, if the two tangents are drawn to the same
branch of the hyperbola? To different branches?
4. The point (2, 0) is a point internal to the hyperbola
a;2_2?/2 = 2. Prove analytically that no tangent can be
drawn from it to the curve.

In each of the following exercises determine how many tan-

gents there are from the point to the conic, and when there are
tangents, find their equations. Use the first method.

^»«-
\2x-y-5 = Q.
188 ANALYTIC GEOMETRY

11. a;2-82/ = 0, (3,2).

12. 2a;2-32/^ = -10, (-2,1).

13. a;2+ 2/2_4a;-2/ = 0, (5,2).

14. x2 + = 25,
2/' (-1,7).

15. Work Exercises 5-10 by the second method.

16. Show, by use of the second method, that the tangents
from the point (2, 3) to the ellipse 4 a;- -f- 9 ?/- = 36 are perpen-
dicular.
EXERCISES ON CHAPTER IX

1. Prove that the slope of the conic

(1 — e2)a;2+ y'^ — 2 mx -\-
m'^ =
at the point (xi, ?/i) is

_ (1
— e-)xi — m
.

A —— •

2/1
«

Hence show that the equation of the tangent at (xi, y^ is

(1 — e-) XiX + 2/i?/

—m (a; + oji) -h m- = 0.
2. Show that the slope of the curve

Ax'- H- Bxy 4- Cy'- -\- Dx + Ey + F=

at the point {xi, y^) is

2Ax, + By, + D
Bx^ -\-2Cyi-\-E

Then prove that the equation of the tangent at (xi, y^) is

Ax,x + |(2/ia; -h x,y)+ Cy^ + ^{x + x,) + ^{y + 2/i) + ^= 0.

3. The following equations contain arbitrary constants.
What does each represent ?
(a) y = Xx-\-S',
Ans. All the lines through (0, 3) except x = 0.
(b) 3ax-\-2y + a-S = 0',

(c) 7a;+ 5?/ — c + 3 = 0;

(d) {2a + 5)x+(7a-S)y = 9a + 2',
 CERTAIN GENERAL METHODS 189

(e) lx-\-{2l + m) — 3 = ?/ ?7i ;

(/) (2 Z + 3 m) x-{4.l + 6 m) y-{-5l = 0.

4. A line moves so that the sum of the reciprocals of its
intercepts is constant. Show that it always passes through a
fixed point.

5 A line with positive intercepts moves so that the excess

of the intercept on the axis of x over the intercept on the axis
of y is equal to the area of the triangle which the line forms
with the axes. Show that it always passes through a fixed
point.

6. Prove that the straight lines,

5a; -2?/ + 6 = 0,
2x-4.y + 3 = 0,
Sx + 2y-^3 = 0,
meet in a point, by showing that the equation of one of them
can be written in the form hi -\- mv = 0, where u = and v =
are the equations of the others.
7. Show that the three lines,

x-\-Sy- = 0j 4.

5x-3y-\- = 0, 6
3a; -92/ 4- 14 = 0,
meet in a point.

8. Prove that the three lines

ka — 1/3 = 0, IfB — my = 0, my — ka = 0,
where a=0, /S=0, and y=0 are themselves equations of
straight lines and k^ I, and m are constants, meet in a point.

9. Find the equation of the common chord of the two in-

tersecting circles
x^+ 2/' + 6a; -8?/ 4- 3 = 0,
2a;2 +2 ?/2 - 3a; + 4?/ - 12 = 0.
10. Show that the two circles,
a;2 -f 2/^ —4 —4 — a; ?/ 10 = 0,
a;2 _|_ 2/2 _|_ (J a; _|_ 6 2/ + 10 = 0,
 :

190 ANALYTIC GEOMETRY

are tangent to one another. Find the equation of the common

tangent and the coordinates of the common point.

11. Find the equation of the circle which goes through the
points of intersection of the two circles of Ex. 9 and through
the origin.
12. Find the equation of the circle which is tangent to the
circles of Ex. 10 at their common point and meets the axis of
X in the point x = 2.
13. What is the equation of the circle which passes through
the points of intersection of the line

2a;-2/ +4=
and the circle

and goes through the point (1, 1) ?

14. Determine the equation of the ellipse which passes

through the points of intersection of the ellipse

+4 =4
a;2 2/2

and the line 3a; — 4?/ — 3 = 0,

^and goes through the point (2, 1). By a transformation to

parallel axes (cf. Ch. XI, § 1), prove that this ellipse has axes
parallel to those of the given ellipse and has the same
eccentricity.

15. Find a single equation representing both diagonals of

the rectangle whose center is at the origin and one of whose

vertices is at the point (a, h).

16. What is the condition that the equation

a^x"^ - bY =
represent two perpendicular lines ?

17. Find the locus of each of the following equations

(a) 6a;2 4- 5xy =

- 4?/2

(b) 4:X''-20xy-}-25y'^ =
(c) a;^+ xy+ 6f- =
 CERTAIN GENERAL METHODS 191

18 Prove that the equation

(1) Ax"^ + Bxy + Cy =

represents the origin, a single straight line, or two straight
lines, according as the discriminant, B'—4:ACj is negative,
zero, or positive.

19. Show that, if equation (1), Ex. 18, represents two

straight lines, the slopes of these lines are the roots of the
equation
CX?+ B\-{-A = 0.
20. Prove that the equation
14a;2-45.T?/-142/- =
represents two perpendicular straight lines.
21. Show that equation (1), Ex. 18, represents two perpen-
dicular straight lines if and only ii A-{- C= 0.

22. Prove that the equation

y"^ — 2xy sec -\- x''- =
represents two straight lines which form with one another the
angle 6.

23. A regular hexagon has its center at the origin and two
vertices on the axis of x. Find a single equation which repre-
sents all three diagonals.' Ans. y^ —Sx-y = 0.
24. Determine the points of contact of the tangents drawn
to an ellipse from the points on the conjugate axis which are
at a distance from the center equal to the semi-axis major.

25. Find the equations of the common tangents of each of

the following pairs of conies :

(a) a;2^?/2=16, y^ = 6x;

^^^ 25'^9~ '
16^25"^
(A^
^
^4-1^=1 ^-^
25 16
= 1.
16 9 '

Draw a good figure in each case, showing the common tangents

192 ANALYTIC GEOMETRY

26. Show that the line

|=l
(2)
J+
is tangent to the circle
X" -\- y"^ = a?
if and only if -_- + -—=—.

27. Find the condition that the line (2), Ex. 26, be tangent
to the ellipse

^ + y-=l.
o? h'-
Ans. ^+^
A^ B"
=L

28. What will the condition obtained in Ex. 27 become in

the case of the hyperbola

^ _ .^ = 1 ?
29. Prove that the line (2), Ex. 26, is tangent to the
parabola 2/" = 2 mx, if and only if 2 jB- + Am = 0.
30. Find the condition that the line y = \x + /8 be tangent
to the conic
(1 — e^)x'- + 2/2 - 2 ma; + iii^ = 0.
Ans. {(3 + mA)- - e- {(3- + m-) = 0.

31. In an ellipse there is inscribed a rectangle with sides

parallel to the axes. In this rectangle there is inscribed a
second ellipse, with axes along the axes of the first. Show
that a line joining extremities of the major and minor axes of
the first ellipse is tangent to the second.
 X CHAPTER X
POLAR COORDINATES
1. Definition. It is possible to describe completely the
position of a point in a plane by telling its distance and its

direction from a given point. This idea forms the basis of the
system of polar coordinates.
Let be the given point, and draw from a ray, OA, from
which to measure angles. Let P be any point of the plane.
Denote its distance from by r, and the
angle AOP by 6. Then (r, 6) form the
polar coordinates of the point P. is

called the pole or origin ;OA, the prime ^„.j^-

direction or initial ray ; and r, the radius
vector (pi. radii vector es).
When r and 6 are given, one, and only one, point is deter-
mined. When, on the other hand, a point is given, r is com-
pletely determined, but 6 may have any one of an infinite
set of values differing from one another by multiples of 360°
(or 27r). Thus, if 6' is one value of $, the others will all be
comprised in the formula

^ = ^'±360n (or (9'±27r7i),

where n is a whole number.

For the point 0, r = but there is no more reason for
;

assigning to 6 one value rather than another. As the coordi-

nates of 0, therefore, we take (0, 6), where 6 may be any num-
ber whatever.
It is possible to define polar coordinates so that r can be
negative. Thus the point (—2, 30°) would be obtained by
193
 194 ANALYTIC GEOMETRY

drawing the ray which makes an angle of 30° with OA and

then laying off on the opposite ray a distance of 2 units. This
system of polar coordinates is not
widely used in later work in mathe-
matics, and even in analytic geometry
it is a matter of custom rather than
""
p 2
of any logical necessit;f. We shall,
therefore, adhere to the original
definition and exclude negative values of r, unless an explicit
statement to the contrary is made.

EXERCISES
For use in these and later exercises the student should pro-
cure polar coordinate paper, ruled like a cobweb. Otherwise
he should use a scale and protractor.
Plot the following points :

1. (1, 0°). 4. (5, - 30°). 7. (3, 180°).

"^

2. (0,1°). 5. (2,200°). 8. (4,1^). f^tr

3. (5,30°). 6. (2,-90°). 9. (6, Jtt).
\^^

4CH 10.
whose center
What are the coordinates of the vertices of a square
is at 0, the prime direction being perpendicular
to a side, if the length of one side is 2 a ? VJ '

11. Write down the coordinates of the vertices of an equi-

and one vertex
lateral triangle, the pole being at the center
lying on OA.
12. The same for a regular heptagon.

13. What loci are represented by the following equations?

(a)r = 5; (6)cos^ = 0; (c) (9 = 90°.
2. Circles. Among the simplest curves^n polar coordinates
\\ are
(a) the circles with center 0. The equa- / ^
tion of one of them is

(1) r = a,
where a is the radius. Fig. 3
 ;

POLAR COORDINATES 195

«

(6) the circles which pass through 0.

Begin with one whose center lies on OA.
If its radius is a, then^ evidently its equa-
tion is

r = 2 a cos B.
Fig. 4
If the coordinates of the
center of an arbitrary circle through are
(a, y), then the equation is
i-V J \

^^^ZI7 A (3) r = 2 a cos {Q - y).

Fig. 5
EXERCISES
v^' Plot each of the following curves (making a
directly
convenient numerical choice of a as, for example, 2 cm.): ;

(a) r = 2 a sin Q ;

(6) r = —2a cos ^

(c) r = — 2 a sin ^.

1/2. Obtain each of the equations in Ex. 1 as a special case

under (3), by choosing y properly.
3.*" Circles are described with their centers at the vertices
of the equilateral triangle of Ex. 11, § 1, each circle passing
through the center of the triangle. Eind their equations.
4. A circle of radius 2 has its center on OA at a distance 3
from 0. Show that its equation is

7^ - 6 r cos ^ + 5 = 0.
5. A circle whose radius is 4 has its center at the point
(5, 90°), -Show that its equation is

r2 — 10 r sin ^ -f- 9 = 0.
.jo. Show that the equation of any circle is

r2 — 2 cr cos (^ — y) -f c2 = p2^
where p denotes the radius and (c, y) are the coordinates of the
center.
What curve is represented by each of the following equa-
tions ?
196 ANALYTIC GEOMETRY

7. r2+8rcos^ = 9.
8. r2 4-8rsiii^ = 9.
9. r'^-Sr sin = 9.
10. r2 4- — 2 r sin = 7.
2 r cos ^ r
11. r2— 2rcos ^ + 2rsin^ = 7.
12. — 6 r cos ^ — 8 r sin ^ = 11.
7-2

, straight Lines. Let us consider first a line L which does

pass through 0, and assume, to begin with, that L meets
the prime direction at right angles at the
distance h from 0. The equation of L is,

evidently, /

(1) r cos 6 =ih.

If L is parallel to the prime direction

and at a distance h above it, it is easily
shown that its equation is
'

(2) r sin (9 = h.
Let L, now, be any line not going through 0. Draw a line
through perpendicular to L, and let B be the point in which
it cuts L (Fig. 7). Denote the length ^
of the line-segment OB by h, and the
^ AOB by y. Let P:{r, 6) be any
point of L.
Then :^BOP=0-y.
Consequently, we have
(3) r cos {e-y) =h
Fig
as the equation of L. i

Mays from 0. The equation of a ray, or half-line, emanating

from 0, is •
'

6 = a,
where a is a constant angle. Thus ^ = is the equation of the
prime direction, and = 90° is the equation of a ray drawn
 POLAR COORDINATES 197

from at right angles to the prime direction. There are two

such rays ; the equation of the other one is ^ = — 90°.
To the right-hand side of any of these equations can be added
any positive or negative multiple of 360°, without altering the
locus.

Lines through 0. The equation of the line through O per-

pendicular to the prime direction is

(4) cot ^ = 0.
For then ^ = 90^ or 6 = — 90°
and we have just seen that these are equations of the two rays
making up the line.
The equation of any other line through is

(o) tan 6= c,

where c is a constant. Thus

tan ^ = 1

represents a line through 0, for the points of which one or the

other of the two equations

^ = 45° or ^=225°
holds.
EXERCISES
1. Establish equation (2). ,

2. Derive equations (1) and (2) from equation (3). '^^^ ^

3. If a line is perpendicular to the prime direction but
does not cut it, what is its equation ? Let h be its distance
from the pole.

4. Eind the equation of a line which is parallel to the prime

direction and a distance h below it.

5. Find the equation of the line which cuts the prime

direction at a distance of 5 units from the pole and makes an
angle of — 45° with the prime direction.
What does each of the following equations represent?
Make a plot in each case.
198 ANALYTIC GEOMETRY

J.6. r cos 6 = 4:. 7. r sin ^ = 4.

8. r cos = — 4. 9. r sin ^ = — 4.
10. r sin ^ + r cos = S 6. 111.
11. r sin — r cos ^ = 3.
'12. 4r sin ^ + 3?' cos ^ = 5. ^4l3. 5r sin^ — 12r cos 6 = 26.

14. tan ^ = — 1. 15. 2 cos = 0. (9

16. ^ = 60°. 17 = 180°.

X
4. Graphs an equation in polar coordinates
of Equations. If
is given, which, cannot be reduced to one of the forms recog-
ized as representing a known curve, it is necessary, in order
to determine what curve is defined by the equation, to plot a
reasonable number of points whose coordinates satisfy the
equation. But considerations of symmetry will often shorten
the work.
Example 1. Consider the equation

(1) r2 = 16 sin 6.

This equation is equivalent to

(2) r = 4 Vsin 6,
where we have taken only the positive square root, since nega-

tive values of r have for us no meaning.

When ^ = 0, = as ^ increases, r increases, and when
r ;

= 90°, r = 4. Using a table of sines and a table of square

roots, we compute the following coordinates of further points
of the curve.
d 10° 20° 30° 40° 50° 60° 70° 80°
1.67 2.34 2.83 3.21 3.50 3.72 3.88 3.97

More computations are unnecessary.

For, the curve is symmetric in the ray
$ = 90°. To prove this, Ave note that, if

P : any point of the curve, then

(r, 0) is

the point P' (r, 180°-^), which is

:

symmetric to P in the ray = 90°, is

also a point of the curve, inasmuch as

sin (180° - ^) = sin 0,

 POLAR COORDINATES 199

We now have
points on the curve for values of from 0° to
180°. These points determine the entire curve, since, if 6 is
greater than 180° (and less than 360°), sin 6 is negative and (2)
is meaningless.

iioj

Fia. 9

To make sure that the curve has not sharp corners at O and
B, we must compute r for small values of and also for values
of near 90°.

d
 :

200 ANALYTIC GEOMETRY

creases, r decreases, and when 6 = 30°, = 0.

?* The coordinates
of intermediate points of the curve are
Hence every point on the
POLAR COORDINATES
first lobe of the curve yields a point
(^
on the second lobe by merely rotating the radius vector
through 120°.
The second lobe again gives rise to a third lobe congruent
to itand advanced by 120°. If this last lobe were again ad-
vanced, it would yield the first. Hence
the three lobes complete the curve.
If we admit negative r's, the curve is

unchanged. The points which we then

get, for example, for values of between
30° and 90° lie on the third lobe of the
curve. Thus, for values of 6 from 0° to
360°, each point of the curve is obtained
twice, once for = 6' and once for Fig. U
6 = 0' + 180°.
Tests for Symmetry. Let the student show that the test for
symmetry in the ray = 90°, given in Example 1, also insures
symmetry in the ray = 270°, and that the test for symmetry
in the prime direction, given in Example 2, also yields sym-
metry in the ray = 180°.
These tests are general, and can be stated as theorems.

Theorem 1. A cAirve is symmetric in the line of the prime

direction if, on substituting — for in its equation, the equa-

tion is unaltered.

Theorem 2. A curve is syimnetric in the line through the

pole perpendicular to the prime direction if, on substituting
180° - for in its equation, the equation is unaltered.

EXERCISES
5iot the following curves.
(l^ The lemniscate (take a =5 cm.),

^ cos 2
9-2 0^2 0.

2. The cardioid (take a = 2i cm.),

r = 2a(l — cos(9).
 202 ANALYTIC GEOMETRY

(a generalization of the cardioid),

f^ The limaQon
r = 4 — 3 cos 6.

4. A second type of limaqon,

r = 3 — 4 cos 6.
Show that if negative r's are admitted, a piece is added to the

curve.

5. r2 = 16cos(9. 6. r = 10 sin 3^. 7. r2 = a'^sin2^.

8. How are the curves of Exs. 5, 6, and 7 related, respec-
tively, to the curves of Examples 1, 2 of the text and the lem-
niscate of Ex. 1 ?

9. r = —+ ^—'
-
1 cos (9
10. r = sec2|.
^

^X -

11. r = _ f3 - -' 12.

' r =
1 I cos 61 1 + 2 cos ^
13. r = 5cos2^.
Show that this curve has two lobes, but that it would have
four lobes, if negative r's were admitted.
14. r=:5cos4^. 15. r = acos9i^.
16. Show that the curve of Ex. 15 has n lobes
but, ;
if n is

even and negative r's are admitted, it has 2 n lobes.

17. The spiral of Archimedes,

r = e,
taking in degrees and -J^ cm. as the unit of length.

^ The
^.j hyperbolic spiral,
1

taking in radians and 2 cm. as the unit of length.

19. r = l-0\ 20. r- + ^2 = i.

5. Conies. The equation of a conic section, when the defini-

tion of Ch. VIII, § 7 is used, is simple in polar coordinates.

 POLAR COORDINATES 203

Let the pole be taken at the focus F, and

let theprime direction be chosen perpendicular
to the directrix D and away from D\ let

If P : (?', 0) is a point lying to the right of D

(Fig. 15) and on the conic, then

FP= r, MP = r cos -{- m.

Now, by definition,
FP = e,
MP
and hence the equation of the locus of P is
r
= e.
?* cos ^ +m
or, if we solve for r,

= em
(1) r
1 — e cos
Ellipses and parabolas lie to the right of D and hence are
represented by when e < 1 and
^ (1),
e= 1, resj)ectively.

When e > 1, however, (1) repre-

^ sents ]ust the right-hand branch of a
hyperbola. The equation of the left-
Fia. 16 hand branch is

r = em
(2)
1 -f e cos 6

For, if P: (r, 6) lies on the left-hand branch (Fig. 16), it is to

the left of D and
PM= — r cos — m.
We then have, since FP/P3f= e,

r
= e.
— r cos 9 —m
which reduces to (2).
If negative r's are admitted, the single formula (1) gives
both branches. For, in this case we may take r, for a point
204 ANALYTIC GEOMETRY

\D P on the left-hand branch, as nega-

tive (Fig. 17). Then we have
K
"' F -J FP=-r, PM =— r cos — m.
PA--"' M and hence = e.
Fig. 17 m — r cos
which reduces to (1).
It is seen, then, that the choice we have made, admitting
only positive or zero more discriminating, for we are
7''s, is

able to represent a single branch of the curve by a simple

formula, —
(1) or (2). In analytic geometry we do not usu-
ally care to do this, the curve that interests us being the pair
of branches. But in applied mathematics it often happens that
one branch of a hyperbola plays a role and the other has no
meaning. Thus when a comet is traveling in a hyperbolic orbit,
it is only one branch of the hyperbola which forms the path.*

New Choice of Prime Direction. If the prime direction had

not been chosen along KF
produced, but at an angle y with
it, as shown in Fig. 18, then evidently

6 —y would take the place of 6 in the

foregoing formulas, but there would be
no other change. The final equations
would now read
= em
(3) r
1 — e cos (6 — y)
= em
(4) r
1 + e COS (6 — y)
* We note that this is not the only way in which we are able, by
simple formulas, to discriminate between the two branches of a hyper-
bola. Thus the equation
- x2 + ?/2 _ 1

represents a hyperbola on the axis of y. The equation

y = Vl +x^
represents one of its branches, and the equation

2/ = — Vl -h ic2
the other.
 POLAR COORDINATES 205

Example. What curve is represented by the equation

= 6 „
r ?
1 + 2 sin ^
The equation can be reduced to the form (3) by choosing y
so that
cos ($ — y)= — sin 6.
Obviously, y must be 270° what amounts to the same
or,

thing, — 90°. and m = 3.

Moreover, e =2
Thus the equation represents one branch of a hyperbola ^

whose eccentricity is 2. The transverse axis is perpendicular

to the prime direction and the branch in question is the one
opening downward. The center of the hyperbola is at the
point (4, 90°). Its asymptotes make angles of 60° with the
transverse axis. The lengths of the semi-axes are a =2 and
h = 2 V3. The vertices are at the points (2, 90°) and (6, 90°).
The second focus is at (8, 90°) and the equations of the direc-
trices are r sin ^ =3 and r sin = 5.
Let the student verify each of these statements, using the
formulas of Ch. VIII, § 6, and then draw the curve to the
scale of 1 cm. as a unit, marking each of the points mentioned
with its coordinates and drawing in the asymptotes and direc-
trices. What is the length of. the latus rectum ?

EXERCISES
What conic or, in the case of a hyperbola, what branch is

represented by each of the following equations ? Draw a

rough figure showing the position of the curve.

1. r =
1 —
3
|- cos
^2
AJ.

3. r= 12
l-3eosO
7.2
5. r =
l-.8sin(9'
24
7. r =
l-t-4sin6>*
 206 ANALYTIC GEOMETRY

10
9. r = 10. r =
5 + 3 cos ^ — 4 sin ^ 1 + sin ^ + cos 6

11. r=^
-2
1 + sin — cos 6

12. Draw an accurate figure, to scale, for each of the curves

of Exs. 5, 6, and 7, marking the coordinates of all the impor-
tant points and drawing in all the important lines.

6. Transformation to and from Cartesian Coordinates. Let

P be any point of the plane, whose coordinates, referred to a
pair of Cartesian axes, are Let the (x, y).

polar coordinates of where P be (r, 0),

the origin, 0, is taken as the pole, and
the positive axis of x as the prime direc-
tion. Then it is clear from the figure that
(1) x = r cos $, y = r sin 0.
i Thus X and y are expressed in terms of r and 0. To express
r and in terms of x and y, we have, for r :

(2) r'2 = x^-{- 2/2, or r = \/x^ -j-

y
and, for 0, the pair of equations

x ^
(3) cos 6 = sin e =
V a;2 -f- 2/^^ y/x^ + 2/2

For ^ we have, also, the equation,

(4) tan
X
But not all values of satisfying this equation are admissible.
Some determine the ray OP, as they should the others give ;

the opposite ray and are to be excluded. If 6 = 6' is one ad-

missible value, the others are 6 =^ 6' -{ 360 7i, where n is a
whole number.

Example 1. What are the polar coordinates of the point

(-5,-5)?
 :

POLAR COORDINATES 207

L
Here, r = VS^ + 52 = 5 V2, tan = 1.
But = 45° is not a correct value of 0, for the point lies in the
third quadrant. The values of are, then
= 225° + 360° n.
It is frequently of importance to obtain, from the equation
of a curve in one system of coordinates, its equation in the
other system. We illustrate the method of doing this by a
number of examples.

Example 2. Find the equation of the equilateral hyperbola

x^ — ?/2 = Cl-

in polar coordinates.
Replacing x and y by their values as given by (1), we have :

r2 cos2 d — r"- sin2 $ = a^,

or r2 cos 20 = a-.
Example 3. Transform the equation of the lemniscate,
r2 = a2 cos 2 0,
to rectangular coordinates.
We perform the transformation piecemeal, first getting rid
of 0. Write the equation as
r2 = a"(cos2^ - sin2^),
and then replace cos and sin by their values, x/r and y/r,
from (1) on multiplying both
; sides of the resulting equation
by r2, we have
r4 = a2(x2-2/2).
Finally, we replace ?-2 by its value x^ + y"^, and obtain
(a;2 + = a2(cc2_
2/2)2 2/2).

This is an equation of the fourth degree in x and y.

Example 4. Transform the equation of the curve of Exam-

ple 1, § 4

(5) r2=a2sin6'
to rectangular coordinates.
 —
208 ANALYTIC GEOMETRY

Eeplacing sin 6 by y/r and multiplying through by r, we

have
(6) r^= ah^.

This becomes

(7) ( Va;2 + 2/2)3 = a?y^ or V(a;- -f 2/')^ = «^2/-

Negative Values of r. If we admit negative values of r, then

(2) and (3) become
r = ± Vic2 + 1/2,

cos^ =^ . sin^ '

± VX^ + ?/^' ± ViC2 -I- 2/2

where the plus signs are to be taken if ?- is positive, the

minus signs if r is negative ; (1) does
not change, as Fig. 20 shows. The
admissible solutions for of (4) are
'
, ^^"^ ^ those determining the ray OP (Fig.
19), if r is positive, or those deter-

FiG. 20 mining the ray OP (Fig. 20), if r is

negative.
If in (5) negative values of r are admitted, then (6) be-

comes, since now r = ± -\/x- + y^,

± V{x' + y-y = a%
which may be written as

(8) (x'^^fy=aY-
The fact that (5) transforms into (7) when negative r's are
excluded and transforms into (8) when negative r's are ad-
mitted corresponds to the fact that in the first case the curve
(5) consists of a single loop, whereas in the second it is made
up of two loops (§4).
This situation is not met with in the case of the lemniscate,
Example 3, since this curve is the same whether negative r's

are excluded or admitted.

 . : :

POLAR COORDINATES 209

EXERCISES
1. Find the Cartesian coordinates of the points
(a) (2,60°); (6) (5,120°); (q) (10,225°);
(d) (3.281, 110° 32'); (e) (2.847, 242° 27').
Plot the given point each time and check the results by direct
measurement.
2. Find the polar coordinates of the points
(a) (6,6); (6) (-2,-2); (c) (2,3);
(d) (-4,3); (e) (7,-8); (/) (-12,-5).
Transform the following equations to polar coordinates.
3. x = S. 4. 2/ = — 4.
5. y=Sx. e. 2x — Sy = S.
7. ?/2 = 4a;. 8. xy = a^.
9. x'^-\-y'^-2x + 4:y = 0. 10. 4x2 + 3z/2=12.
11. Transform the equation of the cardioid,
r = 2a(l — cos 6),
to Cartesian coordinates. Of what degree is the resulting
equation ?

Ans. (x2 + 2/^ + 2 axy = 4 a^ (a;2 -f- y'^), of the fourth degree. ^

Find the equation in Cartesian coordinates of each of the

following curves.
12. r'^=a^Gose, 13. = 0^81^12$.
r"-

14. r = cicsc^. 15. = 4 — 3cos^.

r
16. r = acos3<9. Ajis. y'-y = ax (x^ - S y^)
(x^- -\-

17. r = a sin 3 0. Ans. + y^y = ay (3 — y^)

(x'^ x'^

tlS^ r = asm2e. Ans. V(a;2 + y^y = 2 axy

19. r = a sin 2 6, if negative r's are admitted.

20. r = a cos 2 6. Ans. V(a;2 -f 2/2)3= ^j (^^,2 _ ^2)

21. r = a csc2 6 ^?is. 2/2 — 4 ao; — 4 a2 = 0,
210 ANALYTIC GEOMETRY

22. ,. = __!!^,ife<l.
1 — e cos

Ans. (1 — e^) x^-i-y"- — 2 e^mx — e^-m? = 0.

23. r =
1
—
— e cos
,ife>l. Ans. ^x"" + y^ = e{x-\- m).
24. Same as Ex. 23, if negative r's are admitted.
^?is. That to Ex. 22.

EXERCISES ON CHAPTER X
1. Show that the distance between the two points (?'i, Oi),

(tz, O2) is given by the formula

D = Vri2 + ro^ - 2 r^r^ cos (^1 - ^2)-

2. Deduce a formula giving the area of a triangle, one of

whose vertices is at the pole.

3. Determine the angle of intersection of the two lines

r(sin e + cos 0)= 3, r (4 sin 6*

-f 3 cos 0)= 5.

Suggestion. Put the equations into the normal form,

rcos(^ — y)= h,
and thus find the value of y for each line.

4. Show that the line,

a
is represented in polar coordinates by the equation

= ah
r
a sin 6 -\-h cos 6

5. What lines are represented by the following equations ?

Plot the line each time.

(q\ y
^^ = *
(1)) T = •

sin(9 + cos^' ^^ 2 sin (9 -3 cos ^

Eind the equations in polar coordinates of the following
conies.

6. 2/^ = 2 mx. Ans. r = 2mcosO csc^ 0.

 POLAR COORDINATES 211

a262

a- ft' -a2sin2(9 + 62cos2^

9. What curves are represented by the following equations ?

(a) r2 = 400 ^, ^ 4
25 sin2 (9 + 16 cos2 ^' ^
^
cos2 ^ - sin2 e '

''
2 cos2 ^ + 3 sin2 ^l
cos^ (9- 2 sin2^'

^^ ^-^^
sin26l' cos^d
10. Transform the equation of the circle,

{X - a)2 + (3/
- hy = p2,
to polar coordinates; represent the polar coordinates of the
center by (c, y). Ans. r^ — 2 cr cos (^ - y) -|- c2 = p^.
CONICS
11. A
comet moves in a parabolic orbit with the sun as
focus. When
the comet is 40,000,000 miles from the sun, the
line from the sun to it makes an angle of 60° with the axis of
the orbit (drawn in the direction in which the curve open's).
How near does the comet come to the sun ?
12. A comet is observed at two points of its parabolic orbit.
The which is the vertex
focal radii of these points, neither of
of the parabola, make an angle of 90° with one another and
have lengths of 10,000,000 and 20,000,000 miles, respectively.
Find the equation of the orbit and determine how near the
comet comes to the sun.
13. An ellipse which has a focus at the pole and its trans-
verse axis along the prime direction passes through the two
points (4, 60°) and (2, 90°). What is its equation? Where is
the second focus?

14. A hyperbola has its transverse axis along the prime

direction and a focus in the pole. The branch adjacent to this
212 ANALYTIC GEOMETRY

focus goes through the points (^V2, 45°), ( V2, 90°). Find the
equation of this branch.
15. Show that in a parabola a focal radius inclined at an
angle of 60° with the direction in which the curve opens is
equal in length to the latus rectum.
16. Show that a focal radius of a hyperbola which is parallel
to an asymptote is equal in length to a quarter of the latus
rectum.
17. Prove that in any conic the sum of the reciprocals of
the segments of a focal chord is constant.

Prove that the length of the

18. focal chord of any conic is

given by the formula

2 em
l-e2cos2^o'
where 6q is the angle which the chord makes with the trans-
verse axis.
19. Show that the sum of the reciprocals of the lengths of
two perpendicular focal chords of a conic is constant.

EOTATION OF THE PrIME DIRECTION

20. Let the prime direction OA through
be rotated about
the angle 6q into a new position 0A\
Let an arbitrary point
have the coordinates (r, 6) with respect to as pole and OA as
prime direction and the coordinates (r', 6^) with respect to as
pole and OA^ as prime direction. Show that ^
r' = r, 0' = e- 0,.

21. Equation (1), § 5, when e < 1, represents an ellipse with

one focus in O and the other on OA. From it obtain, by rota-
tion of the prime direction, the equation of an ellipse with one
focus at and the other on the ray = 90°.
22. Equations (1) and (2), § 5, when e > 1, represent a
hyperbola with one focus in and the other on the ray ^ = 180° ;

obtain from them the equations of a hyperbola with one focus

at and the other on the ray = 90°.
 POLAR COORDINATES 213

23. Obtain equations (3) and (4), § 5, from equations (1)

and (2), § 5, by a rotation of the prime direction.
24. What does the equation of the line

r(sin ^ H- cos ^) =3
become, when it is referred to the perpendicular to it from the
pole as the new prime direction ?
25. By rotating the prime direction through a suitable angle
reduce the equation of the circle, ?• = 6 cos {6 — 30°), to simpler
form.
26. The same for the circle, r = 4 cos ^ -f 3 sin 6.
27. The same for each of the conies :

(a) r = : (h) r =
3 _ 3 cos - 4 sin
(9
;
|9'
V2 + sin ^ - cos

28. Prove that the curves r = a sin 2 and r = a cos 2 6 are

the same curves, referred to a common pole and to prime direc-
tions making an angle of 45° with one another.

29. Show that the equations 7'=acos3^ and r = asin3^

represent the same curve.
30. Show that the equation of the curve r = a cos 3 remains
unchanged if the prime direction is turned through any angle
which is an integral multiple of 120°.

31. The same for the curve r = a sin 4^, if the angle is 90°.

Pole in an Arbitrary Point

32. Given the point (2, 3) in the Cartesian plane. The polar
coordinates of a point F, referred to (2, 3) as pole and to the
directed line through (2, 3) in the direction of the positive axis
of X as prime direction, are known to be (2, 13°). What are
the rectangular coordinates of P?
Ans. (2 + 2 cos 13°, 3 + 2 sin 13°)
33. The polar coordinates of a point P, referred to the point

(^oj Vo) ^s pole and the directed line through {Xq, ?/o)
in the
214 ANALYTIC GEOMETRY

direction of the positive axis of x as prime direction, are

(r, 6). Show that the rectangular coordinates, (ic, y), of Pare
x = XQ-\-r cos 9, y = yo-\-'^ sin 9.

34. By the direction 9 is meant the direction which the posi-

tive axis of X would assume if it were rotated about one of

its points (in the positive sense of rotation) through the angle 9.

From one proceeds 2 units distance in the

the point (5, 2)

direction 135°, and from the point thus reached one proceeds 3
units distance in the direction 60°. What are the coordinates
of the final position?

35. Prove that the equations of Ex. 33 can be considered as

the equations of a transformation of coordinates, from (x, y) to
(?', 9), which consists first of a change of origin to the point
(xo, ?/o)
— cf . Ch. XI, § 1 — and then of the introduction of polar
coordinates.

36. By shifting the origin to the point (2, 1) and then in-

troducing polar coordinates, identify the locus of the equation

2/2- 2a;- 2?/ + 4 = 0.

37. By shifting the origin to a suitable point and then in-

troducing polar coordinates, identify the locus of the equation

.(a.2 _|_ y2 _ 2a; - 2?/ -h 2)2 = 25(ic2 - 2/2 -2x-\-2y).
Ans. A lemniscate.

Loci
Solve the following problems in loci, using polar
coordinates and excluding negative r's. In Exs.
39-41, determine when two equations are necessary
to represent the locus.

38. Cissoid of Diodes. OA is, a fixed diameter

of a circle. A meets
variable secant through
the circle in M and the tangent De- at A in N.
termine the locus of the point P, so situated on the
segment ON that OP = MN, and plot it.

39. Limagon of Pascal. A variable secant through a fixed

 )

POLAR COORDINATES 215

point of a circle of diameter a meets the circle again in R.

The constant length h is laid off in both directions along the

secant from R. Find the locus of the two points thus reached.
Show that, if a = b, the locus is a cardioid. Plot the locus
(a) when a = 4, 6 = 5 (b) when a = 4,
;
= 3. ft Cf. § 4, Exs. 2,
3,4.
40. Conchoid of Nicomedes.
A variable straight line through
a fixed point meets a fixed
straight line, at the distance a
from 0, in Q. From Q the
constant length h is laid off
in both directions along QO.
Find the locus of the two b>a b=a b<a
points thus reached. Plot it

for each of the following pairs of values of a and h :

a = 4, 6 = 6 ; a =6=4 ; a = 4, 6 = 3.
41. Ovals of Cassini. Given two points, Fi and F2, with
coordinates (a, 0°), (a, 180°). Determine the locus of a point

6=a
O G b<a
b>a
P which moves so that the product of its distances from Fi
and F2 is constant, and equal to ft\ Show that, if a- = 6-, the
locus is a lemniscate. Plot the locus (a) when a — Q, b =l
(b) when a = 6, 6 = 5.
 ;

CHAPTER XI
TRANSFORMATION OF COORDINATES

1. Parallel Axes.It sometimes happens that it is desirable

to shift from a given system of Cartesian axes to a new
v system of axes having the same direc-
P
c tions as the old, but with a different
origin.
Let P be any point of the plane
let the coordinates of P, referred to
-X
X' the old axes, be (oj, y),and let the
Fig. 1 coordinates be {x\ y') with respect to
the new axes. Let the new origin,
0', have the coordinates {xq, y^ in the old system. Then it is

easy to show that

I
ic = 4- Xq^
a;'
(1)
y = y' + yo;
or
=X ix' X.0)
(2)
= y- {
I
y' 2/0.

For, consider the line-segment OP and the broken line, 00' P,

which has the same extremities. Then
Proj. 0P= Pro j. 00' -f Proj. O'P,

no matter what direction is chosen, along which the projection

is to take place (Introduction, § 3).

If the direction is taken, lirst, as the positive axis of x and

then, again, as the positive axis of y, we obtain, by applying
the definition of coordinates (Ch. I, § 1), the equations (1).

Example 1. Find the equation of the curve

(3) 2/^ + 22/ -40^4- 9 = 0,
216
 :

TRANSFORMATION OF COORDINATES 217

referred to parallel axes, with the new origin v

at the point (2, — 1).

Here, Xq = + 2, 2/o =- 1,

and we have
(4) x = x' + 2, y = y' - 1.

Hence
(2/' -1)2 + 2(^-1)- 4(0.' + 2)+ 9 = 0, Fig. 2

or, on simplification,

Thus the curve is seen to be a parabola whose vertex is at

the new origin. Referred to the old axes, the vertex is at the
point (2, — 1) and the focus at (3, — 1).
Example 2. What curve is represented by the equation

9a;2 + 42/2 + 18x - 16 y= 11 ?

We can rewrite the equation in the form

9(a;2_|_2a; )+4.{y''-4.y )=11.

The first parenthesis becomes a perfect square if 1 is added.
This means that 9 must be added to each side of the equation.
Again, the second parenthesis becomes a perfect square if 4
is added. This means that 16 must be added to each side of
the equation. Hence, finally,

9(a;2 -h 2a; -f 1)+ 4 (2/2- 4?/ + 4)= 11 + 9 + 16,

or 9(a;-f l)2 + 4(2/-2)2 = 36.

If we transform to parallel axes, setting

\x! = x-\-l, Xo = — 1,
l2/' = 2/-2, 2/0 = 2,.
the equation becomes
9a;'2 + 4/2 = 36,
or
4 9
 :

218 ANALYTIC GEOMETRY

This equation represents an ellipse with its center at the

new origin, (% 2/0) =( — I5 2) its semi-axes are of lengths 2
;

and 3 and its foci lie on the i/'-axis at the

points
ix',y')=^{0, ±V5),
or (x,y) = {-l,2±V5).
If, in Example 1, the position of the
new origin (the vertex of the parabola)
had not been given, it could have been
found by the method employed in Ex-
ample 2. Equation (3) can be written as
Fig. 3
y2^2y = ^x — 9.
To complete the square of y"^ -^2y, add 1 to each side of the
equation
2/2 + 2^ + 1 = 403-9 + 1.
Put this into the form :

(2/ + 1)2 = 4(03 -2).

Hence we are led to set

x' = x-2, 2/' = + 1,

2/

that is, to transform to parallel axes with the new origin at

the point (2, — 1). But this is precisely the transformation

(4) which was applied in Example 1.

EXERCISES
1. Find the coordinates of the points (3, 2), (—2, 5),
(—4t, — 1),
(0, 0), referred to new axes having the
same direc-
tions as the old, if the new origin is at the point

(a) (^'0,2/0) = (1,1); W N, 2/0) = (5, -3).

In each of the following exercises transform the given equa-

tion to parallel axes having the same directions, the new origin
being at the point specified. Thus identify the curve repre-
sented by the equation, and describe carefully its position with
respect to the original axes. Draw the curve roughly.
 ;

TRANSFORMATION 0F\ COORDINATES 219

Equation New Origin

2. 2/^ = 2x^4, K,2/o) = (-2,0).
Ans. A parabola with its vertex at (—2, 0), with its focus
at (— f , 0), and with a; = — f as its directrix.

3. 2/2 = 2.2/ + 4 X, (x„ 2/o) = - h 1).

(

4. 3x-2-6.r-42/ + ll = 0, (a^o, 2/o) = (l, 2).

5. 9a;2 4-25.?/2 4-18x-50?/-191=0, (a^o, 2/o) = (- 1, !)•
ylrzs. An ellipse, center at (—1, 1); foci at (3, 1) and
(—5, 1) ; semi-axes of lengths 5 and 3.
6. 0^2 _ 4^2 -6x-32y-o9 = 0, (o^o, 2/o) = (3, - 4).

Show that each of the following equations represents a conic

section. Draw a rough graph of the conic and find, when
they exist, the coordinates of the center and the foci, the
equations of the directrices and the asymptotes, and the value
of the eccentricity.*

7. x'2— 2ic — 42/ + 5 = 0.

'

8. 3^2 -6a: +02/ + 13 = 0. .

9. 2/^-12 + 42/ + 28 = 0. a;

10. 42/2 + 3a: -242/ + 42 = 0.

11. 2a;2+32/2-4a;-62/-l = 0.
12. + 42/2 +2a:-242/ + 36 = 0.
a:2

13. 4a:2 - 92/2 - 16a; + I82/- 29 = 0.

Ans. A hyperbola, center at (2, 1) foci at (2 ± Vl3, 1) ;

directrices = 2 ± ^^VlS asymptotes 2x + 3y — 7 = 0,

: a: ; :

2x-3y-l = 0; e = WIS.
14. 2x2 -32/2 + 20a: -12?/- 44 = 0. •

15. + 42/2 + 14a: + 45 = 0.

a:2

^
16. a:2- 252/2-502/ -50 = 0.

^/2. Rotation of the Axes. Let the new (a:', y')-2ixes have
\^ the same origin as the given (x, 2/)-axes, and let the angle

* Further exercises of this type are Exs. 1-9, of Ch. XII, § 1.

 : :

220 ANALYTIC GEOMETRY

from the positive axis of x to the positive axis of x^ be de-

noted by y (Fig. 4). Let P be any point, whose coordinates,
referred to the old and the new axes,
are (a;, y) and {x\ y') respectively, and
let M
and M' be the projections of P
on the axes of x and x' respectively. '

Then is joined with P by two broken

lines, namely, OMP and OM'P. It

follows that the projections of these
broken lines along any direction are equal

(1) Proj. 0M-\- Proj. MP = Proj. OM' + Proj. 3/ 'P.

If the direction is taken, first, as the positive axis of x and
then, again, as the positive axis of y, we have
0M= OM' cos y - M'P sin y,
MP = OM' sin y + M'P cos y.

But, by the definition of coordinates (Ch. I, § 1),

OM=x, MP = y; OM' = x', M'P = y'.

The final result is, then, the following

X X COS y — sm y. 2/
(2)
y = X sm y -h y cos y.
To express x' and y' in terms of x and y, these equations can
readily be solved for the former variables, regarded as the un-
known quantities in the pair of simultaneous equations (2).

Or, the formulas can be deduced directly from the figure by

taking the projections in equation (1) along the positive axes
of x' and y' in turn. The result in either case is

i x' = X cos y + sill y> 2/

(3) {
\y' = — xsiny yGosy.\ -\-

Example 1 Transform the equation of the equilateral

hyperbola,
a;2 _ ?/2 _ q2^
 : :

TRANSFORMATION OF COORDINATES 221

to new axes with the same origin, the angle from the positive
axis of X to the positive axis of a?' being — 45°.

Here, y = — 45°, and formulas (2) become

Hence x'^ = ^ {x'^ + 2 a^y + y'^),

On substituting these values in the given

equation we have
2 aj'y' = a2. Fig. 5

is the equation of an equilateral hyperbola re-

This, then,
ferred to asymptotes as the coordinate axes.
its

If we had rotated the axes through + 45° instead of — 45°,

the transformed equation would have read

Example 2. Transform the equation

or
JB2 A^
to new axes, obtained by rotating the given axes about the
origin through 90°.
Here, y = 90°, and equations (2) become
x = -y' y = ^'•
Hence
B" A'

Thus it appears that the original equation represents a

hyperbola with its center at the origin, its transverse axis
lying along the axis of y cf Ch. VIII, § 8. The length of the
; .

major axis is 2 B, that of the minor axis, 2 A. The asymptotes

are given by the equations
x' _ _y[ and
B A
 :

222 ANALYTIC GEOMETRY

Eeferred to the original axes, they have the equations

^=y- and 4 = -|-

A B A B
\

EXERCISES

Obtain the equations of transformation in each of the fol-

lowing three cases

J 2. When y = - 120°. i 3. When y = 135°.

Draw a figure and deduce from it directly the formulas of

transformation in each of the following three cases. Check
the results by application of formulas (2) and (3) of the
text.

4. y = 90°. 5. y=-90°. 6. y = 180°.

7. Find the coordinates of the points (2, 0), (3, 1), (—2, 4),

(_ 5j — 8), referred to new axes obtained by rotating the old

through an angle of 45° ; of 150°.

8. Show directly by means of a suitable rotation of the

axes that the equation xy = k"^ represents an equilateral hyper-
bola referred to its asymptotes as the coordinate axes. Deter-
mine the coordinates, referred to these axes, of the vertices
and the foci.
9. The same for the equation xy = — k^.
By rotating the axes through an angle of 45° determine the
curve represented by each of the equations :

A 10. 17^2 -16a^?/ + 17?/2=: 225.

11. 3x''-10xy + ?>y' + S = 0.
12. x"^ 4- 4:xy -f 3 = 0.
?/2 -f-

13. By rotating the axes through 30° determine the curve

represented by the equation

^/3xy -y^= 12.

 : :

TRANSFORMATION OP COORDINATES 223

14. By rotating the axes through an angle y of the first

quadrant, whose sine is
f,
determine the curve represented by
the equation
52x^ - 72xy + 73y^ = 100.
15. Show that the equation of a circle whose center is at
the origin not changed by rotating the axes through any
is

angle. Actually carry through the transformation.

3. The General Case. Let it be required to pass from one

system of axes to a new one, in which both the origin and
the directions of the axes have been
changed. Let the {x, 2/)-axes, with
origin at 0, be the given system and
the (x\ ?/')-axes, with origin at 0', the
new system. Let the coordinates of
0', referred to the (x, ?/)-axes, be {xq, 2/o)>

and let the angle frojii the positive axis Fig. 6

of X to the positive axis of x' be y.
The transition from one system to the other can be made in
two steps
(a) Transform first to a system of parallel axes having the
same direction, but with origin at 0'. If the new coordinates /
are denoted by (X, Y), then ./
i"^' lx = X -{- x,
0)

(');
W^ \

)(b) Now rotate the (X, F)-axes through the angle y

J^. j X= x' cos y — y' sin y,

yY =^x' sin y -\- y' cos y.

Combining these results we get, as the final formulas, th^

" " ~
following:
\ x = x' cos y — y' sin y + Xq,
(3)
\ y = x' sin y + 7/' cos y + 2/o-j
The formulas for (^, y') in terms of {x^ y) are

x' = (X - Xq) cos y + - (7/ ?/o) sin y, !

I
^y' = -(x- Xq) sin y + (y - yo) cos y.

Y --
 :

224 ANALYTIC GEOMETRY^

These can be written in the form

I
x' = X cos y y sin y — Xq^
-\-

(4)
\y' = - X sin y + y cos y - yo,
where
^0 = ^0 cos y + ?/o
sin y,
^ ^ '

^0 = - ^0 sin y H- ?/o cos y.

It is to be noted that, inasmuch as Xq, yQ, and y are constants,

so are Xq and yQ.

Example. Identify the curve represented by the equation

(6)
^
a;2 + 6a;?/ + 2/^-100^ — 14?/ +9=
by transforming to new axes through the point (2, 1), the
angle from the old axis of x to the new being 45°.
Here Xq = 2, ?/o = 1, and y = 45°. We might substitute these
values in formulas (3) and then apply the formulas to the
given equation. It is, however, more feasible in general to
make the transformation in the two steps (1) and (2).
Formulas (1) are, in this case,
x=X+2, ?/= F+1.
Hence (6) becomes

(7) X2 + 6Xr+ F^ - 8 = 0.
Since y = 45°, formulas (2) are :

X = ^{x'- y% {x' + 2/0.

V2 V2
Then (7) becomes

1 {x'- y'Y + 3 {x'^ - y'^)

+ i(a;' + ?/0^-8 = O,
or, on simplification,

2x'^-y'^ = 4..

Consequently, equation (6) repre-

Fig. 7 sents a hyperbola with its center at
 TRANSFORMATION OF COORDINATES 225

the point (2, 1) and with its transverse axis inclined at an

angle of 45° to the axis of x.

In deducing formulas (3), we first shifted the origin and

then rotated the axes. We might equally well have proceeded
in the opposite order :

(a) Rotate the (x, ?/)-axes through the angle y into the new
axes of X and y :

/gx f a; = xeosy-\-y sin y,

\y = — X silly -\-y cos y.
The coordinates of 0', referred to
the new axes, are obtained by setting
x = Xq and = ?/ 2/o ^^ (8) ; they are, then, fig. 8
the Xq and y^ given by formulas (5).
(6) Transform from the {x, ^)-axes to the parallel axes of
x' and y', with origin at 0'. Since the coordinates of 0', re-
ferred to the {x, 2/)-axes, are (xq, y^, the equations of this trans-
formation are
\ x' —X— X(\

Eliminating x and y from (8) and (9), we obtain, as the final

formulas
\x'= X cos y + 2/ sin y — Xq,
(
2/' =— a;sin y -h ?/ cos y - ^o-
But these are precisely the formulas (4) which we had before.

EXERCISES
Obtain the equations of transformation in each of the follow-
ing cases. First find the formulas for x and y in terms of x'
and y', and then solve for x' and y' in terms of x and y. kj^± ^
[M%2/o) = (1,1); y=45°. |2. = (-2, 1);
(a^o,2/o) y=30°.
'3.
K,yo) = (0,3); y=-60.° [4. {x,,y,) = {-^, -Z)-, y=120°.

Draw a figure and deduce from it directly the formulas of

transformation for each of the following values of y, the new
 226 ANALYTIC GEOMETRY

origin in each case being at an arbitrary point (xq, y^. Check

the results by the use of formulas (3) and (4) of the text.

5. y=9r. 6. y = 180°. 7. y = - 90°.

8. Find the coordinates of the points (0, 0), (1, 2), (—3, 4),
(— 2, — 5), referred to new axes passing through the point (2, 1),
the angle from the old axis of x to the new being 45°.

9. The same, if the new origin is at the point (— 3, 4) and

the angle from the old axis of x to the new is — 30°.
Identify the curve represented by each of the following
equations by transforming to parallel axes at the point {x^, y^
specified, and then rotating the new axes through the given
angle v. Draw a figure in each case. ^^
'

'lC,rV\
(XAl^ " '^'>„
[A Equation '
2/,) y
"'"

A^^^ + 5?/2-4ic-4?/-4 = 0,
5a;2_6.T?/ (1, 1), 45tro
A- 11. a;^-4a;// + 2/^4-10a;-22/ + 7= 0, (1,3), -45o
12. a;^-10a;2/ + + 46ic + 102/-47 = 0, 2/'^
(2,5), 135°.

13. Find the curve represented by the equation

66a;'2 - 24a:?/ -f 59 ?/2 -f- 108 a; + 94 +2/ 76 =
by introducing (—1, — 1) and then
parallel axes at the point
rotating these axes through the acute angle whose tangent is |.
Draw a graph.

^^ 14. The equation

/ 7 a;2 - 18 a;?/ - 17 2/2 - 28 + 36
a: 2/
4- 8 =
represents a conic whose center is at the point (2, 0), and one
of whose axes has the slope — \. Identify the conic, and
draw a rough graph of it. ,

V /
4. Determination of the Transformation from the Equations
of the New Axes. Consider the general transformation given
ri by formulas (4) of the preceding paragraph. If, in these for-
mulas, we set a;'= and 2/' = 0, we obtain

/ ajcosy + 2/siny-iCo=0,
^ ^ — X sin y -\-y cos y — ^0 = 0.

^^^
 TRANSFORMATION OF COORDINATES 227

These are the equations of the new axes, referred to the old
system the first is the equation of the axis of y\ the second
;

that of the axis of x'.

Conversely, we
set the expressions on the left-hand sides
if

of equations (1) equal to x' and y' respectively, we obtain

the
equations of the transformation.

Problem. Let it be required to find the equations of a trans-

formation which introduces the two perpendicular lines,

(2\
a: -f 22/ -f 1 = 0, ^

.
2a;- ^-3 = 0,
as coordinate axes.
The natural procedure, in order to obtain the required equa-
tions, would be to set the left-hand sides of equations (2) equal
to x' and y\ But, in order to obtain the correct result in this
way, we must first put the left-hand sides of equations
(2) into
the form of those of equations (1).
These latter are of the form

(3)
(^^ + by — ^Q,
-bx-h ay - ^0,
where
(4) a2 + 62 = l.
The left-hand sides of equations (2) will be of the form
(3) if we multiply the second of the equations through
by-1:
x-h2y-\-l = 0,
-2x+ 2/ + 3 = 0.
To bring about the fulfillment of condition (4), we multiply
each of these equations through by a constant p=^0:

(2a)
px + 2py+ p = 0,
-2px-\- py-\-3p = 0.
Thereby we have not changed tli,e lines which the original
equations (2) represent.
228 ANALYTIC GEOMETRY

The value of p is to be determined so that condition (4) is sat-

by the left-hand sides of equations (2 a), that is, so that
isfied

W'+(2p)2=l or 5p2 = l.
Hence p = ± 1/ V5.
We choose p =
1/V5. If this value is substituted for p in
equations (2 a), these equations will be precisely of the form (1).
Hence the equations of a transformation
introducing the lines (2) as the axes are

X ,
_x±2y±l
=
(5)
V5
2/' =—2 a; -h y +3
V5
FiQ, 9 The the lines (2) is the axis of
first of y' ;

the second, the axis of x'.

The new origin is at the point (1, — 1); for, this is the point
of intersection of the lines (2). The old origin (x, y) = (0, 0)

has, according to (5), the coordinates ( , \ referred to

the new axes, and must lie, then, in the Jl7'st quadrant formed
by these axes. Consequently, the new axes must be directed
as shown in the figure.
The slope of the axis of x', the second of the lines (2), is 2
and so its slope angle is 63° 26', or 243° 26'. It is clear from
the figure that it is the first of these angles which is the angle y.
We obtain a second transformation, for which the lines (2)
are the new axes, by taking the value — 1/ V5 for p. For this
transformation the directions of both axes are opposite to those
for (5), and y has the value 243° 26'.
For both transformations the first of the lines (2) is the axis
of 2/' ; the second, the axis of x'. By reversing the roles of the
lines, two more transformations can be obtained. Thus, there
are in all four transformations introducing a given pair of
mutually perpendicular lines as coordinate axes.
 TRANSFORMATION OF COORDINATES 229

EXERCISES
In each of the following exercises find the equations of a
transformation which introduces the given perpendicular lines
as new axes. Find the position of the new origin and the value
of y draw an accurate figure, and indicate the directions of
;

the new axes.

Axis ofy' Axis ofx'

1. 3a; + 42/-ll = 0; 4« - 3?/ + 2 = 0.
2. a;+?/-3 = 0; x-y-l = 0.
3. 2a;-32/ = 0. 3x -f 2^/ + 5 = 0.
4. 6x-2y=0; 2x-\-by= 0.

5. a.-- 2 = 0; = 0.
2/4-3
6. 2/-8 = 0; x-b=0.
7. If the lines of Ex. 1 are introduced as axes, what does the
equation
(4x - 3?/ + 2)2 = 3cc -f 42/ - 11
become ? What curve does it represent ? Draw a rough graph.
By a suitable transformation of axes determine the nature
and position of each of the following curves. In each case
draw a figure showing accurately the new axes, properly
directed; then sketch the curve. -x^^ «•'
"^^ O
8. + 42/ +
(2a;-32/)2 4-6£c 10 = 0. -,-

9. + yy-2x^2y =
{x (}.
k^
^
3 (oa;-2.v)^+(2x + 5^)2 = 1.
10.

11. {_x^y-iy-6{x-y-l)=0. ~\ 'V

12. Find the equations of all four transformations which
introduce the lines

5aj-122/ + 7 = 0, 12a7 + 5?/ - 17 =

as axes of coordinates. Draw the four corresponding figures,
and find the four values of y.
230 ANALYTIC GEOMETRY

13. Obtain the equations required in Ex. 1 by finding the

coordinates of the new origin and a value for y and then ap-
plying formulas (3) of § 3.

5. Reversal of One Axis. There is one more case which

deserves mention. Suppose that the sense of one axis is re-
versed, while the other axis remains un-
V y
changed. Let the axis which is reversed
be the axis of x (Fig. 10). Then, evidently,
X-
O M \ = — x',
x \ x' = — X,
Fig. 10 (1)
\y = y'\ \y' = y'
If the sense of the axis of y had been reversed, the axis of
X remaining unchanged, we should have had :

\ x=. x', j x' = X,

(2)
\y = -y'\ [y' = -y'
Consider, for example, the equation of a parabola in the
normal form,
]f-
= 2 mx.

If the sense of the axis of x is reversed, the equation becomes

2/'2 = — 2 mx'.
We could use this result to interpret the equation,

2/^ = — 2 mix^
if we knew the parabola only in its normal form. Taking the
axis of x' opposite to the axis of and starting with the
it*,

known parabola
y
y''^ = 2 mx',
we see that the transformed equation, X
o
y- = — 2 mx,
represents a parabola on the negative axis of x,

its vertex being at the origin. Fig. 11

 TRANSFORMATION OF COORDINATES 231

EXERCISES
1. Assuming that the equation of the parabola in the form
a;2 = 2 my
is known, interpret the equation
x^ = —2 my
by the method of the text.

2. Plot the so-called semi-cubical parabola

From the graph determine the curve defined by the equation

2/2 =— aj3^

EXERCISES ON CHAPTER XI

Change of Origin
In each of the following exercises prove, by making a suit-

able transformation to parallel axes, that the given equation

represents two straight lines. Find the equations of the lines,
referred to the original axes.

1. a;2_2/2_4a._^62/ — 5 = 0.
2. + 182/-5 = 0.
4a;2-9y^-f-8cc
3. 4tx'-16y^-12x-i-9 = 0.
What does each of the following equations represent ?
4. a;2-f 2?/2-10a;-f 12?/ + 43 = 0.
Ans. The point (5, — 3).
5. 30^2 _|_ 4^/2 _6.T-j- 16?/ + 21 = 0.
6. By completing the cube for the terms in x in the equa-
tion
y = x^ + 3x''- + 3x-2,
and by making the transformation to parallel axes which is
suggested by the result, determine the curve defined by the
equation. Draw the curve roughly.
232 ANALYTIC GEOMETRY

By the method of Ex. 6, identify and plot the locus of each

of the following equations.
7. 6?/ = a^4-6cc2 + 12aj4- 14.
8. a; = 2?/3-6?/2 + 6?/ + 6.
9. a^-Sx'^-6x-{-xy^-y^~-^S=0.

10. Determine the position of the point (xq, yo) such that, if

the straight lines

Sx-4:y-2 = 0, x + 2y = 4:

are referred to parallel axes at {xq, yo), their equations will

contain no constant terms. What will these equations be ?

11. Show that, if the two intersecting straight lines

Aix + B^y + Ci = 0, Aox -h ^.^Z + C'2 =

are* referred to parallel axes at their point of intersection, the
equations of the lines become

Aix' + Biy' = 0, A2X' + B.jj' = 0.

12. Determine the position of the point (xq, yo) such that, if

the curve
xy — 2x — y — 2 =
is referred to parallel axes at (xq, 2/0)5 its equation will contain
no linear terms in x and y. Identify and plot the curve.
13. Identify and plot roughly the locus of the equation

6xy -\-7x — 5y + 3 = 0.
14. Determine the point {xq, y^) such that, if the curve

3a;2 - 7a^?/ - 62/2 - 19a; +2?/ + 20 =

is referred to parallel axes at {x^, 2/0)5 its equation will contain
no linear terms in x and y. Show that the equation will also
contain no constant term and hence that it will represent two
straight lines. Find the equations of these lines with respect
to the original axes.
 TRANSFORMATION OF COORDINATES 233

Rotation of Axes
15. Given the two perpendicular lines through the origin
of slopes i and — ^. Find the equations of a transformation
introducing these lines as axes.
16. The equation 2x'^ -\-Sxy — 2y'^ = represents two per-
pendicular lines through the origin. Show that it may be
transformed into the equation x'y' == by a suitable rotation
of axes.

Identify and plot roughly the curve defined by each of the

following equations. Cf. Ex. 16.

17. 2x'-{-Sxy-2y^ = 6. 18. 12x''' - 7 xy -12y^=25.

19. Determine the equations of a rotation of axes where-
by the axis of y comes into coincidence with the line
4a; + 3?/ = 0.
20. Identify and plot roughly the curve defined by the
equation (4 ic +
SyY =
125 x. Cf Ex. 19.
.

The same for each of the following equations.

21. (x + yy = W2y.
22. 4:X^-\-4:xy -\-y^ = 5V5x.

23. 16x^--j-24:xy -\-9y^ = 60x-S0y.

24. The line 2x — y — is sm axis of the conic

6x'^-4.xy-\-Sy^ = 6.
By a suitable rotation of axes determine the nature and posi-
tion of the conic.

Show that, by a suitable rotation of axes, each of the follow-

ing equations becomes linear in y and hence capable of solu-
tion for y, without radicals.

25. x'^-y^ + 2x-S = 0.

26. 4.x''^—4cXy-\-y^-\-Sx-y = 2.
27. 2x'''-]-3xy-2y'''-\-6x-2y=i8.
234 ANALYTIC GEOMETRY

General Transformation of Axes

28. Prove that the straight lines

when referred to suitable axes, will have equations the first

of wliich is x' = 0, while the second contains no constant term.
29. Find the equations of the circle

a;2 _|. 2/2 + 605 — 82/ + 6 =

and the line 5x-\-12y — 13 = 0,
when they are referred to axes through the center of the circle

parallel and perpendicular, respectively, to the line.

30. Find the equations of the two circles

x^ + y^ -\- 4:X — 6 y — = 0,4.

x'i -^ y2 - ex -iy -
-\- = 0,4.

when they are referred to the point midway between their

points of intersection as origin and the line joining the points
of intersection as axis of x'.

31. What will the equations of the circles of Ex. 30 become,

if they are referred to the mid-point between their centers as
origin and the line of the centers as axis of y' ?

32. A transformation consists of a change of origin to the

point of a rotation of the new axes through the angle
(xq, 2/0),

y, and of a reversal of the sense of the axis of x thus obtained.

Show that the equations of the transformations are

{ x = — x' cos y — y' sin y + ^o>

1 2/ = — sin y + cos y +
a;' 2/' 2/0-

i
 :

CHAPTER XII

THE GENERAL EQUATION OF THE SECOND DEGREE

1. Change of Origin of Coordinates. The aim of this chapter
is twofold : To determine what curves are represented by-
equations of the second degree in x and y and to develop ;

methods by means of which the curve represented by any

particular equation may be easily identified and its size and
position accurately described. The methods used consist
primarily in transformations of coordinates. We begin, then,
by investigating what can be accomplished by a change of
origin, i.e. a transformation to parallel axes.

Example be required to identify and to describe

1. Let it

accurately the curve represented by the equation

(1) 5 a;2 _4 2/- - 20 - 24a; ?/ -f- 4 = 0.

Completing the square of the terms in x and then of the
terms in y, according to the method of Ch. XI, § 1, we obtain
5(x _ 2)2 - 4(2/ + 3)- = - 20.
On setting

x' = x-2, y' = y-\-S,

that is, on changing the origin of coordi-
nates to the point (2, — 3), this equation
becomes
5a;'2_4y2 = _20,
," /'2

or
4 5 Fig. 1

Consequently, equation (1) represents a hyperbola with its

center at the point (2, — 3) and with its transverse axis par-
235
236 ANALYTIC GEOMETRY

allel to the axis of y. The coordinates of the foci, referred

to the new axes, are (0, ± 3) ; consequently, when referred
to the original axes, they are (2, 0) and (2, — 6). The equa-
tions of the asymptotes, with respect to the new axes, are
V5a;'-22/' = 0, V5ic' + 2y = 0;
hence they are, with respect to the original axes,

V5a;-2?/-2V5-6 = 0, ^ V5aj + 2?/ - 2V5 + 6 = 0.

The semi-axis major is V5, and the semi-axis minor, 2 ; the
eccentricity has the value f V5.

Example 2. Consider the equation

(2) Zxy - 6> -f 3?/ - 10 = 0.

We rewrite this equation, first, in the form
^{xy-2x + y )=10,
and then as Z[x{y — 2)-\- (y )] = 10.
If —2 is added to the y in the second parenthesis and, in
equalization, 3 • 1 •
( — 2) or —6 is added to the right-hand
side, this equation becomes
tf'

3(cc+l)(^-2) = 4.

We now change the origin to the point

X' by
O' (—1, 2) setting

x' = x-\-l, y' = y — 2.

The equation thus becomes
O
Fig. 2 3 x'y' = 4.
Accordingly, (2) represents a rectangular hyperbola with the
lines a; -f 1 =
and ?/ 2 = as asymptotes. —
Example 3. The equation,
y V
(3) x'^ + 2x-2y-l = 0,
can, according to the method of Ch. XI, § 1, o
be put into the form o'
(x + iy-2{y + l) = 0, Fig. 3
 : :

EQUATION OF THE SECOND DEGREE 237

and hence represents a parabola with vertex at the point

(—1, — 1) and with axis parallel to the axis of y.
Example 4. Consider, now, an equation in which all three
quadratic terms are present

(4) 6 a.'2 - a-?/ -2 2/2 + 4a; -f- 9 ?/ - 10 = 0.

In this case, completing the squares of the terms 6a;2-j-4x
and of the terms —2y'^-\-9y does not help. Let u§ make an
arbitrary change of origin, sej:ting

(5) x = x'-\-xo, y = y'-\-yo,

and aim to determine the new origin, (xq, y^), so that in the re-
sulting equation the linear terms in x'
and ?/' do not appear.
Setting in (4) the values of x and y as given by (5) and
collecting terms, we have

(6) 6a;'2-a;'?/'-22/'2

+ (12 a'o - 2/0 4- 4)x' +(_ cc^ - 4^0 -f 9)/ + F' = 0,

where
(7) F' = Q V - x,y, _ 2 2/0^ +4 a^o + 9 - 10.
2/0

If the terms in x' and y' are to drop from this equation, Xq and
2/o must be so chosen that

/gx 12a;o- 2/0 + 4 = 0,

^^
-0^0-42/0 + 9 = 0.
Solving (8) simultaneously for x^, 2/0, we have
^0 = ~" T' 2/0 = 'i~'

The value of F', for these values of a^o and 2/0, is 0. Conse-
quently, we have shown that equation (4), when referred to a
new origin at (— 4, 2f), becomes
6x'^-aj'2/'-22/'2 = 0.
The left-hand side of this equation can be factored

(9)
•
(3a;'-22/0(2aj'-f2/')=O.

Thus (4) represents two straight lines through the point

(— 1, 2|-) with slopes | and — 2.
238 ANALYTIC GEOMETRY

If in (9) we set

x' = x-\-\, y' = y- V-,

i.e. if we transform back to the original axes, we obtain, finally,

(3x-2y + 5X2x-\-y-2)=0.
This equation is seen to be precisely equation (4), with the
left-hand side factored into two linear factors. The two
straight lines represented by (4) have, then, the equations

Sx-2y + 5 = 0, 2x-\-y-2 = 0.
It should be noted that the constant term (7) of equation (6)
is the value of the left-hand side of (4) for x = Xq, y = y^. In
this example, this constant term took on the value zero when
Xq, 2/o
were chosen so that the coefficients of the linear terms in
x' and y' vanished. This does not, however, occur in general,
as we shall see in the next paragraph.

EXERCISES
In each of the following exercises identify and plot roughly
the curve represented by the given equation. If the curve is

a conic section, find, when they exist, the coordinates of the

center and the foci, the equations of the directrices and the ^

asymptotes, and the value of the eccentricitv.

T^4.x'^ + 9y^-16x-lSy-ll = 0.:y

%^. 18a;2 -h 122/2 - 12 12?/ - 19 = 0.
a; -f-

3. 4 ^2 -h 3 + 16 — 6
2/^ 31 = 0.
a; 2/ -I- Ans. No locus.

^25 -
aj2 4 -f 50
2/2- 8 - 79 = 0.
a^ 2/

^^a;2_9^2_^20xH- 122/ + 33 = 0.
6. 7a;2-52/- + 2a;-42/-l = 0.

7. 2/--8aj + 62/-h49 = 0.

8. 3a;2_6aj-52/-|-3 = 0.

9. 22/2 + 4a; + 3?/- 8 = 0. ^ \^

lO^^xy + 2a; - 3?/ — 11 = 0. l
11. 5xy — 5x-\-y i-l = 0.
 EQUATION OF THE SECOND DEGREE 239
12. Sxy-\-x~18y-6 = 0.
13. 2x'''-\-5xy-3y''' i-Sx-^16y-5 = 0.

14. x''--\-4.xy-\-3y'-2x-2y=0.
15. Sx'''-xy-^5y^-6x-\-y-\-3 = 0. Ans, The point (1, 0).

16. Prove that every equation of the form

y = Ax' + Bx-{-C, A=^0,

or 01 the form
x=.Ay^+By-hC, A=^0,
represents a parabola with its axis parallel to an axis of coor-
dinates.

17. Show that every equation of the form

bxy + dx + ey +/= 0, b^^O,

represents either a rectangular hyperbola with its asymptotes
parallel to the axes, or two perpendicular straight lines parallel
to the axes. Prove that the latter case occurs if and only if
bf= de.

18. Given the equation

aa;2 -I- C2/2 + dx 4- e?/ -|-/ = 0,
where neither a nor c is : ac^ 0.
(a) If ac >
prove that the equation represents an
0, ellipse,
or a point, or that it has no locus.
(6) If ac < 0, show that the equation represents a hyperbola,
or a pair of intersecting straight lines.

2. Rotation of Axes. Example 1. Let it be required to

identify the curve defined by the equation

(1) 5a;2_6a:7/-h5?/2-8 = 0.
We transform (1) by a rotation of the {x, 2/)-axes through
an arbitrary angle y into the {x', ?/')-axes. For x and y in (1)
we set, then, according to Ch. XI, § 2,

/Qx ' ^= ^' cos y — y' sin y,

y = a;' sin y -\- y' cos
y.
 ;

240 ANALYTIC GEOMETRY

and obtain, after collecting terms and simplifying,

(5 — 6 cos y sin y)x'2 — 6 x'y\Gos^ y — sin2 y) -|- (5 -|- 6 sin y cos y)y'^
- =
8 0,

or,on replacing the trigonometric functions of y by functions

0f2y,
(3) (5-3 sin2 y)a.''2 - 6xy cos 2y +(5 + 3 sin2y)2/'2 -8 = 0.
We now choose y so that the coefficient of x'y' will become :

cos 2 y = 0.
Values of 2 y satisfying this equation are 90°, 270°, 450°, 630°
the corresponding values of y are 45°, 135°, 225°, 315°. We
choose, arbitrarily, the smallest of these values, namely, y=45°.
Equation (3) thus becomes
2a;'2-f 82/'2-8 = 0,
or
n
J2
(4) — + ^=1.
Consequently, equation (1) represents an ellipse with its

center at the origin and with the transverse axis inclined at an

angle of 45° to the axis of x.

Example 2. Consider the equation

(5) 5 ic^ — 6 a;?/ +5 2/^ — 4 — 4 — 4 = 0.

.'c ?/

We proceed, as in § 1, Example 4, transforming (5) to arbi-

trary parallel axes, x^ y, and then choosing the new origin,

(a.'o, y^, so that in the equation resulting

X'
from (5) the linear terms in x and y drop
out. We findnew origin must be
that the
_ at the point (1, 1),and that the resulting
equation then becomes
" -^
(6) 5^2 _ 6x^-1- 5^2 _ 8 = 0,
Fig. 4
where the constant term, — 8, is found as
the value of the left-hand side of (1) for
a; = 1, = 1
?/ ; cf. end of § 1. ^

Now (6) is the same equation in x, y as (1) is in x, y.

 EQUATION OF THE SECOND DEGREE 241

Hence, follows that (5) represents an ellipse with its center

it

at the point (1, 1)and with the transverse axis inclined at an

angle of 45° to the axis of x.
The procedure, then, for any equation similar in form to (5)
consists first in transforming to parallel axes so that the linear
terms in x and y drop out, and then in rotating the new axes
so that the quadratic term in x, y drops out. We shall show
later that this procedure is always valid except in one case.

EXERCISES
Identify the curves represented by the following equations.
Draw a graph in each case, showing the original and the new
axes an^ the curve.
_1^5ic2 -^4»jr+ - 32 = 0.
5?/2

2. 5ic2 + +
26aj?/ = 0.
5?/2-72
(^,,; 7a;2 + 2a??/ + 7?/2 4- 2 = 0.

4. 5 a;24- 2^3xy + 7 - 16 = 0.
?/2

5. 2a;2-f4V3a;?/- 22/2 -16 = 0.

tiZr3aj2 _ 23^2/ + 3/ - 4a; - 42/ = 0.

7. x^ + Qxy + - lOx — 142/ + 14 = 0.

2/^

8. A.x' + 16a;2/ + ^2/2 - 4ic - 82/ + 13 = 0.

9. Show that, iih ^±2a, the equation

ax'^ + hxy + ay'^+f=0, bf^O,

represents an ellipse, or a hyperbola, with its center at the
origin and with its axes bisecting the angles between the
coordinate axes.

3. Continuation. General Case. We propose to develop and

simplify the method of § 2, Example 1, for the removal of the
term in xy. Take the equation
(1) Ax^ + Bxy -\-Cy^-j- F' = 0, B^ 0,

and rotate the axes through the arbitrary angle y, by means of

formulas (2), § 2. The resulting equation can be written as
 : :

242 ANALYTIC GEOMETRY

(2) ax"' -h hx'y' + c?/'^ J^F' = 0,

where
a=^ A cos2 y 4- 5 sin y cos y + C sin^ y,
(3) 6 = -(^- 0)sin2y + 5cos2y,

c=i A sin2 y — B sin y cos y + O cos^ y.

Since y is to be chosen so that 6 = 0,

(4) -(^- C)sin2y + 5cos2y = 0,

(5) cot2y = ^^'.
Of the values of 2y which satisfy this equation, we choose
arbitrarily that one which lies between 0° and 180°. Then y
is a positive acute angle.
If the axes are rotated through this angle y, (2) becomes

(6) ax'^ + c2/'2 + i^' = 0.

The values of a and c to be determined.
are still There is
a simpler way of doing than substituting the value found
this
for y in the formulas for a and c, as given by the first and last
equations of (3). First, add these two equations ; the result is

(7) .a + c = A+C.
Thus we have one very simple equation for the two unknown
quantities a and c.
Next, subtract the second of the two equations from the
first

(8) a - c = (^ - 0) cos 2y + 5 sin 2 y.

Square both sides of (8) and both sides of the second equation
of (3)
b = -(A— C) sin2yH- 5 cos 2 y,
and add the equations thus obtained ; the final result is

(0) (a - c)2 + b^ = (A- (7)2 + m

But y was chosen so that 6 = 0. Consequently, (9) becomes

(a-cy = (A- Cy + B",
 EQUATION OF THE SECOND DEGREE 243

or
(10) a-c = ± V{A - Cy -h B".
Thus we have a second simple equation for the two unknowns
a and c.

From equations (7) and (10) the values of a and c are easily
found in terms of the known coefficients, A, B, and C, of (1).
There are, however, two values of each, due to the double
sign before the radical in (10). Which values should we
take?
If in (8) we substitute for A—C its value as given by (4),
we obtain

a— c= B —
COS'^ 2 YT,
^
,

-f- B sm 2o y,
T3 .

sin2y
or

a —c= B
sm 2
But 2 y lies, by choice, between 0° and 180° and, consequently,
sin 2 y is positive. It follows that a — c must have the same
sign as B.
Accordingly, if we rewrite (10) as

the plus sign must be chosen. Hence, always,

(11) a-c=Byl(^^^+l.
From equations (7) and (11) unique values for a and c can
now be found.
Example. Consider the equation

(12) Ix" -Sxy -\-y'' + l^x-Sy-2 = 0.

By shifting the origin properly, to the point (— 1, 0), (12)
becomes
(13) 7 aj'2 - ^x'y' + y'^ - 9 = 0.
244 ANALYTIC GEOMETRY

Next, rotate the new axes through the positive acute angle
given by formula (5), which in this case is

cot 2 ^
7-1 ^ 3
-8 4'

so that y has the value 63° 26'. Thus (13) becomes

(14) '
ax'"^ + c/'2 _ 9 = 0.

The values of a and c are determined from equations (7)

and (11), which are, here,
a -f c = 8,

«-^=-Wf-; + 1 = _ 10.
\-\
Then the values for a and a= c are : — 1, c = 9. . Conse-
quently, (14) becomes

or
X//2 y
m
(15) -1.
9

Equation (15) represents a hyperbola'

with its transverse axis along the axis of
y". Hence (12) represents a hyperbola
Fig. 5
with its center at (~ 1, 0) and with its

transverse axis inclined at an angle of 63° 26' + 90° = 153° 26'

with the axis of x.

The Expression B^ — 4^ AC. If from (9) we subtract the

square of (7), we obtain

(16) = B'-4.ACj
b^-4:ac
or, since we chose y = 0,
so that 6

(17) -4ac = S2-4^a

TJie Case B" - A AC > 0. If B - AC is 4: positive, ac is, by
(17), negative ; hence a and c have opposite signs. Thus, if
F' ^ 0, (6) represents a hyperbola. If F' = 0, (6) becomes
(18) aa;'2 + c/2 = 0.
 EQUATION OF THE SECOND DEGREE 245

Since a and c have opi^osite signs, the left-hand side of (18)

can be written as the difference of two squares and then fac-
tored. Therefore, (18) represents two straight lines (Ch. IX,
§ which intersect at the origin.
4)
These results for (6) are true for the original equation (1).
They hold not only if B^O, —
the case which we have been
treating, — but also if B = 0. For, if 5 = 0, (1) is itself in
the form (6), and hence may be considered directly. We have,
then, the following theorem.

, _ Theorem 1. — 4 AC > 0, the equation

JVhen 5^
Ax^ + Bxy -irCf' + F' =

represents a hyperbola^ if F' =^0; ifF' = 0, represents it two

intersecting straight lines.

The Case B^ — 4:AC<0.

In this case, according to (17),
ac is and a and c have the same signs. Then if
positive,
F^ ^ 0, (6) represents an ellipse, or, in the case that a, c, and F'
are all of the same sign, has no locus. If F^ = 0, (6) reduces
to (18). But now the left-hand side of (18) can be written
as the sum of two squares, since a and c have the same sign.
Hence it is satisfied only hy x = 0, y = 0. It represents, then,
a single point, or, as we may say, a null ellip)se.*
Not merely a and c have the same signs in this case, but
also A
and C. For, if A
and C have not the same signs, the
product AC < or = consequently, 5^ — 4 AC > 0,
; a con- —
tradiction. It follows, further, from (7), that A and C have
the same signs as a and c.

We can now characterize more fully the two cases which

arise when F' ^ 0. We have seen that equation (6) has no
locus, if i^' is of the same sign as a and c, or, as we can now
say, if F' is of the same sign as A and C, i.e. if AF' (or
CF') > 0. On the other hand, (6) represents an ellipse, if F' is

opposite in sign to A and 0, i.e. if AF' (or CF) < 0.

We summarize our results in the form of a theorem.

* Cf. null circle, Ch. IV, § 2.
246 ANALYTIC GEOMETRY

Theorem 2. When B^ — 4:AC< 0, the equation

Ax"^ + Bxy -f- C^2 j^F' = 0,

*

(/" F' ^ 0, represents an ellipse or has no locus, according as AF'

(or OF) is negative or positive; if F' = 0, the equation repre-
sents a single 2yoint.

The Case B^-4.AC=0. If 5^-4^(7=0, there is no

need of rotating the axes. Consider, for example, the equation

(19) dx"' -6x7j-\-y^-4: = 0,

for which jB2 — 4^(7 = 36 — 4-9 = 0. This equation can be
written in the form
(Sx-yy-A = 0,
or (Sx-y-2){3x-y + 2)=0,
and hence represents two parallel lines of slope 3.

EXERCISES
Identify the curves represented by the following equations.
Draw a graph in each case, showing the original and the new
axes and the curve.

1. 2x^-\-4:xy + 5y^-{'4:X-\-16y + 2=0.

2. 3x2 + + 81/2 + 6a; + 16?/ + 38 = 0.
12.T2/

3. 7Sx'^-^72xy + 52y^-\-74:X -S2y- 4.7 = 0.

4. 2x2 + 3a;2/-2?/2- 16^-122/ + 22 = 0.
5. x^-5xy-\-13y'^-3x-\-21y = 0.
6. 15 xy - 8/- + 450?/ - 450 = 0. .

7. 20x2_l6x?/ + 8?/2 + 52a;-40?/ + 5 = 0.

8. 8x2 + 8x?/-7?/2 + 36?/ + 36 = 0.
9. 7x2-3x?/ + 3?/2 + 5x + 15?/ + 35 = 0. Ans. No locus.

10. 12x2-20x?/-36?/2-22x-26?/-9 = 0.
11. 3x2+2x?/ + 2?/2 + 10x = 0.
12. x'^-\-3xy-y2 + 2x-10y = 0.
 : :

EQUATION OF THE SECOND DEGREE 247

4. The Genei;al Equation, B'^ — ^AC^^O. We consider liere

the general equation of the second degree

(1) Ax'^ + Bxy -h Cy"- + Dx + Ey + F = 0,

assuming that B^ — AC ^ 0. From the results of the
4: preced-
ing paragraph, we should expect that, in general, (1) represents
an ellipse or has no locus, if B^ -4:AC<0, and represents
a hyperbola, if B'^ — ^AC>0. Accordingly, we shall call
(1) an equation of elliptic type or of hyperbolic type, according as
^ — 4 ^(7 is negative or positive.*
To remove the terms in x and y from (1), we set
X = X' +Xoy y = y' -^y^
in (1), obtaining

(2) Ax'' 4- Bxy + Cy'^ + (2 Ax, + By, + D)x'

+ {Bx,-^2Cyo+E)y'+F' = 0,
where
(3) F' = Ax,^ -j- Bx^, + Cy,^ + Dx, -^Ey^ + F
is the value of the left-hand side of (1), formed for x = Xq,
Setting the coefficients of x' and y' in (2) equal to zero

(4)
^Ax,+ By,+ D = 0,
Bxo + 2Cyo + E = 0,
and solving these equations simultaneously for Xq and i/o, we
have
(5) X
:^?_CDj-BE ^ 2AE-BD '

'
B'-4.AC' ^' B'-^AC
Since has been assumed that the denominator, B'^ — 4. AC,
it

of these fractions is not 0, it is always possible to solve equa-

tions (4), and the solution (5) is unique.

If the new origin {xq, y,) is taken at the point (5), equation
(2) becomes
(6) Ax'-^ + Bx'y' -f Cy"^ -\-F' = 0.

* If 52 —
4 J. C 1= 0, We shall say that (1) is of parabolic type. This
case will be treated in the next paragraph.
 : : :

248 ANALYTIC GEOMETRY

Equation (6) is exactly tlie equation treated, in § 3. There-

fore the theorems of § 3 are valid for it and, consequently, for
the original equation (1).

The value of F' as given by (3) can be put in a more con-

venient form. Multiply the first of the equations (4) by Xq^

the second by y^, and add

2 Ax,^ + 2 Bx,y, + 2 Cy^ + Dx, + Ey, = 0.
Multiply this equation by — i and add it to (3)

Finally, substitute the values of Xq and ^o ^s given by (5).

The result is

-^, ^ 4A0F- B^F-AE^ - CD^ + BDE

m-4.AC
The numerator of the fraction is known as the discriminant
of equation (1) and is denoted by A
(7) A= 4 ACF - B^F - AE^ - CD^ + BBE.
In terms of A, F' has the value

=- A
(8) F'
B^-4.AC
It is clear that if F' = 0, then A = 0, and conversely. In
stating the theorems of § 3 for equation (1) above, we can,
F' ^0 and i^' = by A :#= and A = respec-
therefore, replace
tively. Furthermore, in case B^ — 4: AG is negative and F'
and A are not 0, A has the same sign as F\ In this case,
then, AF' (or CF) is positive or negative, according as AA
(or CA) is positive or negative.
We now restate, for equation (1), the theorems of § 3.

Theorem 3. An equation (1) of hyperbolic type :

JB2-4ac>0,
represents a hyperbola, if A=^0. If A= 0, it represents two
intersecting straight liiies.
 EQUATION OF THE SECOND DEGREE 249

Theorem 4. An equation (1) of elliptic tyjje :

i/A^O, represents an ellipse or has no locus, according as A^

(or CA) is negative or positive; i/A = 0, the equation represeyits
a single point.

If an equation of the form (1) is given, and 5^ — 4 AC =f=^ 0,

the type of curve which the equation represents can be de-
termined by finding the sign of B^ — A, and by ascertaining AC
whether or not A = 0. Further investigation is necessary
only in case ^2 _ 4 ^q ^ q ^nd A :^ the sign of AA (or CA) ;

must then be determined.

For example, the equation
x-2 — Sxy -\- 2 y'^ -{- X — 5y + S =
represents a hyperbola, inasmuch as B^—4:AC=9—4: 2=1 >0, •

and A = — 15 ^ 0.
To and size of an ellipse or a hyperbola
find the position
definedby an equation of the form (1), it is necessary to carry
through in detail the work of changing the origin and rotating
the axes. If, however, A= 0, it is sufficient merely to make
the proper change of origin. The equation then takes on the
form (6), where F' = 0. In the
elliptic case, it represents a
single point, the new
In the hyperbolic case, it can
origin.
be factored into two linear equations, which determine the
two lines typical of this case.

EXERCISES
Determine the nature of the. curve defined by each of the
following equations. In case the equation represents two
straight lines or a single point, find the equations of the lines,
or the coordinates of the point, referred to the (x, ?/)-axes.

1. 4.x''— 5xy-\-y'^-{-llx-Sy = 0.
2. 3a;2_4iC2/ + 2?/2-2a; = 0.
3. 3iB2 + 2a;?/ + 2/2-8a;-4?/ + 6 = 0.

Ans. The point (1, 1).

250 ANALYTIC GEOMETRY

2x — y — S = 0,
Ans. The lines + 2?/ — 4 = 0. a:

5. x^ -\- xy -\- = 0.
y"'
-\- 3y Ans. No locus.
-\- 4^

6. Sx'^-xy-2y'^-5x-2y — 56 = 0.
7. 2x'^-xy-^y'^ — 7y-{-10 = 0.
8. 4.x''--3xy-{-9y^--\-lTx-12y-^19 = 0.
9. 10a;2-9icy-9?/24-14a; + 21?/-12 = 0.
10. 4:X^ — 2xy + y^ — 4:X-\-y-l-5 = 0.

11. 2x'^ — 3xy-\-y'^ — 6x-{-5y + = 0. 4:

12. Prove that the general equation is of hyperbolic type,

if AC< 0, i.e. if A and G are of opposite signs.

13. The same, if 5^ and AC = 0.

5. The General Equation, B^ — 4^AC=0. First Method. If
B^— 4:AC has the value 0, the equation

(1) Ax"^ 4- Bxy + Cy''--\-Dx + Ey + F=0

is said to be of parabolic type. The method used in the case
52 _ 4 ^(7:^0, which begins with shifting the origin so that
the linear terms in x and y drop out, is inapplicable here, since
equations (4) of § 4, for the determination of the new origin,
have in general no solution if 5^ — 4:AC = 0.*
Let us begin, not with a change of origin, but with a rota-
tion of axes, assuming that B =^0. Applying to (1) the trans-
formation (2) of § 2, we obtain

(2) ax'^ + bx'y' + cy*^ -{- dx' + ey' + F = 0,

where a, b, c are as given by formulas (3) of § 3, and

,ox d= D cos y + sin y, -EJ

e = — D sin y-\-E cos y.

* They have no solution if the lines

2Ax-\-By+D = 0, Bx + 2 Cy -\- E=
are parallel ; infinitely many solutions, if these lines are identical.
 EQUATION OF THE SECOND DEGREE 251

Since formulas (3) of § 3 are valid, so are the equations which

were deduced from them ; in particular,

(4) + c = A-\- C,
•
a
(5) = B'-4.AC.
b''-4:ac

Here, B^ — A AC = 0, and hence — 4:ac=0. It follows, b"^

then, that we can make 6 = by choosing y so that 6 = 0, or

by choosing y so that either a = or c = 0. The second of
these two methods is in the end the simpler. We will follow
it and, in particular, choose to make a = 0.

If a = 0, we have, by the first of the formulas (3) of § 3,

^cos-y + ^siny COS y + Csin^y = 0.

B^ *
Divide by -^
sin^ v, substitute for
''
C its value and clear of
4.A
fractions ; the result is

4:A''cot''y + 4.ABcoty-hB^ = 0,
or (2 ^ cot y -h By = 0.
Hence

(6) cot y =—
2A
We choose that value of y satisfying (6) which lies between
0° and 180°.
If the axes are rotated through this angle y, then a = 0,
6 = and, from =^+
(4), c (7. Thus (2) becomes

(7) {A + C) + dx' y'-' +ey'-{-F=0,

where the values of d and e are to be computed from (3).
Equation (7) can now be treated by the method of § 1, Ex-
ample 3.

Example. The equation

(8) 30^2 4- 12xy -f- 12?/2 + 10a; -f 10?/ - 3 =
is of parabolic type, since B'^ — AC = 144 — 4 3
4= • • 12 = 0.
* J. ^ 0, for otherwise B= ; and we have assumed B ^ 0.
252 ANALYTIC GEOMETRY

Here, (6) becomes

C0ty = -V = -2,
whence
y = 153°26', smy = 4V5, cosy = -|V5.
and compute the values
E-otate the axes through this angle y
which A-\- and e have in this case. There results, as
C, d,
the equation into which (8) transforms,

15 ^2 _ 2V5x' - 6V52/' -3 = 0.
This equation can be rewritten as

15 (2/'2_ I V5 -h i) = 2 V5 + 3
2/' oj' -}- 3,

or 15(2/' - iV5)2 ^ 2V5(aj' + f V5),

or, finally, as

(9) 152/''2 = 2V5aj",

where
(10) K2/') = (-iV'5,iV5)
has been introduced as new origin of coordinates.
It follows from equation (8) represents a parabola
(9) that
with its vertex at the new origin (10) and with its axis inclined
at an angle of 153° 26' to the axis of x.
To find the coordinates of the
vertex (10) with respect to the
original axes, we substitute in for-
mulas (2) of § 2, first, the values
which sin y and cos y have in this
case :

Fig. 6 X
2x' + y' x'-2y'
V5 '

and then the values for x', y' given by (10). We obtain, as
the desired coordinates, (x, y) = (l, — 1).
We return now to the general case, li d=^0, equation (7)
represents a parabola ; cf. Ex. 16, § 1. If d = 0, (7) can be
written in the form

(11) y'^ + 2ky' = l,

 EQUATION OF THE SECOND DEGREE 253

where we have divided through by A -\- C* and introduced

simpler notations for the resulting constants. Equation (11)
becomes immediately
(12) ^y'^k)2 = Jc2 + l',

consequently, it represents two parallel lines, a single line, or

has no locus, according as A;^ + ^ is positive, zero, or negative.
To obtain the condition for the exceptional case, d = 0, in
terms of the coefficients of (1), we note that, since B- — 4,AC
= 0, AC> and A and C are of the same sign. We assume
that A and C are positive if they were negative, equation (1)
;

could be multiplied through by — 1. Since

(13) B = ± 2^ AC,
(6) can be written as

COty = T^-|,
whence it can be shown that

smy cos

Hence, from (3),

 :

254 ANALYTIC GEOMETRY

We collect our results in tlie form of a theorem.

Theorem 5. An equation (1) of parabolic type:

B'-4:AC = 0,
represents a parabola, if A ^ 0. /f A = 0, it represents two parallel
lines, a single line, or has no locus.

We have proved the theorem on the assumption that B :^0.

If B = 0, then either ^ = or (7=0, and the content of the
theorem is easily verified.
If A= 0, the given equation may be treated directly, with-
out change of axes. An equation of this type is

Sx" + 24a^ + 18?/2 _ 14a; -21y + S = 0.

It can be written in the form

2(2ic *+ 3^)2- 7(2x + Sy)+S=0',

the left-hand side can then be factored :

[2(2x + 32/)-l][(2a;-f 32/)-3]=0.

The equation, therefore, represents the two parallel lines

4a; -f- 6?/- 1 = 0, 20.' + 31/- 3 = 0.

Second Method. We notice that equation (8) can be written
in the form
(16) 3(a; + 22/)2 + 10.^ + 10?/-3 = 0.
There are two linear expressions in (16), namely, that in the
parenthesis and that consisting of the remaining terms in the
equation. If the lines represented by these expressions, set
equal to zero, were perpendicular, (16) could be simplified by
introducing these lines as coordinate axes. As the equation
stands, these lines are not perpendicular. We can, however,
rewrite it in a form in which they will be.
Add an arbitrary constant k to the expression in the paren-
thesis in (16) and, in equalization, subtract 6 k(x -f- 2 ?/) -f 3 A;-
from the remaining terms

(17) 3(x 4- 2?/ + ^')2-f (10 -6k)x-^ (10 -12k)y - 3 -3k^ = 0.

 EQUATION OF THE SECOND DEGREE 255

Determine k so that the two lines defined by the linear ex-

pressions in (17) are perpendicular, that is, so that

10-6A: ^^
10-12A;~'"
Thus -10 + 6A: = 20-24A; and A: = 1.

For A: = 1, (17) becomes

(18) 3(x J^2y + iy^2{2x-y- 3)- 0,

and the two lines in question are perpendicular. The equations
of a transformation introducing these lines as axes were found
in § 4 of Ch. XI, and are given by
formulas (5) of that paragraph.
Referred to the new axes, (18)
becomes
3(V5a;7 + 2(- V52/0= 0,
or 15ic'2_2V52/' = 0.
Fig. 7
Hence we have shown again that
(8) represents a parabola with its vertex at the point (1, — 1)
and with its axis inclined at an angle of 153° 26' to the axis
of X*
The first of the two methods described is more direct and
more in keeping with previous methods. Its application to
a particular equation, however, is handicapped by the early

*To treat the general equation (1) by this method, assume that A and
C are positive. Then, since B = ± 2 VJ-C, (1) can be written as

{\/Ax ± VCyy + Dx-{- Ey+ F = ().

From this point the discussion proceeds as in the example in the text.
It can be shown that the exceptional case arises when and only when

VAx ± VCy = 0, Dx-^Ey+Fz=0

are parallel ; i.e. when and only when
EVA T I>VC = 0.

But this is precisely the equation (14) obtained by the first method.
From it follows that A = is the condition for the exceptional case.

Thus we have, in sketch, the proof of Theorem 5 by the second method.

256 ANALYTIC GEOMETRY

introduction of radicals. The second method avoids this dis-

advantage, and is the more elegant, though perhaps theoreti-
cally the more difficult, of the two.

EXERCISES
Identify and plot roughly the curve defined by each of the
following equations. If a change of axes is necessary, show
the new axes on the graph.
1. 16aj2_24a.-2/+9?/2-38a;-34?/ + 71 = 0.
2. 9x^-24.xy-\-16y'^-\-Sx-Ay-Q = 0.
3. 2oa;2 -f + lUy'^ + 86 - 233y + 270 = 0.
120xy a;

4. - 20xy + 20y^ -\-2x + y-{-S=0.

5.^•2

5. 25x^-\-30xy-{-dy'~-\-10x + 6y +1 = 0.
6. x'^—2xy + + Sx — y—4: = 0.
y''^

7. x''-^xy-\-4.y^-\-3x-6y-10 = 0.
8. 27x''- 36xy-}-12y^-4:0x-\-lSy +32 =0.
9. 2x^+12xy + + 13y + = 0.
+lSy'^ x 9
10. 4a;2 + 12a;^ + 9y'^ + 2x + 3y + 2 = 0. Ans. No locus.

6. Summary. Invariants. The content of Theorems 3, 4, 5

we summarize in the following table.
 EQUATION OF THE SECOND DEGREE 257

a cylinder as its and the plane approaches a position

limit,
parallel to the rulings of the cylinder. But the section of a
cylinder by a plane parallel to the rulings is two parallel lines,
a single line, or nothing. These sections of the cone or cylin-
der are called degenerate ; those first mentioned, non-degenerate.
From the above table we can now draw a general conclusion.

Theorem 6. An equation of the second degree, if it has a

locus, represents a conic section, which is non-degeyierate if A ^ 0,
and degenerate if A= 0.

Invariants. We have seen that the value of the quantity

^ -f C is unchanged by a rotation of axes [§ 3, (7) and § 5, (4)].
This is true also of the value of the quantity ^ — 4^C[§ 3,
(16) and § 5, (o)]. We say that ^ + (7 and 1^-4. AC are
invariant under a rotation of axes. They are also invariant
under a change of origin, since we saw, in § 4, that the quadratic
terms in the general equation are not affected by a change of
origin.
Consequently, A -\- C and B^ — 4: AC are invariant under any
change of axes. For, any change of axes consists. of a change
of origin, combined with a rotation of axes.
It can be shown that the discriminant A is also invariant
under any change of axes.
The importance which these quantities, A, B^ — 4t AC, smd
A -\- C, have assumed in the course of the treatment is closely
related to the fact that they are invariants with respect to any
change of axes. For, it is clear that a quantity whose value
varies with the choice of axes can have no particular signifi-
cance in a theory which deals primarily with properties of
the curve which are independent of the choice of axes, whereas
it is to be expected that an invariant quantity would play an

important role.

EXERCISES ON CHAPTER XII

In each of the following exercises, determine the nature of

the curve represented by the given equation, and then find its
258 ANALYTIC GEOMETRY

position. Draw a figure, showing the curve, the original axes,

and any new axes used.

2. 7x'^-Sxy + y^ + Ux-Sy + W = 0.
3. Sx'' + Sxy-^2y'--6x-3y-5
= 0.
4. 4.t2 + 8x'?/ + 4?/2 + 13a; + 32/ 4-4 = 0.
5. 9aj2-8a;2/ + 242/—
32x-16?y + 138 = 0.
6. + xy-2y^-llx-y -h2^=0.
x'^

7. 9a;2 + 24x2/ + 162/2 +

8iK- 62/ + 3 = 0.
8. 3x'^ + 4:xy-^10x + 12y +
7 = 0.

9. 3a;2 + 4x2/ + 22/2+8x + 42/ + 6 = 0.

10. 9^2 -15x2/ + + 63^ = 0.
2/^

11. 10x2 - 12x2/ + 52/2 -84x + 562/ -14 = 0.

12. 25x2- 20x2/ + 42/2 + 2OX-IO2/
+ 5 = 0.
13. 6x2 + 12x2/ + 2/'-36x-62/
= 0.
. 14. 32x2 + 48x2/ + 182/2 -57x- 242/ + 6 = 0.

15. 4x--Tx2/-22/2 + 22x + 2/ + 10 = 0.

16. 7x2-18x2/-lT2/'-28x + 362/+8 = 0.
17. 9x2-12x2/ + 42/2 + 4x-592/
+ 38 = 0.
18. 7a;2_5x2/ + 2/'-42x + 152/ +
63 = 0.

19. 14x2 + 24x2/ + 212/2 + 52x + 662/

+ 14 = 0.
20. 9x2 + 12x2/ + 42/2 -4x- 72/
-4 = 0.
+ + IO2/- 4 = 0.
21. 20x2 + 23x2/ + 62/2 11a;

22. 25x2-7x2/ + 2/'-407x + 162/

+ = 0.
13

23. 49x2- 28x2/ + 42/2-42X + 122/

+ 9 = 0.
24. 4x2 + 6x2/ + 52/2 + 2x + 72/ +
3 = 0.

25. a;2-2x2/-6x + 42/ + 4

= 0.
26. 2x2-x2/ + ^2_72/ + 6=0.
degree represents
27. that an equation of the second
Show
perpendicular lines, if and
an equilateral hyperbola or two
only ifA + C = 0.
 EQUATION OF THE SECOND DEGREE 259

28. If the equation (1), § 3, represents a hyperbola, show

that the asymptotes are defined by the equation

Ax^ + Bxy + Cy^ = 0.

29. If the general equation of the second degree represents
a hyperbola, prove that the asymptotes have the directions of
the lines defined by the equation of Ex. 28.

30. Prove that every equation of the form

AB («2 - 2/2) _ (A" - B^)xy = 0,

where C^O and not both A and B are 0, represents a rectan-
gular hyperbola with the lines

Ax + By = 0, Bx — Ay =
as asymptotes.

31. Show that the equation of every rectangular hyperbola

can be written in the form
4)

AB(x^ - y2)_(^2 _ 52)a;i/ + jr>x + ^?/ + F= 0.

32. Find the equation of each of the rectangular hyperbolas
(a) 12x^-7xy- 12?/^ - ITa.- + 31 - 13 = 0,
1/

(6) 6a;2 J^^xy-^y'^- 39a; + 26?/ - 13 = 0,

referred to the asymptotes as axes.

33. Show that for just one value of \ the equation

A.a;2 -f 4x?/ + 2/2 _ 4a; _ 2 y — 3 =

represents two straight lines. Find the equations of the lines.

34. If the general equation of the second degree represents

an ellipse or hyperbola, what is the condition that the center
be at the origin ?

35. If the general equation represents a parabola, show that

the vertex is at the origin if and only if

AD' + CJ^2 4_ BDE = and F= 0.

Suggestion. Write the equation in the form (7), § 5.
260 ANALYTIC GEOMETRY

36. Pyove that, if the equation (tL), § 3, represents an ellipse

or hyperbola, the axes are defined by the equation
5(^2 _ ^2) _ 2(A - C)xy = 0.
37. Show that, in case B^ — 4:AC==0 and A = 0, the gen-
eral equation represents two parallel lines, a single line, or
has no locus, according as the expression

is positive, zero, or negative.

Suggestion. Consider equation (7), § 5, where cZ = 0, and
use (15), § 5, in simplifying the result.
38. Prove that the expression in Ex. 37 can be replaced by
D'^-4.AF,ifA^0', and by JSJs - 4 CF, ii C ^ 0.

Definition. Two conies are said to be similar and similarly

placed, if their eccentricities are equal and their corresponding
axes are parallel.
39. Prove that the conies,
llx' + 6 xy -\- S y^ - 12 X - 12y - 12 = 0,
11 a;2 + 6 a;?/ +3 ?/2 - 34 - 18 + 29 = 0,
cc 2/

are similar and similarly placed.

40. Show that, if the coefficients of the quadratic terms in
two equations which represent non-degenerate conies are re-
spectively equal or proportional, the conies are of the same
type. Prove further that, if they are ellipses or parabolas,
they are similar and similarly placed, and that, if they are
hyperbolas, they are similar and similarly placed or each is
similar and similarly placed to the conjugate of the other.
 CHAPTER XIII

A SECOND CHAPTER ON LOCI. AUXILIARY VARIABLES.

INEQUALITIES

Extension of the Method for the Determination of Loci.

1.

If welook back over the locus problems which we have thus

far solved, we find that there is invariably but a siiigle essen-
tial condition governing the motion of the point tracing the

locus. For example, the sum or the difference of two dis-

tances or of their squares is required to be constant ; or the
slope of one line is given as proportional to that of another.
This single condition it is comparatively simple to express
analytically and thus to determine the locus.
In a great many problems, however, the motion is governed
by not just one, but by two or more essential conditions, in-
terdependent on one another.
An example of such a problem 't
~
is the following : A triangle
has a lixed base AB and its
vertex F moves on an indefinite
straight line L, parallel to the
base. Tind the locus of the
point of intersection P of the
A : (-a,0) O C B:{a,0)
altitudes.
Fig. 1
Here the motion of P is •

governed by tico essential first, by the motion of

conditions ;

V, and secondly, by the fact that P is the point of intersection

of the altitudes. The two conditions are interrelated, since
the position of V determines the position of the altitudes and
hence of their point of intersection. Consequently, they are
261
262 ANALYTIC GEOMETRY

in substance equivalent to a single condition, which, when

expressed analytically, would give the equation of the
locus.
Our problem, then, is to reduce the two conditions to a single
condition. This it is, in general, very difficult to do geometri-
cally. Analytically, however, the task is simpler. For, condi-
tions expressed in analytical form, in terms of equations, are
usually more easily combined than when they are in geometri-
cal form.
Accordingly, we proceed to express analytically the two
conditions governing the motion of P. Take the mid-
point of the base of the triangle as the origin and the
axis of X along the base. Let the length of the base be
2 a and the distance of L above the base, h. The coordinates
of A and B are (— a, 0) and (a, 0). Denote those of P by
(X, Y). ^

The first of the two conditions is that V move along L.

But then the distance, KV, of V from the axis of y varies.
Accordingly, we can express the motion of F' along L by tak-
ing the abscissa of F'as a variable. Denote this variable by y.
The coordinates of "Fare, then, (y, h).
We now have coordinates for the three vertices of the tri-

angle. Hence we can find the coordinates (X, Y) of the

point of intersection of the altitudes. Thus we shall have
expressed the condition that P be this point.
The coordinates (X, Y) of P will be obtained in terms of
the constants, a and and the variable y. If we eliminate y
h,

from the two equations which give the values of and Y, the X
resulting equation will contain only a and h, and and Y, and X
will be the equation of the locus of P.
The variable y is known as an auxiliary variable, or param-
eter. It helps in expressing analytically the conditions
governing the generation of the locus. The method involving
its use, which we have just described, general in scope, andis

may be applied with advantage to any locus problem contain-

ing multiple conditions.
 A SECOND CHAPTER ON* LOCI 263

2. One Auxiliary Variable. In illustrating the method by

examples, let us first complete the problem of the previous

paragraph.
Consider P as the point of intersection of VC and BD.
The equation of VC is
(1) X=y.
The slope of A V is
h

hence the equation of the perpendicular, BD, to AVis

(2) 2,_0 = -'!^'(x-«).

Solving equations (1) and (2) simultaneously, we obtain the

coordinates of P,

(3) X=y, Y='^^,

in terms of the constants a and li and the auxiliary variable y.

By eliminating y from equations (3), we obtain

(4) X2 = -/ir-+-a2
as the equation of the locus.
The locus of P
is, then, a parabola, with its axis along the

perpendicular bisector of the base of the triangle it goes ;

through the extremities of the base and opens away from the
line L. Every point of it is included in the locus.*

* In the locus problems considered hitherto, particularly in Ch. V,

care was taken to emphasize that two things are necessary (a) to :

determine the curve, or curves, on which points of the locus lie; (6) to
show, conversely, that every point lying on the curve, or curves, obtained
isa point of the locus. In the problems of the present chapter, for —
example, in the one above, —
part (6) of the proof is usually omitted. It
consists, as a rule, in retracing the steps of part (a) and so presents, in
general, no difficulty. And it is more important, now, that the student
gain facility in deducing the equation of the curve, or the equations of the
curves, which turn out, in the great majority of cases, to be precisely the
locus.
 :

264 ANALYTIC GEOMETRY

Remark. It was not necessary to find the actual coordinates

(3) of P. The fact that P is the point of intersection of the

altitudes might have been expressed by writing down the
conditions that {X, Y) satisfy equations (1) and (2), namely,
X=y, liY=-{y + a){X-a).
If we eliminate y from these equations, we obtain equation
(4) of the locus.

Example 2. A straight line L passes through a fixed point

Pq ; find the locus of the mid-point P of the portion of L inter-
cepted by two given perpendicu-
lar lines, neither of which goes
through Pq.
Take the two given lines as
(x,r) axes, and let the coordinates of
Pq, referred to them, be {xq, y^).

X The conditions governing the

X
motion of P are, first, the rota-
tion of L about Pq, and secondly,
the fact that P is the mid-point
of the segment AB.
Fig. 2 We express the rotation of L
by taking its slope, A, as aux-
iliary variable. The equation L is,
of then,

2/ - 2/0 = (^ -
>. ^'^^'o)-

The coordinates of the points of intersection of L with the

axes are
A'.(xQ-y^,o)', B :{(), Vq-Xxq).
A

Hence the coordinates of P, the mid-point of AB, are

^-1--!. Y=h{yQ-^^^)'

Toeliminate X from these two equations, we might solve

the first for X and substitute its value in the second. But we
 : — ;

A SECOND. CHAPTER ON LOCI 265

notice an easier metliod ; rewriting the equations in the form

2x-«„ y^,
A

and multiplying together the left-hand sides and then the

right-hand sides, we obtain the equation

(5) {2X-x,){2T-y,)=x,y,,
devoid of A.

The equation of the locus, in this form, or better, in the

form

suggests that we change to parallel axes, with the new origin

at(t.|):

2' "^
2

The locus, referred to the new axes, has the equation,

4:xy = XoyQ,
and is, therefore, a rectangular hyperbola. It follows from (5)
that the hyperbola goes through and Pq.
To describe the locus independently of the coordinate sys-
tem : Let be the point of intersection of the given lines
the locus is a rectangular hyperbola through O and Pq, with
its center at the mid-point of OPq and with its asymptotes
parallel to the given lines.

EXERCISES
1. line L parallel to the axis of x.
Given a Through the
origin draw a variable line meeting L in Q, and on this vari-
able line mark the point P whose ordinate equals the abscissa
of Q. What is the locus of P?
Ans. The parabola y^ = hx, where h is the algebraic distance
from the axis of x to L.
266 ANALYTIC GEOMETRY

2. A line-segment AB of fixed length moves so that its

extremities lie always on two perpendicular lines. Find the
locus of the point dividing AB in the ratio 2 : 1, using as
auxiliary variable the angle which the moving line makes with
one of the two perpendicular lines.

Ans. An ellipse, center in the point of intersection of the

given 4ines, axes along them, with length and breadth in the
ratio 2 : 1.

3. Determine the locus described in Ex. 2, when the given

ratio is mi : m2.

4. The line L is the perpendicular bisector of the fixed

horizontal line-segment AB. The points li and S are taken
on L, with E always below S, so that the distance MS is one
half the distance AB.
Find the locus of the point of inter-
section of AR and BS, taking the axes of x and y along AB
and L. Ans. The hyperbola, 2xy -\-x^ = a^, through A and B.
5. A variable line is drawn through a fixed point Pi meet-
ing a fixed line L in Pg. Points P are taken on this line so
that the product of the distances PiP and P^P^ is constant.
Find the locus of these points.
Ans. Two circles, tangent at Pi to the line through Pi par-
allel to L.
6. Find the locus of points from which the tangents drawn
to a parabola are perpendicular.
Suggestion. Use the equation of the tangent with given
slope, Ch. IX, § 6, eq. (10).

7. The same for the ellipse.

8. The same for the hyperbola.

9. Determine the locus of the mid-points of all the chords

drawn from the vertex of a parabola.

3. Coordinates of a Point Tracing a Curve, as Auxiliary Vari-

ables. In the problems of the previous paragraph one of the
conditions governing the motion of the point tracing the locus
was the auxiliary motion of a line or of a second point. In
 .

A SECOND CHAPTER ON LOCI 267

each case this auxiliary motion could be expressed analytically

by the introduction of one auxiliary variable.
Suppose, now, that the auxiliary motion consists of the trac-
ing of a given curve, not a straight line, by a point R. Let
the curve be, for example, the circle

(1) a;2 + 2/2 =a 2

The motion of R on the circle might be represented analyti-

cally by the introduction of a single auxiliary variable, e.g.
one of the coordinates of R ; but it is in general simpler, ana-
lytically, to represent the motion by tivo auxiliary variables,
namely, by both the coordinates {x\ y') of R. These will be
connected by the equation,

(2) x"^ -f 2/'2 = a?^

which states that R is on the circle.

The reason for this choice of auxiliary variables lies partly

in the fact that we thereby avoid radicals ;
* partly in the
principle of algebraic symmetry. term we mean to By this
signalize the fact that equation (1) bears equally on x and y,
and so it is well to carry the solution through in such a man-
ner that it, too, will bear equally on the two coordinates of
each of the principal points involved.f

Example 1. Let AA' be a fixed diameter of a given circle

and let RR' be a variable chord perpendicular to AA'. What
is the locus of the point of intersection, P, of AR and A'R' ?

Choose the center of the circle as the origin and the axis of
X along AA\ Then (1) is the equation of the circle.

* If we had taken x' as a single auxiliary variable, the coordinates of

R would be {x\ ± \/a2 — x'2).

t It is possible to represent the motion of i? by a single auxiliary vari-

able and at the same time and preserve symmetry, by
to avoid radicals
choosing as the auxiliary variable the angle d which the radius drawn to
R makes with a fixed direction, e.g. the axis of x ; the coordinates of R
are then : x = a cos 6^ y' = asin^. We however, to use as aux-
prefer,
iliary variables the coordinates of R connected by equation (2)

\
 : :

268 ANALYTIC GEOMETRY

Take, as the auxiliary motion, the tracing of the circle by

the point It and, as auxiliary variables, the coordinates (x', y')

of R. These are connected

^ ^ by the equation (2). The
^P:{X Y) coordinates of R' are, evi-
dently, {x', - y').
The equations of AR
and A'R' are

2/-0=-yL(aj4.a),
X -\- a

y
y (x— a).
X —a
Fig. 3 Since P is the point of in-
tersection of AR and A'R'j
its coordinates {X, Y) satisfy both these equations

(3) T=-y—{x+d).
x'-{-a

(4)
x'

We have, then, three equations, (2), (3), and (4), involving,

besides the constant a, the coordinates (X, Y) of the moving
point and the auxiliary variables x', y'. To obtain an equa-
tion in X, Y" alone, we must eliminate x', y'. We shall do this
by solving two of these equations, preferably (3) and (4),
simultaneously for x' and y\ and substituting the values
obtained for them in the third equation, (2).
To this end we rewrite equations (3) and (4) as follows

(3a) Yx'-{X^a)y' = -aY,

(4a) Yx' + {X - a)y' = aY.

Hence x' =— = «^-

X'
2/'
^ X
Substituting these values in (2) and reducing, we obtain

as the equation of the locus.

 A SECOND CHAPTER ON LOCI 269

The locus is thus seen to be a rectangular hyperbola with the

given diameter of the circle as major axis. It evident from
is

the figure, however, that if R is restricted to the upper half of

the circle, and R' to the lower half, only the upper half of one
branch and the lower half of the other belong to the locus. It
is only when R and R' are each permitted to trace both halves

of the circle that the locus consists of the entire hyperbola.

Remark. The points A

and A^ do not belong to the locus.
For, the only possible which P can take on the position
way in
of the point A\ for example, is for R and R^ to coincide in A^ ;

but then there is no chord RR' and also no line A'R\ so that
no point P on the locus is determined.
Let us return now to equations (3a) and (4a). In. solving
them for x', we actually obtain
TXx' = a?T.
But Y^ 0, since P cannot lie on the axis of x^ in either A or
A' hence we were justified in dividing by Y, and the result,
;

jc' = a^/X, is correct.

In subsequent problems we shall lay no stress on exceptional
points such as A and A'. Their importance for the student at
this stage is relatively small.

Example 2. A point R traces a pa- R:{x,'y')

rabola. Find the locus of the point of
intersection, P, of the line through the
focus and R
with the line through the
X
vertex perpendicular to the tangent 'F:{f, 0)
at R. [P (X, Y)••

The parabola, referred to the coor-

dinate axes shown in the figure, has
the equation
= 2 mx.
2/2
Fig. 4

The motion of R can be expressed by taking, as auxiliary

variables, its coordinates (x', i/), connected by the relation

(5) y''- = 2mx',

270 ANALYTIC GEOMETRY
which states that R, in moving, stays always on the pa-
rabola.
The slope of the tangent at R is m/y', Ch. IX, § 2, eq. (5);
consequently, the line through perpendicular to the tangent is

y = — ^-x
m
As the equation of FR we have

y = —
X —m
-[0^-11

The equations expressing the fact that P:{X, T) is the

point of intersection of these two lines are, therefore.

(6) Y=-^X,
m
771

77l''
(7) (x'-f^Y=y'^X
2,

From equations (5), (6), and (7) we have to eliminate x' and
y'. Solving (6) and (7) for x^ and y', we have :

y'z=—m
^
— a;' = _v 1.
X 2 X
Substituting these values for y' and x' in (5) and reducing the
result, we obtain
X2 -|_ Y^ - mX =
as the equation of the locus.
The locus is therefore a circle, passing through the vertex of
the parabola and having its center at the focus. The vertex,
0, is not a point of the locus.*
Elimination of x', y'. In each of the above examples we
eliminated the auxiliary variables x', y' by solving the last
two of a set of three equations for x' ^ y' and substituting the
values thus obtained for cc', y' in the first equation, — the
* It
au exceptional point, similar in type to the exceptional points,
is

A andJ.', of Example 1. For, when E is at 0, FU and OT coincide and

consequently determine no point on the locus.
 :

A SECOND CHAPTER ON LOCI 271

equation stating that the point {x\ y') lies on the given curve.
This method is valuable because of its general applicability.
The student should, however, be on the alert for short cuts in
the elimination. For example, he might have noticed by close
inspection that, in Example can be eliminated easily
1, x', y'

from equations (2), (3), (4) by multiplying equations (3) and

(4) together

^' = -J^S^'-^)^
x'^ — a-

and by noting, from equation (2), that the quantity

x'^ — a?
has unitv as its value.

EXERCISES
1. Let A A' be the major axis of an ellipse and RR' be a
variable chord perpendicular to AA'. Find the locus of the
point of intersection of AR and A'R'.
2. Given and a variable chord
a fixed diameter of a circle
parallel to Find the locus of the point of intersection of
it.

the line through the mid-point of the chord and one extremity
of the diameter with the radius drawn to the corresponding
extremity of the chord. What is the locus if the radius is

drawn to either extremity of the chord ?

Ans.. Part of a parabola ; the parabola.
3. Find the locus of the point of intersection of the line
drawn through a given focus of an ellipse perpendicular to a
variable tangent with the line joining the center to the point of
tangency.
Ans. The directrix corresponding to the given focus.

4. Let i? be a point tracing an ellipse. Find the locus of

the point of intersection of the line drawn through the center
perpendicular to the tangent at R with the line drawn through
R parallel to the conjugate axis.
Ans. An ellipse, similar to and having the same axes as
the given ellipse, but with foci on the opposite axis.
272 ANALYTIC GEOMETRY

5. The preceding problem for a hyperbola.

6. Find the locus of the point of intersection of the line

drawn through a variable point i? of a parabola parallel to the
axis and the line through the vertex perpendicular to the tan-
gent at R.

7. A variable tangent to an ellipse meets the transverse

axis in the point T. Determine the locus of the point of
intersection of the line drawn through T parallel to the con-
jugate axis and the line joining the point of contact of the
tangent to a vertex.

8. The preceding problem for a hyperbola.

9. Let RR' be an arbitrary chord of an ellipse parallel to

the conjugate axis; let the normal at R meet the line joining
the center to R' in the point S. Find the locus of the mid-
point of RS.
10. The preceding problem for a hyperbola.

4. Other Problems Involving Two or More Auxiliary Vari-

ables. There are problems in which it is convenient to use
two auxiliary variables other than those of the type which
we considered in the preceding paragraph.

Example. The points A and B are fixed and the line L is

perpendicular to AB at its mid-point, 0; R and S are two

points on L, both on the same side
of AB and moving so that the
(0,s):S
product of their distances from
P\{X,Y) is constant, and equal to Ir. Find
the locus of the point of intersec-
AR and BS.
tion, P, of

B:{a,0) Take the axes as shown in the

figure and let AB = 2 a. The
motions of R and S can be repre-
sented by taking their ordinates, which we denote by r and s,
as auxiliary variables. The condition that R and S are on the
 :

A SECOND CHAPTER ON LOCI 273

same side of AB and relatively so situated that OE OS =

•
b^
is then given by the equation,

(1) rs = 62.
The equations of AM and BS are
—a r as
Since P : (X, Y) is the point of intersection of these lines, we
have

(2) 1+^
a
= ^,
r

(3) 1-^=1,
a s

To eliminate the auxiliary variables r and s from equations

(1), (2), (3) is now our problem. We notice that in the product
of equations (2) and (3)
^_X2_F2
a" rs

r and s enter only in the form rs, and that the value of rs is
given by (1) as 6\ We have, therefore, as the equation of the
locus

a" 62

The locus of P is, therefore, an ellipse, with its axes along

AB and L, and passing through the points A and B. These
points are not, however, points of the locus.

EXERCISES
1. What is the locus of P in the problem in the text, if E
and S are always on opposite sides of AB ?
2. The points Pi and P2 are fixed, and the Jines Li and L2
are perpendicular to P1P2 in Pi and P2 respectively Qi and ;

Q2 are two points on Li and L2 respectively, both on the same

side of P1P2 and moving so that the product of their distances
274 ANALYTIC GEOMETRY

from Pi and Pg respectively is constant. Find the locus of

the point of intersection, P, of P1Q2 ^^^ P^Qi-
3. What is the locus of P in the preceding example, if Qi
and Q2 are always on opposite sides of P1P2 ?
4. Do Ex. 2, § 2, using the intercepts of the moving line AB
on the two given perpendicular lines as auxiliary variables.

5. The same for Ex. 3, § 2.

6. The points R and S move, one on each of two fixed
perpendicular lines, so that the segment RS subtends always
a right angle at a fixed point, not at the intersection of the
two lines. Eind the locus of the mid-point of RS.
Ans. Perpendicular bisector of the line-segment joining the
fixed point with the intersection of the fixed lines.

7. Two right angles, having their vertices in fixed points

A and B, rotate about these points, so that the point of inter-
section of two of their sides traces a line parallel to AB.
What is the locus of the point of intersection of the other two
sides ?

5. Use of the Formula for the Sum of the Roots of a Quad-

ratic Equation. The sum of the roots of a quadratic equation
(Ch. IX, § 5),
Ax'^ + Bx-^-C^O, A^O,
is the negative of the ratio of the coefiicients of the terms in
X and a?- ; that is,

(1) a;i + »2 = -^5

where Xi and 072 are the roots.

As a simple example of the way in which this fact may be

used to advantage, let us find the coordinates of the point P
midway between the points of intersection. Pi and Pj, of a
line and a conic. Take, for example, the line

(2)
*
2x-y = l,
and the ellipse

(3) 3a;2 + 42/2 = 3.

 : : ;

A SECOND CHAPTER ON LOCI 275

The coordinates {xi, 2/1) and {X2, 2/2), of P^ and P2, are the
simultaneous solutions of equations (2) and (3). Substituting
in (3) the value of y from (2) and
collecting terms, we have the quad-
ratic equation, Pi'^^vVi)

19a;2-16.T-|-l = 0, X
for the determination of x^ and x^.

We are interested, not in the

actual values of x^ and X2, but in K '{Xnty^)

half their sum ; for this is the

Fig. 6
abscissa of the mid-point, P, of
P1P2. By (1) the sum is if. Then the abscissa of P is
^^
and, since P lies on the line (2), its ordinate is

2/ = 2(t%)-1=-A-
Consider now the following locus problem : A variable tan-
gent to the parabola,
2/^ = 2 mx,
ii:i^',y')
meets the hyperbola,
xy = c%
in the points Pj and Pg.
What is the locus of the mid-
point, P, of P1P2 ?
As auxiliary variables we
take the coordinates {x', y') of
the point R tracing the parab-
YiQ, 7 ola ; they satisfy the equa-
tion of the parabola

(4) ?/'2 = 2 mx'.

The tangent at R has the equation
(5) y'y = m{x -\- x').

To find the coordinates of Pi and Pj, we solve (5) simultane-

ously with the equation of the hyperbola. Eliminating 2/, we
have
Tiix^ + mx'x — (?y' = 0.
 276 ANALYTIC GEOMETRY

Half the sum of the roots of this equation is X; hence,

by (1),

(6) X=-2 x'

Since Y is the ordinate of the point on the line (5) whose

abscissa is given by (6), we have

+ xA mx'
x' ^
, (7) y'Y=m{-- ,

We now have three equations, (4), (6), and (7), from which
to eliminate the auxiliary variables x' and y'. We solve (6)
and (7) for x' and y\ obtaining

x' = -2X, y' = --^.

Substituting these values of x' and y' in (4) and simplifying

the result, we have
4r2=-mX.
Consequently, the locus of P
is a parabola with vertex at

the origin and opening out along the negative axis of x as

axis. The origin is not a point of the locus.

EXERCISES
1. A variable tangent to the circle
x'^ + f- = a?
meets the hyperbola
2xy = a?

in the points Pi and Pj- Find the equation of the locus of the
mid-point of PiP2. Plot the locus.

^iis. —— = —
£c2
1

2/^ a^
, a curve which does not, despite its ap-

pearance, consist of two conjugate rectangular hyperbolas.

2. Two equal parabolas have the same axis and vertex, but

open in opposite directions. Find the locus of the mid-points

of the chords of one which, when produced, are tangent to the
other.
 A SECOND CHAPTER ON LOCI 277

3. Find the locus of the mid-points of the focal chords —

chords through the focus — of a parabola.
Ans. A parabola with its vertex in the focus of the given
parabola, with the same axis, but half the size.

4. Determine the locus of the mid-points of one set of focal

chords of an ellipse.

5. The same for a hyperbola.

6. A variable tangent to a parabola meets the tangents at

the extremities of the latus rectum in the points Pi and P2.

Find the locus of the mid-point of PiPj-
7. The asymptotes of a hyperbola intercept the segment
P1P2 on a variable tangent. What is the locus of the mid-
point of PjPo ? Ans. The hyperbola itself.

6. Loci of Inequalities. Though we are concerned primarily

in mathematics with equalities, it is not infrequent that in-
equalities become important. Accordingly, it is not out of
place to consider here the loci of some inequalities.

Example 1. The equation x —1= represents all the

points of the line parallel to and one unit to the right of the
axis of y, and no other points. Consequently, the inequality,
x — 1 ^ 0, represents all the points of the plane not on this
line. In particular,
a;-l>0
represents all the points to the right of it and
a;-l<0
represents all the points to the left of it.

Example 2. What is the locus of points whose coordinates

satisfy the inequality

(1) 5aj + 122/ + 6>0?

The equation obtained by replacing the sign > by the sign
of equality represents the line L shown in the figure. From
278 ANALYTIC GEOMETRY

Example 1 we should expect that the locus of (1) would con-

sist of all the points on one side of L. This is, in fact, the
case : If the quantity,

(a^i,2/i)
is positive for a certain point
{xq, ?/o), then it is positive for all
points on the same side of.i> as

{^Q, Vo)-
Fig. 8
We prove this by showing that
the opposite assumption leads to a contradiction. Suppose
that F becomes negative for some point {xi, y^) on the same
side of L as (xq, 2/0) • Joiii (^oj Vq) to (a^i, 2/1) by any curve G
not cutting Z, and let a point {x, y) trace this curve. For
(o^o, 2/0)5 -^ is positive ; but when (x, y) has reached (xi, yi), F
has become negative. Consequently, for some intermediate
point B on C, F has the value zero, inasmuch as its value
changes continuously as {x, y) moves along C. Hence must R
lie on L, — a contradiction, since we took C as a curve never
cutting L.
To ascertain on which side of L the points represented by
(1) lie, we have but to find the value of F for one point not
on L. In this case the simplest point to take is the origin.
But, when x = and y = 0,Fis positive. Therefore the locus
of (1) consists of all points on the same side of L as the origin.

Example 3 What is the locus of the inequality

(2) y^> 2x?

The equation, obtained by replacing the sign > by the
equality sign, represents a parabola. By the reasoning of
Example 2, then, the inequality represents all the points
within, or all the points without, the parabola. The latter is
clearly the case, since (2) is not satisfied by the coordinates
of the point (1, 0), — a point which is within the parabola.
 :

A SECOND CHAPTER ON LOCI 279

EXERCISES
Find the loci of the following inequalities. Draw a figure
in each case and shade the area of points represented by the
inequality.
1. + 2 > 0.
a; 2. + 3< 0.
2?/

3. a;+2/+ l>0. 4. 3x-4:y-2>0.

5. 2x-3y <0. 6. a;2 + ?/2< 1.

7. + 7a.'>0.
2/2 8. 3x^ -\- 4.y^ > S.

9. a;2-2/2<l. 10. 3aj2-22/2< -6.

7. Locus of Two or More Simultaneous Inequalities.

Example 1. Find the locus of points whose coordinates

satisfy simultaneously the two inequalities

(1) 5.T+ 122/ + 6>0,

(2)' 3.T-42/-2>0.
Denote the left-hand sides of (1) and (2) by F^ and F^, re-

spectively. By Example 2 of § 6, the points whose coordi-

nates satisfy (1) are all the
points which are on the same
side the line Li:Fi
of as =
the origin similarly,
;
the points
whose coordinates satisfy (2)
are all the points which are on
the opposite side of the line
L2'F2 = from the origin.
The points whose coordinates Fig. 9
satisfy (1) and (2) are the
points common to these two sets, namely, those of region I
of the figure.
Lying between the lines Li and L2 there are four regions,
III, IV.
I, II, It is clear from the foregoing that the pairs of
simultaneous inequalities representing these regions are

f
F, > 0, IV f
F, < 0,
II III
li^2>0; F,<0; Fo>0.
 : ;

280 ANALYTIC GEOMETRY

Example 2. Find the points satisfying simultaneously the

inequalities
?/2-2a? < 0, x + y -1 <0.
The equations obtained by replacing the signs < by signs of
equality represent a parabola and a line intersecting it. The
locus of the first inequality is the interior of the parabola;
that of the second is the half-plane bounded by the line and
containing the origin. Common to these two regions is the
finite region contained between the parabola and the line
this, then, is the locus of the two inequalities taken simul-

taneously.

EXERCISES
Find the locus of points whose coordinates satisfy simul-
taneously the following sets of inequalities. Draw a figure in
each case, and shade the region represented.

f4x-3 < 0,
2.
l5x-12tj-{-26 >0,
1.
\Sx-\-2y-6 < 0. 3X + 42/-10 > 0.
l2x-y-\-3<0, f
x'^-\-y^ < 4, Ans.
4.
3.
\4.x-2y + 9 >0. {

I
aj— 3 > 0. No locus.
+ 42/^-12 >0,
|3a!2 ix^-Ty < 0,
6.
5.
\2x-3y-\-12>0.
{

[ iC2 - 2/2 + 1 > 0.

i2x-y-S<0,
7. )x + Sy-5 < 0, 8. y>o,
^5x + + 3>0. 2/
2x — y>0.

Each of the following pairs of curves divide the plane

(minus the points on the curves) into a number of regions.
Find the pairs of simultaneous inequalities representing these
regions.

9. 5?/+ 8 = 0, 3a;+ 8^-2=0.

10. 5a;-121/ + 26 = 0, 3a;+ 4?/ - 10 = 0.
11. 2x2 + = 8,
2/^ 4a;-32/-2 = 0.
12. y'^=z2mx, x — 0.
 :

A SECOND CHAPTER ON LOCI 281

13. = 4,
a:'^-2/- 2x-y -2 =0.
14. + 4?/ = 0,
.T- 2x-Sy-6=zO.
15. if-{-Sx = 0, + y^ = 9.
a;2

8. Bisectors of the Angles between Two Lines. The two lines

A: 5a; + 12?/ + 6 = 0,
Xs: 3a;- 4^-2 = 0,
are given. It is required to find the equations of the lines bi-
secting the angles between them.
We solve this problem by finding the locus of the point
P : (X, Y) moving so that its distance Di from Li equals its

distance D2 from L2 :

Di = A.
According to Ch. II, § 8, Di and D2 are

n^ ^5X + 12F+6 ^^ ^3X-4F-2

' '
13 5
where, in each case, that sign is to be chosen which will make
the distance positive.

The lines Li and L2 are those of Example 1, § 7. It follows,

from the results there given in connection with Fig. 9, that
the signs which must be taken to make Z>i and Z>2 both positive
are
If P is in I, + for A, + for A ;

if P is in II, + for D^, - for D. ;

if P is in III, — for Di, — for D2 ;

if P is in ly, - for D„ + for A-

For, if P lies, for example, in the region I, then the numerators
in the expressions for A ^^^ both positive and the +
Di and
sign must be taken in each case make A ^^^ ^2 to positive.
If, now, P is in I or III= Aj we have
and Di

5(5 X+12F+6)=13(3X-4F-2),
or, on reducing,
(1) X-8F-4 = 0.
 ;

282 ANALYTIC GEOMETRY

Thus (1) is that bisector of the angles between Lx and L^, which
lies in the regions I and III.
If P is in II or = Z^g, we have
IV and D^
5(5X+12r+6)=-13(3X-4F-2)
or
(2) 16X + 2r+l = 0.

This is the bisector which lies in the regions II and IV.

Simplification. We now give in condensed form the method

of finding the bisectors. By equating D^ and D2, we have
5X+12r+6 ^3X-4F-2
13 5

If we take both signs positive or both negative and reduce the

result, we get (1). If we take the plus sign on the right and
the minus sign on the left or vice versa, and then simplify, we
get (2). The equations (1) and (2) represent the bisectors
which equation represents a chosen bisector is easily deter-
mined by making a plot.

EXERCISES
1. Find the equations of the bisectors of the angles between
the following pairs of lines, and draw a figure which shall
indicate each bisector.

.. |5a;-122/ + 26 = 0, _ \4.x + 3y-2 = 0,

W |3a;+ 42/-10 = 0; ^^^ \
^- - 4 = 0.
2/

2. Find the equation of that bisector of the angle between

the two lines,
4a: -3?/ +3= and 3ic-42/-6 = 0,
which passes through the region between the two lines which
contains the origin.
Find the equations of the circles tangent
3. to the lines of
Ex. 1, Part (a), and having their centers on the line y = S.
4. Find the equations of the circles tangent to the lines of
Ex. 2 and passing through the point (1, 0).
 A SECOND CHAPTER ON LOCI 283

Given the triangle ABC with the sides

AB: 3a; 4- 41/- 3 = 0,

BC: SX-4.1J- 3 = 0,
CA: 12x-5y-\-15 = 0.
5. Prove that the bisectors of the interior angles of the
triangle meet in a point. Pind its coordinates. Ans. ( — 3^, 0).
6. Pind the equation of the circle inscribed in the triangle.
Ans. 12 1 {x'- + 2/') + 88 - 65 = 0.
a;

7. Show that the bisector of the interior angle at the vertex

A and the bisectors of the exterior angles at the vertices B and
C meet in a point. Pind its coordinates. — 5).Aiis. (1,

8. Pind the equation of the circle tangent to BC, and to AB

and AC produced. Ans. — 2 x + 10 y -^ 10 = 0.
x"- -\- y"-

9. How many circles are there tangent to three lines?

Draw a figure showing these circles.
10. Given the triangle with vertices A, B, and C in the three
points (1, 0), (—2, 4), and (— 5, — 8). Prove analytically that
the bisector of the interior angle at A divides the side BC into
segments proportional to AB and AC.
See also Exs. 26-30 at the end of the chapter.

EXERCISES ON CHAPTER XIII

Let AA' be a fixed diameter of a circle and E a point

1.

tracing the circle. Pind the locus of the point of intersection

of A'R and the line through A perpendicular to the tangent at
B.
2. Pind the locus of the point of intersection of the nor-
mals to an ellipse and to the auxiliary circle at corresponding
points. Take the eccentric angle (Ch. VII, § 10) as the aux-
iliary variable.

3. The circle x- + y~ = a- cuts the axis of 2/ in A: (0, a).

A point S traces the tangent at A and the second tangent from
S touches the circle in JR. Pind the locus of the point of in-
tersection of the altitudes of the triangle ARS.
284 ANALYTIC GEOMETRY

Suggestion. Take the abscissa of S and the coordinates of

R as auxiliary variables, and use the fact that OS is perpen-
dicular to AE.
4. Let AA' be a fixed diameter of a circle and R a point
tracing the circle. Find the locus of the point of intersection
of AR and the line joining A' to the point of intersection of
the tangents at A and R.
5. The normal to a hyperbola at a variable pointR meets
the transverse axis in JV. Determine the locus of the mid-
point of RN.
6. Find the locus of the point of intersection of the line

drawn through one focus of an ellipse perpendicular to a

variable tangent and the line drawn through the point of
tangency parallel to the transverse axis.

Ans. An ellipse, center in the focus chosen, with axes

having the same directions as those of the given ellipse.

7. Find the locus of the point of intersection of the line

drawn through one vertex of a hyperbola perpendicular to a
variable tangent and the line drawn through the point of tan-
gency parallel to the transverse axis.

8. Find the locus of the point of intersection of the line

drawn through the focus parabola perpendicular to a
of a
variable tangent and the line joining the vertex with the
point of tangency.
Ans. An ellipse, whose minor axis is the line-segment join-
ing the vertex of the parabola to the focus.

9. Two lines, passing through the points A and B respec-

tively, are originally in coincidence along AB. They are
made to rotate in the same direction about A and B respec-
tively, the first twice as fast as the second. What is the locus
of their point of intersection ?

10. Find the locus of the center of a circle which touches

one of two perpendicular lines and intercepts a segment of
constant length on the other.
 A SECOND CHAPTER ON LOCI 285

11. Find the locus of the point of intersection of the line

drawn through the vertex of the parabola 2/2 =z 2 mo; perpen-
dicular to a variable tangent and the line drawn through the
point of tangency perpendicular to the axis.
Ans. The semi-cubical parabola, my'^=2x^.
12. A vertex of a quadrilateral and the directions of the
sides through are fixed. The two angles adjacent to are
right angles and the diagonal joining their vertices has a fixed
direction. Find the locus of the fourth vertex.
Ans. Straight line through 0, perpendicular to the line
through which makes an angle with the fixed direction equal
to the sum of the two angles which the sides through make
with the fixed direction.
13. A parallelogram has sides of constant length a and b
and has one vertex fixed at a point 0. It opens and closes
so that the two sides through are always equally inclined
to a fixed line through 0. Taking the angle which these sides
make with the fixed line as auxiliary variable, find the locus
of the vertex opposite to 0.

14. Each of two straight lines moves always parallel to

the product of the distances of the lines from a
itself so that
fixed point O is constant. Find the locus of their point of
intersection, taking the axes so that is the origin and the
directions of the two lines are equally inclined to the axis of x.
Ans. Two conjugate hyperbolas, center at O, with asym-
ptotes parallel to the fixed directions.

15. Find the locus of the center of a circle which passes

through a fixed' point on one of two perpendicular lines and
intercepts a segment of constant length on the other.

16. Find the locus of points from which it is possible to

draw two perpendicular normals to a parabola.
17. Find the locus of the point of intersection of the tangents
to an ellipse at points subtending a right angle at the center.
18. The preceding problem for the hyperbola.
286 ANALYTIC GEOMETRY

19. Determine the locus of the mid-point of a variable

chord of an ellipse drawn from a vertex.
20. Find the locus of the mid-point of a variable chord X)f

a parabola which subtends a right angle at the vertex.

21. A R traces an ellipse, of which A and A' are the
point
vertices. Find the locus of the point of intersection of the
lines drawn through A and A' perpendicular to AR and A'E
respectively.
Ans. An ellipse, similar to and with the same center as the
given ellipse, but with opposite transverse and conjugate axes.
The asymptotes of a hyperbola intercept the segment
22.
AB on a variable tangent. What is the locus of the point
dividing AB in a given ratio, mi m2 ? :

Ans. A similar hyperbola, with the same transverse and

conjugate axes.

Exercises 23-25. Determine the equations of the desired loci

by use of rectangular coordinates. To identify the locus from
its equation introduce polar coordinates.
23. Find the locus of the point of intersection of a variable
tangent to' a rectangular hyperbola with the line through the

center perpendicular to the tangent. Ans. A lemniscate.

24. Find the locus of the point of intersection of a variable
-\- 2 ax = and the perpendicular
y''-
tangent to the circle x^^ -\-

to this tangent from the origin. Ans. A cardioid.

25. What is the locus of the mid-points of the chords of the
circle = which,
a;- -|- ?/' <^" when produced, are tangent to the
hyperbola 2 xy = c- ? Ans. A lemniscate.

26. Show that the line

a; cos 30° -f?/ sin 30° =5
is 5 units distant from the origin and that the perpendicular

from the origin to it makes with the positive axis of x an angle

of 30°. Prove that the distance of the point {xq, ?/o) from the
line is
- (ajo cos 30° + 2/0 sin 30° - 5),
 A SECOND CHAPTER ON LOCI 287

if (iCo, 2/0) is on the same side of the line as the origin, and is

iCo cos 30° + 2/0 sin 30° -5,

if (xq, yo) is on the opposite side of the line from the origin.

27. State and prove for the line

(1) iccosc^ H- 2/sin<^ =p, P^O

the results corresponding to those given in the preceding exer-
cise for the particular line for which (f>
= 30°, p = 5. Prove
that the equation of every line can be written in the form (1).

28. Two lines, with their equations in the form (1), are
given. Let « = 0, ^= be the abridged notation (Ch. IX, § 3)
for these equations. Prove that the bisectors of the angles
between the two lines are given by the equations w — /? =
and a-\- 13 = 0. Show that, if neither line goes through the
origin, the bisector a — (3 = passes through that opening
between the lines in which the origin lies,
29. The equations of the sides of a triangle, given in the
form (1), are ct = 0, ^ = 0, and y = 0. Assuming that the
origin lies within the triangle, find the equations of the bisec-
tors of the interior angles and prove that they meet in a point.

Prove that the bisectors of two exterior angles of the

30.
triangle of the preceding exercise and the bisector of the
interior angle at the third vertex meet in a point.
 CHAPTER XIV
DIAMETERS. POLES AND POLARS

1. Diameters of an Ellipse. By the axes of an ellipse we

may mean either the transverse and conjugate axes, indefinite
straight lines, or the major and minor axes, the segments of
these lines intercepted by the ellipse cf. the dual definition,
;

Ch. VII, § 1.

By a diameter of an ellipse we may mean, also, one of two

things, either an indejinite straight line through the center of the
ellipse, or the segment of this line intercepted by the ellipse and ;

we agree to adopt this dual definition. The length of the seg-

ment is called the le7igth of the diameter ; its end points, the
extremities of the diameter.

Problem. What is the locus of the mid-points of a set of

parallel chords of an ellipse ? i
.

In the special case of a circle, the locus is a diameter, con-

sidered as a line-segment. This is true, also, for the general

ellipse. For, if the chords are parallel to an axis of the

ellipse, the theorem is geometrically
obvious ; if they are oblique to the
axes, as is generally the case, we
resort to an analytical proof.
Let the ellipse be .

(1) '

Fig. 1 a2 b\

and let X (=?t: 0) be the slope of the chords. Consider a variable

chord of slope X moving always parallel to itself. Its mo-
288
 DIAMETERS. POLES AND POLARS 289

tion we express analytically by taking jS, its intercept on the

axis of y, as auxiliary variable. The equation of the chord is,
then, .

(2) y=Xx-{-l3,
where \ is constant and (3 is variable.
The work now proceeds according to the method of
Ch. XIII, § 5. If in (1) we set for y its value as given by
(2), we obtain the equation

or (a2X2 -f 62)aj2 + 2 a^XfSx -\- a\f3'' - b^) = 0,

whose roots are the two points of intersection
abscissae of the
of the line (2) with the ellipse. Half the sum of these roots
is Xy the abscissa of the mid-point, P, of the chord. Hence,
by the formula, Ch. XIII, § 5, (1), for the sum of the roots of
a quadratic equation,

(3) X= ^^M_.

Since, moreover, P: (X, T) lies on the chord (2), we have

(4)
-^
T=XX-\-l3.
It remains to eliminate (3 from (3) and (4). Substituting
its value, as given by (4), into (3) and simplifying the result-
ing equation, we obtain
(o) b''X-\-a^\Y=0.
This is the equation of a line through the center of the
ellipse, that is, a diameter. It is clear geometrically, however,
that it is not the indefinite line which is the locus, but merely
the portioij of it lying within the ellipse. We have thus ob-
tained the following result.

Theorem 1. Tlie locus of the mid-points of a set of parallel

chords of the ellijyse (1) is a diameter, considered as a line-seg-
ment (exclusive of the end points). If the slope of the chords is

A.(^ 0), the slope X.' of the diameter is

(6) y= —^.
290 ANALYTIC GEOMETRY

EXERCISES
1. Find the locus of the mid-points of the chords of the
ellipse

which are inclined at an angle of 135° to the axis of x. First

draw an accurate figure, showing the chords and the locus then ;

solve the problem analytically, using the method, but not the
formulas, of the text.
2. Prove the converse of Theorem 1, namely, that every
diameter of the ellipse (1) bisects some set of parallel chords.
Show that, if X'{=^ 0) is the slope of the diameter, then the
chords which it bisects are of slope A., where

a?X'

3. Prove analytically that the tangent to an ellipse atan

extremity of a diameter is parallel to the chords which the
diameter bisects.
Suggestion. Let (iCj, 2/1) be the coordinates of the extremity
of the diameter and find, by using (6), the slope A of the
chords in terms of Xi and y^.

2. Conjugate Diameters of an Ellipse. Two mutually per-

pendicular diameters of a circle have the property that each
bisects the chords parallel to the other. The axes of an el-
lipse have same property. Are there other pairs of
this
diameters of the ellipse which have it? This question is
answered in the affirmative by the following theorem.
Theorem 2. If one diameter bisects the
chords parallel to a second, the second di-
ameter bisects the chords parallel to the first.

The two diameters stand in a reciprocal

relationship ; each bisects the chords paral-
lel to the other. We call them a pair of
conjugate diameters, and say that each is conjugate to the
other.
 DIAMETERS. POLES AND POLARS 291

We now prove Theorem 2. Let the diameter D' bisect the

chords parallel to the diameter D to prove that D bisects
;

the chords parallel to D'.

Denote the slopes of D and D' by A and V. By hypothesis,
the diameter of slope X' bisects the chords of slope A conse- ;

quently, by Th. 1, § 1,

But then A =—
a-A'

This equation says that the diameter of slope A bisects the

chords of slope A' that is, D
bisects the chords parallel to D',
;

q. e. d.
Incidentally, we have also proved the following theorem.

Theorem 3. Tivo diameters D and D' of the ellijyse

(1) , 5: + ^'=!
are conjugate, if and only if tJiey are the axes or have slopes A
and A' related hy the equation

(2) AA'= - 62
a2

The symmetry of (2) in A and A' corresponds to the sym-

metry in the geometrical relationship of D and D'.
To each diameter D there
corresponds a conjugate diameter
— the diameter parallel to the chords which bisects. There D
are, then, infinitely ma.ny pairs of conjugate diameters. Since,
by the product of the slopes of any pair, other than the
(2),
axes, is negative, the two diameters of the pair pass through
different quadrants.
The axes are the only mutually perpendicular pair, unless
the ellipse becomes a circle. For, if any other pair were per-
pendicular, the product, AA', of their slopes would be — 1, and
this is impossible, according to (2), unless h- = a'^\ but then
the ellipse becomes a circle.
292 ANALYTIC GEOMETRY

We now find the conjugate diameters which are equally

inclined to the axes. It is evident, geometrically, that these
will also be the conjugate diameters of equal lengths. If they
exist, their slopes must be equal except for sign : A' = — X.
Hence, by (2),

and
a2 a

We see, then, that there is a single pair of conjugate diameters

ivhich are equally inclined to the axes, or have equal lengths.
They are the diagonals of the rectangle circumscribed about
the ellipse (Fig. 3). We denote
them by Di and Di'.
Now let a diameter D, starting
from coincidence with the trans-
verse axis AA', rotate about
into coincidence with Z>i ; then
the conjugate diameter, D', starts
from coincidence with the con-
FjG. 3
jugate axis BB', and rotates into
coincidence with D/. But D' rotates more quickly than D,*
so that the angle from D to D% at first 90°, becomes obtuse
and steadily increases. When D continues to rotate from Z>i
to BB', then D' rotates from Z>/ to AA' but now D' rotates;

less quickly than D, so that the angle from D to D' decreases

and becomes again 90° in the final position.

EXERCISES
1. Draw accurately an ellipse whose axes are 10 cm. and
7 cm. Construct the axes and the pair of conjugate diameters
equally inclined to the axes. Then draw the diameters in-

* Since the slope of A is -

a
< 1, ^ ^' OA < 45^ and X B' OA' > 45° ;

hence D has a smaller angle through which to rotate than D'. Con-
sequently, it is to be expected that D' will rotate more quickly than D.
A proof of the fact may easily be given later, when the student studies
.the Calculus.
 DIAMETERS. POLES AND POLARS 293

clined at angles of 10°, 20°, 30°, 40°, 50°, 60°, 70°, and 80° to the
transverse axis. For each of these diameters compute the
slope,and the angle of inclination, of the conjugate diameter
and construct it. Find the angles between the successive
diameters of this new setj and also the angles between the
successive pairs of conjugate diameters. Mark clearly the
pairs, and study the results and the figure in light of the text.

2. Prove that, if one of a pair of conjugate diameters of

an ellipse has the slope e or — e, where e is the eccentricity,
the other joins two extremities of the latera recta.
3. If the equal conjugate diameters of an ellipse form with
one another an angle of 60°, what is the eccentricity of the
ellipse ?

4. The axes of an ellipse are the axes of coordinates and

the slopes of two conjugate diameters are -| and —f . What is

the eccentricity ?

5. The same, if the slopes of two conjugate diameters are

— -| and J.

6. Prove that the line joining a focus to the point of inter-

section of the corresponding directrix and a diameter is per-
pendicular to the conjugate diameter.

3. Diameters of a Hyperbola. A diameter of a hyperbola is

defined in the same way as a diameter of an ellipse, § 1.

Certain diameters of a hyperbola, however, do not meet the

curve. Special definitions of the length and extremities of such
a diameter must, then, be adopted. These we shall consider
later.

The locus of the mid-points of a set of parallel chords of

slope A (^ 0) of the hyperbola

(1) ^-g=i
can be found by the method of § 1. It is, however, unneces-
sary to repeat the work there given. For, this work becomes
294 ANALYTIC GEOMETRY

valid immediately for the hyperbola (1) if, in it, we replace

h"^ by — h^. It follows, then, that the locus now required is

the diameter

(2) - 62a; + a'^xy = or y^x - a^Xy = 0.

The locus consists of all the points of this diameter only if

the given chords connect points of opposite branches of the

Fig. 4 Fig. 5

hyperbola (Fig. 4). If the chords connect points on the same

branch (Fig. 5), the locus is merely the points of the diameter
which lie within the curve. The result can be stated as follows.
Theorem 4. Tlie locus of the mid-points of a set of parallel
chords of the hyperbola (1) is a diameter, or so much of a diame-
ter as lies ivithiji the curve. If the slope of the chords is X (=^ 0),
the slope X' of the diameter is

(3) ^' = Jx-

There are chords of an ellipse with any given direction.
This is not true, however, for a hyperbola. For, there are no
chords of a hyperbola parallel to an asymptote, since a line
parallel to an asymptote meets the curve in but one point.
Consequently, the slope, A., of the chords of Theorem 4 cannot
have either of the values, ± b/a.
EXERCISES
1. A set of parallel chords of the rectangular hyperbola
x^ — y- = 6

are inclined at an angle of 30° to the positive axis of x.

What is the inclination of the diameter which bisects them ?

 DIAMETERS. POLES AND POLARS 295

First draw an accurate figure, showing the chords and the

diameter ; then solve the problem analytically, without refer-
ence to the formulas of the text.
2. If a set of parallel chords has a slope nearly equal to
that of an asymptote S, then the diameter D bisecting the
chords has a slope nearly equal to that of S, and when the
chords approach a limiting position of parallelism to S, then
D approaches S as its limit. Draw a figure showing the rea-
sonableness of this theorem and then prove the theorem ana-
lytically by use of (3).

3. Prove the converse of Theorem 4, namely Every :

diameter of a hyperbola, not an asymptote, bisects some set

of parallel chords. Cf. § 1, Ex. 2.

4. Show that the mid-point of a chord of a hyperbola is also

the mid-point of the chord of the conjugate hyperbola which
lies on the same line, pence show that the mid-points of the
chords of a given slope lie on one and the same diameter,

whether the chords are chords of the given hyperbola or of

its conjugate.
6. Prove Ex. 3, § 1, for a hyperbola.

4. Conjugate Diameters of a Hyperbola. Let the diameter,

D', of slope X', bisect the chords of the hyperbola

(1) ^-t=l
which are parallel to the diameter D, of slope \(=^ 0). Then,
by Th. 4, § 3,
V=^ or AA'=^^

Since these equations are symmetric in \ and A.', it follows that

the diameter D bisects the chords parallel to D'.
Thus Theorem 2, § 2, is established for the hyperbola, and
the two diameters D and D' are, in the sense of that theorem,
conjugate diameters ; each bisects the chords parallel to the
other.
296 ANALYTIC GEOMETRY

We have also proved, incidentally, the following theorem.

Theorem 5. Two diameters, D and D', of the hyperbola (1)

are conjugate if and only if they are the axes or have slopes, \
and A', related by the equation

(2)
a?

There are infinitely many pairs of conjugate diameters, as

in the case of the ellipse. But here the two diameters of a
pair, not the axes, pass through the same quadrants, since the
product, AA', of their slopes is positive.
The value, b~/a^, of this product is the square of the slope
of an asymptote. The slope of an asymptote, therefore, is a
mean proportional between the slopes
of any two conjugate diameters, not
the axes. Consequently, two such
conjugate diameters, D and D', are
always separated by the asymptote S
which lies in the same quadrants with
Fig. 6 them, and the nearer D lies to S on
the one side, the nearer Z>' will lie to
S on the other side. If D approaches yiS as a limiting position,
then so will D\ Thus, an asymptote is often spoken of as a
self-con jugate diameter actually, however, it has no conju-
;

gate, since, as we ha,ve seen, there are no chords parallel to it.

It is now clear that if a diameter, Z), starting from coinci-
dence with the transverse axis, rotates in one direction about
O into coincidence with an asymptote, then the conjugate
diameter, D', starting from the conjugate axis, will rotate in
the opjyosite direction about into coincidence with the same
asymptote.

Conjugate Hyperbolas. Consider now the hyperbola.

(3)
* 2/ _ -I

a^ 62

conjugate to the hyperbola (1). Since the two hyperbolas

have the same center, they have the same diameters, con-
 DIAMETERS. POLES AND POLARS 297

sidered as iudefiuite straight lines. Moreover, they have the

same pairs of conjugate diameters. For, the chords of (3) of a
given slope and the chords of (1) of the same slope are bisected
by one and the same diameter, Ex. 4, § 3.
We see now a suitable definition for the extremities of a
diameter which does not meet the given hyperbola. They
shall be the points in which the diameter meets the conjugate
hyperbola (Fig. 7), and the distance between these points shall

be the length of the diameter.*

Conjugate Diameters of a Rectangular Hyperbola. special A

ellipse, all of whose conjugate diameters are mutually perpen-
dicular, is the circle. There is no special
hyperbola with this property, since two
conjugate diameters of a hyperbola, other
than the axes, always pass through the
same quadrants. For this reason, too,
there are no conjugate diameters equally
inclined to the axes.
There may, however, be conjugate di- Fig. 7
ametere,each of which has the same
inclination to one axis as the other has to the other axis. The
product, A.A', of the slopes of two such diameters is 1 ; hence,
by (2), such diameters exist only if

or 0? = &2,
a2

that is, only if the hyperbola is rectangular. In this case

XA' = and every pair of conjugate diameters are in the re-
1,

quired relation. Consequently, the two diameters are equally

inclined to the asymptotes, inasmuch as the asymptotes are
now the bisectors of the angles between the axes. They are
also equal in length, as considerations of symmetry immedi-
ately show (Fig. 7). We have thus proved the following
theorem.
* An asymptote, considered as a diameter, we shall not think of as
having length or extremities.
298 ANALYTIC GEOMETRY

Theorem 6. Two conjugate diameters of a rectangular hyper-

bola are ahvays equally inclined to the asymptotes and always
equal in length.
shown that the rectangular hyperbola is the only-
It can be
one with either of these properties. Cf. Ex. 6, below, and § 6,

Ex. 5. Thus the rectangular hyperbola plays a role among the

hyperbolas which is somewhat similar to that played by the
circle among the ellipses.

EXERCISES
1. Draw accurately the hyperbola for Avhich 2 a = 10 cm. and
26 =7 cm. Construct the axes, AA! and BB', the asymptote
S passing through the first quadrant, and the diameters i)i, Z)2,
D^ inclined at angles of 10°, 20°, 30° to the transverse axis.
Compute the slopes and angles of inclination of the conjugate
diameters, D/, A'? A'? ^^^ draw these diameters. Find the
angles between the successive diameters, BB\ Di, A'? A'j S,
and compare them with the corresponding angles between the
diameters, AA', A, A, A, 'S- Study the results and the fig-
ure in light of the text. ,

2. Prove Ex. 2, § 2, for the hyperbola.

3. Prove that the asymptotes of the hyperbola

are conjugate diameters of the ellipse

^
a2
V-^
^'
52
= 1
-^•

4. The axes of a hyperbola are the axes of coordinates, and

the slopes of two conjugate diameters are 2 and |. What is

the eccentricity of the hyperbola ? Two answers.
5. Prove Ex. 6, § 2, for the hyperbola.

6. Show that two conjugate diameters of a hyperbola are

never equally inclined to the asymptotes unless the hyperbola

is rectangular.
 DIAMETERS. POLES AND POLARS 299

5. Diameters of a Parabola. When one focus and the cor-

responding directrix of a central conic an ellipse or a hyper- —
bola —
are held fast and the center is allowed to recede in-
definitely along the transverse axis, the limit of the conic is a
parabola and the limit of the diameters of the conic is a set
of lines parallel to the axis of the parabola.
Accordingly, by a diameter of a parabola we
shall mean any line in the direction of the axis
of the parabola.
If this definition is really in accord with
that of a diameter of a central conic, we
should find that the mid-points of a set of
parallel chords of a parabola lie on a line in
the direction of the axis. This is the case. yig 8
If the chords are perpendicular to the axis,
their mid-points evidently lie on the axis ; if the slope of the
chords is X{=^ 0), and the equation of the parabola is

(1) 2/^ = 2 mx,

the mid-points of the chords lie on the line

(2) y = ,•
f,
as may easily be shown.

EXERCISES
1. Establish the result embodied in formula (2).

2. What is the equation of the diameter of the parabola

2/2 -f 6 a; = 0,
which bisects the chords of slope ^?
3. Prove Ex. 3, § 1, for the parabola.
4. There are no conjugate diameters for a parabola. Why ?
6. Extremities and Lengths of Conjugate Diameters. Ellipse.
Let the coordinates of one extremity of a diameter D (not an
axis) of the ellipse

(1)
^+^' = 1
300 ANALYTIC GEOMETRY

be (iCi, 2/1). The slope of D is then A = yi/x^. From (2), § 2,

the slope A' of the conjugate diameter D' is

Consequently, the equation of D' can be written in the form

(2)

It follows that D' is parallel to the tangent to the ellipse at

(xi, yi). In other words, the tan-
( ay, hx a diam-
gents at the extremities of
eter are parallel to the conjugate
'^'
^
diameter.
The coordinates of the ex-
tremities of D' may be found by
solving equations (1) and (2)

Fig. 9
simultaneously. The solutions
are found to be
'

_ayj hxA «2/i hx^

a

We summarize the foregoing results in a theorem.

Theorem 7. If {x^, ?/i) is one extremity of a diameter D of

the ellipse (1), then (2) is the equation of the conjugate diameter

D', and one extremity (cc/, yi) of D' is

(3)
3. / _ «2/i
2/1
, _hxi
a
Suppose, now, that we denote the length of Z> by 2ai and
that of D' by 2 hi. Then it can be shown, by application of
(3) and the equation which states that {xi, 1/1) is on the ellipse,

that

(4) ai2 = 62 + 62^,^2 and W- = a^ - e^x^^

We have, then,
ai2 + ^^2 = a2 + 62 or (2 a^f + (2 61)2 = (2 af + (2 6)2.
This result we express as a theorem.
 DIAMETERS. POLES AND POLARS 301

Theorem 8. Tlie sum of the squares of the lengths of any tivo

conjugate diameters of an ellipse isconstant, and equals the sum
of the squares of the axes.
Hyperbola. Of two conjugate diameters, D and D', of the
conjugate hyperbolas
/v2 7/2

(5) («) -.-!;

a2 62
=1 ^ ^ '
a? Z>2

one meets the one hyperbola the other, the other hyperbola. ;

Suppose that D meets {oa) and D' meets (5 6), and that the
coordinates of an extremity of on {pa) are {x^, D y^), while
those of an extremity of D' on (5 5) are (a^/, ?//).
Then the equations of D' and are, respectively, D
(a)
x^'x y^y ^Q
(6) (^)
a2 62 '
a2 62 '

as is evident, from analogy to the

corresponding equation (2) in the
case of the ellipse. From (6) it
follows that the tangents, at the
extremities of a diameter, to the
hyperbola on which these extremi-
ties lie are parallel to the con-
jugate diameter.
Fig. 10
The coordinates of the extremi-
ties of D' can be found by solving equations (6 a) and (5 6)
simultaneously. The solutions are

^a.Vi hx;\ ^^^

and
/ ay^ bxA
\b '
a J \ b a
One of these extremities is (x/, ?//) let us say, the first ; one.
Then the values of x^ and ?// in terms of Xi and y^^, and vice
versa, are

(7) (a) x,' = y\

I
=— 6a;,
Vi
6 a 6 a
where the equations (6) are obtained by solving equations (a)
for Xi and 2/1.
302 ANALYTIC GEOMETRY

The student should note the symmetry of formulas (6) and

2/i) and {x^, y^).
(7) in (xi, The results embodied in these
formulas we state as a theorem.
Theorem 9. If (xi, yi) is one extremity of a diameter D
meeting the hyperbola (5 a), and {xi,yi') is a properly chosen one
of the extremities of the conjugate diameter D', then (6) are the
equations of D and D', and (7) give the relations between the two
extremities.

Let 2 ai be the length of D and 2 bi that of D'. It can be

shown that
f
a^s = -b'^ = e^x^'^ + a^
e2a?i2

^ ^
1 ^i^ = ^"^^ -a? = e^x{^ + 62.

Hence a^- - b^ = a^ - 6^

and we have the following theorem.

Theorem 10. The difference of the squares of the lengths of
any two conjugate diameters of a hyperbola is constant, and equals
the difference of the squares of the axes.

EXERCISES
Establish the following formulas.
1. Formulas (3). 3. Formulas (7).

2. Formulas (4). 4. Formulas (8).

5. Show that two conjugate diameters of a hyperbola are

never equal unless the hyperbola is rectangular and that in
this case they are always equal. ,

6. Prove that the product of the focal radii to any point of

an ellipse equals the square of half the diameter conjugate to
the diameter through the point.
7. State and prove the corresponding theorem for the
hyperbola.

7. Physical Meaning of Conjugate Diameters. Ellipse.

Consider a flat bar of iron, on the end ABCD of which a
circle is drawn. Let D and D' be any two mutually perpen-
 DIAMETERS. POLES AND POLARS 303

dicular,aud therefore conjugate, diameters of the circle.

Imagine that the bar is subjected to heavy pressure. Then
the lengths of all lines parallel to
AB and CD will be shortened in the
same and the lengths of lines
ratio,
parallel to AD and BC will all be
lengthened in the same ratio the ; jt^q^ h
two ratios will not, however, be
equal.* The circle will thereby be 1 I i 1

carried over into an ellipse, and the

diameters D and D' will become
conjugate diameters of this ellipse. t T t t t
f f T t t
A proof of these facts will be given pj^^ j^2

shortly.
The student can perform a suggestive experiment by taking
an ordinary four-sided eraser, dra.wing a circle and the diame-
ters D, D' on one of the broader faces of it, and then pinching
the eraser in a vise. The circle will go over into an oval that
looks like an ellipse, and D and D' will remain sensibly straight
lines.
If the vise is set too hard, the bulging will be considerable.
But imagine the ends of the eraser cut off square
and the
eraser then fitted snugly into a tube orchamber of rectangular
cross-section, with the broader faces and the ends in contact
with the walls of the chamber. Let the chamber be closed at
one end by a rigid, plane diaphragm, against which the eraser
is to be pressed.
now, a plunger, which just fits the chamber, is introduced
If,

and pressed down, the deforjnation will be much like that de-
scribed in the opening paragraph; the circle will become a
true ellipse, and D, D', remaining straight lines, will become

* Near the ends AB and CD of the cross-section these statements will

be only approximately ti'ue, since there will be a slight bulging
and, in- ;

deed, there will also be a slight bulging of the ends themselves. But near
the middle of the cross-section the deformation will be, to a high degree
of approximation, as described.
304 ANALYTIC GEOMETRY

conjugate diameters of the ellipse. But there will be one es-

sential difference, in that in the first case lines parallel to AD

are lengthened, whereas in this second case they remain un-
changed cf. rigs. 11 and 12, which have been drawn for the
;

second case, rather than for the first. The first case can, how-
ever, be reduced geometrically to the second if, after the def-
ormation has been made, the new figure is reduced in scale,
so that lines parallel to AD
again assume their original
lengths.
We shall confine ourselves to the second case. The defor-
mation of the plane of the circle may, in this case, be called a

compression in one direction or a sim2:)le compressioyi. All line-

segments in the direction of compression are shortened in the
same ratio, the ratio of compression. All line-segments in the
perpendicular direction remain the same in length they are ;

all moved parallel to themselves, with the exception of one

which remains In the case described this one rests

fixed.

against the diaphragm, either along AD or BC. If, however,

the diaphragm is replaced by a second plunger, the fixed line
might be AD
or BC or any parallel line such as EF, depending
on the manner in which the pressures on the two plungers are
applied. This line, perpendicular to the direction of compres-
sion and having all its points fixed under the compression, we
shall call the central line.
In studying the a compression let us take the cen-
effects of
tral line as the axis of x and the ratio of compression as Z;
/ is a positive constant < 1. We prove first that the compres-
sion carries a straight line L into a
straight line L. This is obvious if L
is parallel to either axis. If L is any
other line, the similar triangles in
Fig. 13 show that it goes over into a
line L, and that if L is of slope X, L is
of slope IX.
Next, consider an arbitrary circle, with center (Fig. 14).
The diameter A' A of the circle which is parallel to the central
 DIAMETERS. POLES AND POLARS 305

line, the axis of x, goes over into a parallel line-segment, A' A,

of equal length, and the mid-point, 0, of A'A goes over
into the mid-point, 0', of A'A.
y.Y,y
Let the circle be referred to
axes (x, y) with the origin at O,
the axis of x lying along A'A.
Its equation will be

(1) 2? + y^ = a\
Let the curve into which the circle
is deformed be referred to axes
(X, T) with the origin at 0', the
axis of X
lying along A'A. Then
any point {x, y) on the circle goes over into a point (X, Y)
such that
x Y=Vy,
-
or x=X, Y
=— '
V

It follows, then, that the circle (1) is transformed into the

curve
X^ + = a',
-f
or
X'2 Y2
(2) b = la.
a2 &^

Thus the circle is seen to be carried into an ellipse.

It remains to prove that the lines D and D', into which two
conjugate diameters D and D' of the circle are carried by the
compression, are conjugate diameters of the ellipse. If the
angle <^ isshown in Fig. 14, the slopes of D and D' are
as
tan <^ and tan((^ -f- 90°)= — cot <^. Hence the slopes of D
and D' are
\ = I tan (f)
and X' =— I cot <^.

Then kX' =-l^ tan cf>cotcf>=- P,

or, since, by (2), 1 = h/a.

XX' =-
a'
306 ANALYTIC GEOMETRY

Consequently, D and D', according to Th. 3, § 2, are conjugate

diameters of the ellipse, q. e. d.

If the center of the circle lies on the central line (Fig. 15),

the circle is the auxiliary circle of the ellipse (Ch. VII, 10),
and the angle <^ is the eccentric
angle for the extremity P of the
diameter D. The eccentric angle
for the extremity P' of the con-
jugate diameter D' is, clearly,

<^-f90°, or <^ + ?i90°, where n is

an odd number. Consequently, we

have proved the following theorem.
Theorem 11. The eccentric angles

for two points of an ellipse ichich are

Fig. 15
extremities of two conjugate diame-
ters differ by 90°, or by an odd multiple of 90°.
The theorem is essentially the geometrical equivalent of the
physical property of conjugate diameters which we have been
discussing. It furnishes a method of constructing rapidly as
many pairs of conjugate diameters of an ellipse as may 'be
desired.
The parametric representation of the ellipse can be used to
great advantage throughout the study of conjugate diameters.
The extremity P of the diameter D (Fig. 15) has, by Ch. VII,
§ 10, the coordinates

(3) X = a cos <^, y = b sin <^.

Then the extremit}^ P' of the conjugate diameter D' has, by

Th. 11, the coordinates

(4) x' = — a sin <^, y' = b cos ^.

Hence we obtain, for the squares of the half-lengths of D and
= a2 cos2 + 62 sin2
a^ <^ b^- = a- sin2
<^, <f)-\-b''- cos2 <^.

Therefore, a{^ + b^^ = a2 + b""^,

and we have a simple proof of Th. 8, § 6.

 DIAMETERS. POLES AND POLARS 307

Hyperbola. Consider, now, a set of four steel girders A, B,

C, and form of a square, and a cross girder E through
D in the
the center of the square parallel to A and C. Suppose that
tie-rods are spanned into the frame along the diagonals of
the square and along pairs of lines making equal angles with
the diagonals. These pairs of tie-rods,
then, lie along conjugate diameters of a
rectangular hyperbola, of which the
diagonals of the square are the asymp-
totes (Th. 6, § 4).
Suppose that the girder is firmly E
set in masonry, so that it is immovable,
and suppose that equal tensions are
exerted on the girders A and C as
shown. Then the square is elongated into a rectangle, except
for a slight bulging the diagonal rods come to lie along the
;

diagonals of the rectangle and the other pairs of tie-rods take

on the positions of pairs of lines which are conjugate di-
ameters in a hyperbola having the diagonals of the rectangle
as asymptotes, as we shall presently
show.
In this case we speak of an elongation
in one direction or a simple elongation.
The line of the girder E is the central
line of the elongation, and the ratio
Z(>1), in which all distances perpen-
dicular to E are stretched, is the ratio of
elongation.
Fig. 17 Let us take the central line as axis of x.
Since the slope of the diagonal S of the
square is 1, the slope of the diagonal S of the rectangle is I.

If the angle <^i is as shown, Fig. 16, the slopes of the two
lines D and D' making equal angles with S are tan <^j and
tan (90 —
<^i) = cot <^i. Hence the slopes, X and X', of the lines
D and D', into which D and D' are carried, are
X = I tan </>! and X' = 1 cot <^i.
 :

308 ANALYTIC GEOMETRY

But then W =1%

and, according to Th. 5, § 4, the lines D and D' are conjugate
diameters of a hyperbola, of which the diagonals of the rec-
tangle are the asymptotes. The ratio b/a of the axes of the
hyperbola equals the ratio of elongation, I

Finally, let us show that the elongation carries the rectangu-

lar hyperbola

(5)
^_ _ 2/^ 1

a"- a'
into the hyperbola

(6)
__ II— 1
a' 62

The two hyperbolas have the same

same eccentric
auxiliary circle, and the
angle, <^i, for points, P:(a;i,
y-^) and
Fig. 18
P : (iCj, 2/i)) with the same abscissa.
Hence, according to the method of
parametric representation of a hyperbola (Ch. VIII, § 9), the
coordinates of P and P are

(5a) •
a?! = a sec <^i, yi = a tan <^i ;

(6 a) jCi = a sec <^i. yi = b tan <^i.

Therefore yi = -yi,
a
or, since b/a = 1, yi = Wi- '

Hence the elongation does carry the hyperbola (5) into the hy-
perbola (6).
Let P, with coordinates (6 a), be an extremity of the diam-
eter D. Then the coordinates of an extremity P' of the conju-
gate diameter D' are, by (7), § 6,

(7) x' z= ^=b

a tan d)i, y^' = ^^ = b sec
a
<^i.
 : :

DIAMETERS. POLES AND POLARS 309

Here again, then, the use of the eccentric angle gives sym-
metry to the results.
One-Dimensional Strains. In the case of the ellipse we
might equally well have subjected the given circle to an elon-
gation, and in the case of the hyperbola we might have com-
pressed the given equilateral hyperbola, instead of elongating it.

Compression and elongation in one direction are but two types

of a single kind of deformation, known as a one-dimensional
strain. If the coefficient
of the strain is greater than unity,
I

the strain is an elongation on the other hand, if Z

; 1, the <
strain is a compression.

EXERCISES
1. Repeat Ex. 1 of § 2, drawing the auxiliary circle and
constructing the diameters conjugate to the given diameters
by application of Theorem 11.
2. Draw in pencil the asymptotes and a number of pairs of
conjugate diameters, including the axes, of a rectangular
hyperbola. Construct in ink the lines into which the given
lines are carried by the compression of ratio
J which has
an axis of the hyperbola as central line. What does the re-
sulting figure represent ?

3. Prove Th. 10, § 6, by means of formulas (6 a) and (7) of

the present paragraph.

8. Harmonic Division. Let P^P^ be a line-segment, and let

Qi be one of its points. Then Qi divides PiPg internally in a
certain ratio, /x (Ch. I, § 6)

On PxPi produced construct the point Q2 which divides P1P2

externally in the same ratio

C2A
 : : — —

310 ANALYTIC GEOMETRY

The points Qi and Q^ are said to divide the segment P1P2

harmonically they divide P1P2 internally and externally in
;

the same ratio

(V\ ^iQi _ Q2P1 .

Qif*2 Q2P2

Let us start with and trace the changes in Qi and

jx as given
Q2 a-s fx varies. and Q^ coincide in P^. As fx in-
If /x = 0, Q^
creases from to 1, Qi moves to the right from Pj to the mid-
point M
of P1P2, and Q2 moves to the left and recedes indefi-
nitely. If /x = 1, Qi is at M\ but Q2 ^^^ disappeared. Tlius
there is midpoint of a segment, divides the
no point which, ivith the

segment harmonically. As increases from 1 without limit, Qi ju,

proceeds from M
toward P2 as its limit, and Q^ appears again
from the extreme right, continually moving in and approach-
ing P2 as its limit.

The proportion (1) may be written in the form : -

(2)
PiQi _ Q,P2
Q2P1 Q2P2
this new proportion says that Pj and P2 divide the segment
Q1Q2 harmonically. Thus we have the following theorem.
Theorem If the points Qi and Q2 divide the line-segment
1.

P1P2 harmonically, then, recip)rocally, the points Pi and P^ divide

the line-segment Q1Q2 harmonically.
In other words, the relationship between the two pairs of
points is symmetric.

Suppose that Pi and P2 have the coordinates (xi, y^ and

{x2, 2/2)-Then the coordinates (cc/, ?//) and (x2, 2/2O of Qi and
Q2 are given by formulas (1) and (2) of § 6, Ch. I. If, in
each of these formulas, we divide the numerator and denomi-
nator by 7712 a^d then set mi/m2 = /a, we obtain, as the desired
coordinates
_ — -f fxxo ,,f_ yi +
—
n
Vi
.
•
,

Xi
,
— 0^1
2/1— f^y2 .
,

1 H- /X 1 -f /A

_ ^1 — /^^2 „/_fcii^.
o-
^2' ^
^2
/
—^ l—fX
J
——
2/2 :;

1—fX
I
 : j

DIAMETERS. POLES AND POLARS 311

EXERCISES
1. Four points on the axis of x have, respec-
Pi, P2, Qi, Q2
tively, the abscissas 3, 8, 5, — 7. Show that
Qi, Q2 divide P1P2
harmonically, and find the common ratio /a of internal and ex-
ternal division. Find, also, the value of the ratio, fx', for the
division by Pj and P2 of the segment Q1Q2'
2. Find the point on the axis of x which, with the point
(—1, 0), divides harmonically the segment of the axis joining
the points (- 8, 0), (3, 0).

3. Exercise 1, for the four points Pi, P2, Qi, Q2 with the
respective coordinates (2, 3), (—1, 9), (1, 5), (5, —3).
.4. Find the point which, with the point (2, 1), divides har-
monically the line-segment joining the points (5, —2), (1, 2).

9. Polar of a Point. Consider the following locus problem.

The ellipse

x^
(1) +^=1
62

and the point Pi (xi, y^) are :

given. A line L is drawn through

Pi meeting the ellipse in Q^ and
Q2, and on L the point P: (X, Y) Fig. 20
is marked which, with Pi, divides

Q1Q2 harmonically. What is the locus of P, as L revolves

about Pi ?
Since Pi, P divide Q1Q2 harmonically, Q^, Q2 divide PiP
harmonically. Hence the coordinates (x/, ?//), (x2j 2/2O of Qi,

Q2 are

Qi:
r
iCi'
_Xi-{-jxX
= 2/i + /xF
1+/X 1 +/X

Q2: Xo
, _ Xi-,xX
—^
,
_
-—Vi- txT
i — fX
5 2/2
1 — [X.

As L rotates, the ratio /x varies ; it is, then, an auxiliary

variable expressing analytically the rotation of L.
312 ANALYTIC GEOMETRY

The coordinates of Qi and Q2 satisfy (1). Substituting them

in turn in (1) and clearing each of the resulting equations of
fractions, we have

To eliminate fx, we subtract the second equation from the first,

thus getting
4 fxb'~XiX + 4 /Aa2?/i r = 4 ^a^b^
or, finally, ^ ^
a^
+ b^
= 1.
The locus of P is, therefore, a straight line, or a portion of
a straight line.* This line isknown as the jjolar of the j^oint

Pi with respect to the ellij^se. Hence we may say :

TJie polar of the point {xi, yi) ivith respect to the ellipse (1) has
the equation

(2) a?+M=i.
This equation is identical in form with the equation of the
tangent to the ellipse at the point (a?i, y^, Ch. IX, § 2, (12).
But in the present problem [xi, 2/1) is, in general, not on the
curve, and then (2) represents a line which is not a tangent.
If, in particular, Pj : {xi, y^) is on the ellipse, then (2) does
represent the tangent at Pi. Accordingly, we should like to
say The polar of a point on the ellipse is the tangent at the
:

point. Now there is trouble, geometrically, when P^ is on the

ellipse. For then Qi or Qo coincides with Pj, and P coincides
with them, so that, actually, no polar is defined. Suppose,
however, that Pi is a point near to Pi, but not on the curve.
Then it can be shown ( Exs. 1, 2) that the limiting position
of the polar of Pi, when Pi approaches Pi as its limit, is the
tangent at Pi. Hence the above statement is substantiated,

* If Pi is inside the conic, the locus is the entire line, but if Pi is out-
side the conic, the locus consists of only those points of the line which
are inside the conic.
 DIAMETERS. POLES AND POLARS 313

not as a conclusion, but as a proper definition of the polar of

a point on the ellipse.
If Pi is always the mid-point of Q1Q2, and
at the origin, it is

so there is never a point P which,

with Pi, divides Q1Q2 har-
monically. Consequently,
the origin has no polar.
The foregoing discussion is valid for the hyperbola

(3) ^-^=1,
if in the equations we replace b"- by — 6-. Thus, the polar of
the point (xi, yi) with respect to the hyperbola (3) has the equation

(A\
^
Ei^ _ y^y = 1
^ 52 •
a2

The polar of a point on the hyperbola is defined as the tangent

at the point. The center of the hyperbola has no polar.
We can now state the following theorem.

Theoreivj, 2. Given a central conic, C. Every point in the

plane, except the center of C, has a polar ivith respect to G.

Let the student show that the polar of the point {xi, y^ with
respect to the parabola

(5) y'-^^mx
has the equation

(6) y,y=m{x-\-Xi).

If we define the polar of a point on the parabola as the tangent

at the point, equation (6) shows that there are no exceptions
in this case. Accordingly, we have the theorem :

Theorem 3. Every point in the j^lcine has a polar with re-

spect to a parabola.
From the definition of a polar it is evident that the polar of
a point internal to a conic does not cut the conic, and that the
polar of a point external to a conic does cut the conic. In the
intermediate case, when the point lies on the conic, the polar
is a tangent.
314 ANALYTIC GEOMETRY

EXERCISES

I. Show that the polar of a point P^ external to a conic is

the line Li drawn through the points of contact of the tangents

to the conic from Pj.
Suggestion. Prove that P (Fig. 21)
approaches K as its limit, when the line
jD, rotating about Pi, approaches T.
2. Prove that, if the point Pi of Ex. 1
approaches a point on the conic as its

Fig. 21 limit, then its polar, Li, will approach

the tangent at this point.
3. Establish formula (6).

In each of the following exercises, find the equation of the

polar of the given point with respect to the given conic and
draw a figure, showing the conic, point, and polar.

Conic Point
4. x>'-\-f- =% (0,2).
5. 3a;2 + 52/2:=i5^ (5^6),
6. x^-y''±= 16, (2, 1).

7. 2?/2_5a; = 0, (-3,4).

8. Prove that in any conic the polar of a focus is the cor-

responding directrix.
9. Prove that the polar of a point Pi with respect to a
circle, center at 0, is perpendicular to the line OPi-
10. Show, further, that the product of the distances of
from Pi and the polar of Pi is the square of the radius of the
circle.

II. On the basis of the results of Exs. 9, 10, discuss the

variation in position of the polar of a point P with respect to
a circle, (a) when P moves on a straight line through the
center of the circle ; (6) when P traces a circle, concentric
with the given circle.
 DIAMETERS. POLES AND POLARS 315

Find the equation of the polar of the point (xi, y{) with re-
spect to each of the following conies.
12. The hyperbola : xy = k.
13. The circle: (x - ay + (y — ^y = p^.
14. The conic : (1 — e2) — 2 mx + m2 = 0.
x''--\-y^-

10. Pole of a Line. If, with respect to a given conic, the

line L is the polar of the point P, the point P is known as the
pole of the line L.
Given a conic and a line L ; to find the pole, P, of L with
respect to the conic.
Let the conic be the ellipse,

(1)
^
2a;2-|-32/2 = 6,
and L, the line,

(2) . 4a;- 32/ -2 = 0.

If we denote the coordinates of P by {xi, y^, the polar of P
with respect to (1) is

(3) 2a;ia;+ 3^/12/ -6=0.

But the polar of P was given as the line (2). Equations (2)
and (3), then, represent the same line. Consequently, by
Ch. II, § 10, Th. 5,

2xi_ Syi _—6

Then iCi = 6, ^i =— 3
and so the point (6, — 3) is the pole of the line (2) with re-
spect to the ellipse (1).

We now Has every line a pole with

raise the question:
respect to a given conic ? Let us answer this question first
for the central conies. Equations (2) and (4) of § 9, which
represent the polars of a given point with respect to the central
conies (1) and (3) of § 9, are never satisfied by x = 0, y = 0,
no matter where the given point lies. Consequently, the polar
316 '
ANALYTIC GEOMETRY

of a point with respect to a central conic never passes through

the center of the conic. In other words, a diameter of a central
conic has no pole.
We proceed to show that every other line has a pole, giving
the proof in the case of the hyperbola

(4) t-t=:l.

Any line not a diameter of (4), that is, not passing through
the origin, can be represented by an equation of the form *

(5) Ax + By= 1.

If (iCi, 2/i) is the pole of this line, the line also has the equation
 DIAMETERS. POLES AND POLARS 317

Theorem 4. Given a conic C. Every line of the plane,

tvhich is not a diameter of C, has a pole with respect to C.
By comparing theorem with Theorems 2, 3, § 9, we
this
see that the lineswhich have no poles with respect to a conic
go through the point which has no polar, provided these lines
intersect.
EXERCISES
1. Give the proof of Theorem 4 for the parabola.
In each of the following exercises find the pole of the given
line with respect to the given conic.
Conic Line
2. a;2 + ?/2 = 8, 2rc-3?/-2 = 0. ^

3. 5x2-62/2-30 = 0, 4aj + 22/-7 = 0.

4. 3?/2-8a; = 0, 2 a; - 3 = 0.
5. 7aj2 + 22/2=14, 6a; + 5?/ -8 = 0.
6. Prove that the pole of any line through the focus of a
conic is a point on the corresponding directrix.

7. Given the circle x"- 4- y"- = a^. Prove that the pole, with
respect to this circle, of a line moving so that it is always
tangent to a concentric circle traces a second concentric circle.

Cf. Exs. 9-11, § 9.

11. Properties of Poles and Polars. The poles and polars *

discussed in this paragraph are all taken with reference to an
arbitrarily given conic. For the
sake of brevity mention of the
conic is, in general, suppressed.

Theorem If a point Pi lies

ba.
on the polar of a second point P2,
the7i. conversely, Po lies on the ^
polar
'
^' ^ Fig. 22
.
ofPi-
Let the polars of the points Pi and Pg be Li and i^2- Then
the theorem says that, if Pi lies on X2, P2 lies on Li. But this
* Only those points which have polars and those lines which have poles
are considered.
318 ANALYTIC GEOMETRY

is the same as saying that, if L^ goes through P^, L^ goes

through P25 or vice versa. This second form of the statement

we enunciate as a theorem.

Theorem 56. If a line Li goes through the pole of a second

line L2, then, conversely, L2 goes through the pole of Li.
Since the two theorems are equivalent in content, and differ
only in point of view, a proof of one also proves the other.
We choose to prove Theorem 5a, and to give the proof in the
case that the given conic is the hyperbola

a;2 _ ?/2 _ .

The proofs in the other two cases are similar.

Let Pi and P2 have the coordinates (xi, y^) and {x2, y<^.

Then Li and L2 have the equations

^_M = l and ^_M=i
The condition that Pi lies on Zg is

Ml _Ml=l '
a2 62

and the condition that P2 lies on Li is

f^i^_M2 = l
a2 62

But these two conditions are the same. Hence, if Pi lies on

on Xi, and conversely, q.
L2, then P2 lies e. d.

Suppose, now, that we join the points Pi and P2 of Fig. 22

by the line L. Since Pi lies on L, it follows, by Th. 5a, that
the pole, P, of L on Li. Similarly, since P2 lies on L, P
lies

lies also on ^o- Hence, P is the point of intersection of Li and

L2. Thus we have the theorem :

Theorem 6a. Tlie pole of the line joining tivo points, Pi and
P2, is the point of intersection of the polar s of Pi and P2.

Starting again, we bring the lines Li and L2 of Fig. 22 to

intersection in P. By Th. 56, since each of the lines Li and
 : : :

DIAMETERS. POLES AND POLARS 319

L2 goes through P, it follows that the polar, X, of P goes

through each of the poles, P^ and Pg? of Ly and L^. Conse-
quently, L is the line joining Pi and P2 and we have proved
Theorem 66
Theorem 65. The ])olar of the point of intersection of two
lines, Li and X2, is the line joining the poles of Li and L^,

By application of either Ths. ha, 5b or Ths. 6a, 6b, the

student can easily prove the following theorems.
Theorem 7a. If a number of points
all lie on a line, L, their polars all go
through a point, namely, the pole of L.

Theorem lb. If a number of lines cdl

go through a point, P, their poles all lie on
a line, namely, the jyolar of P.
Finally, take a line L which cuts the Fig. 23
given conic in two points, Pi and P2.
Since L is the line joining Pi and P2, the pole, P, of L is the
point of intersection of the polars of Pi
and P2 (Th. 6a), that is, of the tangents
to the conic atPi and Po. Thus we have
proved the theorem
Theorem 8a. TJie pole of a line inter-

Fig. 24
secting the given conic in tivo points. Pi
and P2, is the point of intersection of the
tangents to the conic at Pi and P-i.

Let the student prove the mate of this theorem, namely

Theorem 86. Tlie polar of a point external to the given

conic is the line joining the points of contact of the tangents to the
conic from the point.

Theorem 8a furnishes a means of constructing the pole of

a line which meets the given conic ; Theorem 86, a means of
constructing the polar of a point external to the conic.
To construct the pole, P, of a given line, L, which does not
meet the conic (Fig. 25), choose any two points, Pi and P2, on
 ;

320 ANALYTIC GEOMETRY

L, and construct their polars, L^ and L^.

Since L is the line joining Pj and P2, its
pole, P, is the point of intersection of Li
and Z/2.

The student should now establish the

analogous construction for the polar of
a given point which is internal to the
Fig. 25
conic*

EXERCISES
1. Prove Theorems 7a, 76.

2. Prove Theorem 86.

3. Show how to construct the polar of a given point which

is internal to the conic. Prove the validity of the construction.
4. On the basis of Theorem 86 develop in detail a method
for finding the equations of the tangents to a conic from an
external point.

By means of this method find the equations of the tangents

required in each of the following exercises of Ch. IX, § 7.

5. Exercise 5. 6. Exercise 6.

7. Exercise 9. 8. Exercise 12.

12. Relative Positions of Pole and Polar. Central Conies.

The following theorem is in-

structive concerning the rela-

tive positions of pole and polar

with regard to a central conic.

Theorem 9. Let the point Pi

and the line Li he pole and polar
in a central conic, center at ;

let D he the diameter through Fig. 26

* These methods are not very serviceable if accurate constructions are

desired, since they involve the construction, not only of the tangent at a
given point of the conic, but also of the tangents from an external point
of. § 13. They are, however, useful in rough work.
 DIAMETERS. POLES AND POLARS 321

Pi, D' the diameter conjugate to D, and P^ the point of inter-

section of Li with D. Then L^ is parallel to D' and the half-
length, d, of D is a mean proportional betiveen OPi and OF2:

(1) OP, ' OP. = d\

We will prove the second part of the theorem first. By the
definition of the polar of a point, P1P2 is divided harmonically
by the points, Q, and Q^, in which D meets the conic. Conse-
quently, by (1), § 8,

QiPi ' Q2P2= Q2P1 • P2Q1'

Expressing each of the four distances in terms of OP,, OP2,

and OQi = OQ2 = d, we have
(OPi - d)(0P2 + d) = (OP, + d){d - OP2).

On multiplying out and reducing, we obtain equation (1),

q. e. d.
We will give the proof of the first part of the theorem in
the case that the conic is the ellipse

— -1-^=1
a- W
Let the coordinates of P, be {x,, y,) and those of Qi, {x2, 2/2)-

Then the equations of ij and D' are, respectively, by § 9, (2),

and § 6, (2),

(2)
a^ 62
o"- ^2
a'
^ 52

Since P, and Q, are on a line witl^ 0, their coordinates are

proportional :

Xi:y-, = X2: 2/2,

and hence so are the left-hand sides of

equations (2). Consequently, Li and Z>'

are parallel, q. e. d.

The proof of the second part of the

theorem assumes that D meets the conic.
This is not true, however, if the conic is

a hyperbola and. Pi lies in an opening Fig. 27

322 ANALYTIC GEOMETRY

between the asymptotes not containing a branch of the hyper-

bola. In this case, too, the theorem is valid, but the points
Pi and P2, instead of being on the same side of 0, are on
opposite sides.*

Corollary. The points P^ and P^ are on the same or oppo-

site sides of 0, according as D meets or does not meet the conic.
Given, now, a point, P, and its polar, L, with respect to a
central conic. If P traces a diameter D, then L moves always
parallel to the conjugate diameter.
In the case that D intersects the conic, and P is an intersec-
tion, L is the tangent at P. If P then moves in along D
toward the center as its limit, L ceases to meet the conic, and
recedes indefinitely. On the other hand, if P moves out
along D, receding indefinitely, L moves in toward the center,
and approaches the diameter conjugate to D as its limit.
The case in which the conic is a hyperbola and D intersects
the conjugate hyperbola remains. If P is at one of the inter-
sections, L is the tangent to the conjugate hyperbola at the
other ; cf. Ex. 6. If P then moves in toward the center, L
moves away from it, and so forth, as before.

Parabola. Corresponding to Theorem 9, we have the fol-

lowing theorem, the proof of which is left to
the student.

Theorem 10. Let Pi and L^ he pole and

a parabola, and let the diameter
polar in
through P^ meet Li in P^ and the parabola in
Q. TJien L^ is parallel to the tangent at Q,
and Q is the mid-20oint 0/P1P2.
Fig. 28 Consequently, if a point P traces a diameter
* Let the student give an analytical proof of these facts and hence of
the corollary ; cf. Exs. 1, 2. There is no geometrical proof, analogous
to that of the text. The roles of Qi and Q2 in that proof cannot be
played here by the points in which D meets the conjugate hyperbola ;

these points do not divide P1P2 harmonically: Xi is the polar of Pi

with respect to the given hyperbola, and not with respect to its conjugate.
 DIAMETERS. POLES AND POLARS 323

Z) of a parabola, its polar L moves always parallel to the

tangent at the point in which D
meets the curve. If moves P
along D
in either direction, receding indefinitely, then L moves
in the opposite direction, and recedes indefinitely.

Theorems 9 and 10 furnish new methods for the construc-

tion of the polar of a given point or the pole of a given line.
These we shall consider in the next paragraph.

EXERCISES

1. Give an analytical proof of the second part of Theorem

9, in the case thatD meets the conic.
2. The same if the conic is a hyperbola and D does not meet it.
3. Theorem 9 is no longer valid if the conic is a hyperbola

and Pi on an asymptote. Prove that, in this case, Li is

lies

parallel to the asymptote, and that the product of the distances

OPi and OP2 is constant, where P2 is the point in which Li
intersects the other asymptote.

4. Prove Theorem 10.

5. A pair of conjugate hyperbolas and a point P are given.

Show that the polars of P with respect to the two hyperbolas
are parallel to, and equally distant from, the diameter conju-
gate to the diameter through P.

By applying Th. 9, the Corollary, and Th. 10, prove the fol-
lowing theorems.
6. Let (7 be a hyperbola, C
the conjugate hyperbola, and a D
diameter meeting C. Then the polar of an extremity of with D
respect to C is the tangent to C at the other extremity of D.
7. The polar of a point with respect to a central conic (not
a circle) is perpendicular to the line joining the point to the
center, if and only if the point is on an axis of the conic.

8. If a line is normal to a parabola at one extremity of the

latus rectum, its pole lies on the diameter passing through the
other extremity.
324 ANALYTIC GEOMETRY

13. Construction Problems. In problems of construction re-

lating to conies, diameters play an important role. To con-

struct a diameter of a conic, one has but to draw two parallel
chords and join their mid-points.

Center^ Axes, Foci. Consider first a central conic, drawn on

paper. Construct two diameters ; their point of intersection
will be the center, 0, of the conic. With as center, describe
a circle cutting the conic in four points ; the lines through
parallel to the sides of the rectangle determined by the four
points will be the axes.
If the conic is an ellipse, the lengths a and h are now known,
and c = Va- — &' can be constructed by means of a right tri-

angle ; cf. Ch. VII, Fig. 2. Thus the foci will be located.
If the conic is a hyperbola, only a is known. But then h
can be found by reversing the construction of Ch. VIII, § 9.
Hence the asymptotes and foci may be accurately constructed.
Let a parabola be given. Construct a diameter and two
chords perpendicular to it; the line joining the mid-points of
these chords is the axis of the parabola. The construction of
the focus we postpone until we have given that of a tangent.
Tangents. To construct the tangent to a central conic at a
point P, construct the center 0, and then draw OP and a chord
parallel to OP. Let K
be the mid-point of this chord. Then
the line through P parallel to OK is the tangent at P. Why ?
If the conic is a parabola, construct the
axis. Let K
be the foot of the perpendicular
dropped from P on the axis, and make OM
equal to OK
(Fig. 29). Then, by Ch. VI, § 3,
Ex. 8, MP
is the tangent at P. The focus can
now be constructed by use of the focal prop-
erty, namely, by constructing the focal radius

Fig 29
^ ^^ ^^ *^^ ^^^^ making ^ MPF = ^ TPS.
Of course, if the focus, or foci, of a conic
are given, the tangent at a point can be constructed by means
of the focal property of the conic.
 DIAMETERS. POLES AND POLARS 325

To construct the tangents to a conic from an external point

is a more difficult problem. We shall give presently a solu-
tion involving poles and polars and refer the student else-
where for one based on more elementary, though less elegant,
principles.*

Poles and Polars. Given an elementary construction for the

tangents from an external point, we can carry through ac-
curately the constructions of § 11 for the polar of a given point
and the pole of a given line.
We are more interested, however, in the constructions of
poles and polars, based on the theorems of § 12. We will
describe, for example, the construction by this method of the
polar of a given point Pi with respect to a central conic.
Draw the diameter I) through Pi (Fig. 26) and, by drawing a
chord parallel to D
and bisecting it, construct the diameter D'
conjugate to Z>. On a separate sheet construct the third pro-
portional to the length OPi and the half length, d, of D.-f
Lay off the resulting length from on D in the proper direc-
tion, according to the Corollary of Theorem 9, and through the
^oint thus reached draw the line parallel to D'. This line is

the polar of P^
The construction is the same whether P^ lies inside, on, or
outside the conic.

Tangents from an External Point. Let the point be P and

construct its polar L by the method just described. The lines
joining P to the points of intersection of L with the conic are,
by § 11, Th. 86, the required tangents.

* Cf., for example, Wentworth's Plane and Solid Geometry, Ed. of

1900, pp. 416, 435, 455.
t If D does not meet the conic (as may happen in the case of a hyper-
bola) and the conjugate hyperbola is not given, the length dl (Fig, 27)
is unknown, so that a separate construction for it is necessary. Here d
equals the 6i of the formula, ai"^ hi^ = a'^ —
b% of § 6, Th. 10. The —
lengths a, 6, and ai are known or can be constructed by methods already
given ;
the length k = Va^ — 62 is found by using a right triangle and,
finally, that of bi = Vai^ — A:2, in the same way.
326 ANALYTIC GEOMETRY

EXERCISES
1. Using a templet, draw an ellipse. Carry through in de-
tail thd constructions for (a) the center, (6) the axes, (c) the
foci. Devise a method for constructing the directrices.
2. The same problem for the hyperbola. Construct also
the asymptotes.
3. Construct the axis, a tangent, the focus, and the directrix
of a parabola.

4. Construct the tangent to a hyperbola at a given point by

use of the focal property. Use a templet to draw the hyper-
bola and consider that the foci are given.
5. The same for an ellipse.

6. Perform in detail the construction, based on Theorem 9,

§ 12, of the pole of a given line with respect to a central conic.

7. Carry through carefully the construction, based on

Theorem 10, § 12, of the polar of a given point with respect
to a parabola.

8. The same for the pole of a given line.

EXERCISES ON CHAPTER XIV

Diameters
1. Prove that two similar ellipses with the same center

and the same transverse axis have the same pairs of conjugate
diameters.
2. A line meets a hyperbola in the points Pj and Po and

meets the asymptotes in the points Qi and ^2- Prove that

the segments P1P2 and Q1Q2 have the same mid-points.
3. Using the result of Ex. 2, show that any two hyperbolas

with the same asymptotes have the same pairs of conjugate

diameters.
4. Prove that the line-segment joining two extremities of
conjugate diameters of a hyperbola is parallel to one asym-

ptote and is bisected by the other.

 DIAMETERS. POLES AND POLARS 327

5. The chords of an ellipse from a vertex to the extremi-

ties of the minor axis are parallel to a pair of conjugate
diameters. Prove this theorem.

6. Two chords connecting a point of a central conic with

the ends of a diameter are called supplemental chords. Show
that chords of this nature are always parallel to a pair of con-
jugate diameters.

7. Show that, if a parallelogram has its vertices on a cen-

tral conic, its center is at the center of the conic ; hence prove,
by Ex. 6, that the sides of the parallelogram are parallel to a
pair of conjugate diameters.

8. Prove that the angle which a diameter of an ellipse,

not an axis, subtends at a vertex is the supplement of the

angle which the conjugate diameter subtends at an extremity
of the minor axis.

9. A parallelogram is circumscribed about an ellipse by

drawing the tangents at the ends of a pair of conjugate diam-

eters. Prove that the area of this parallelogram is the same,
no matter what pair of conjugate diameters is chosen.
Suggestion. Compute the area of the triangle with one
diameter as base and an extremity of the other as vertex.
10. State and prove the corresponding theorem for the
hyperbola.
11. Show that in the case of the hyperbola the parallelo-
gram of the two preceding exercises always has its vertices on
the asymptotes.
12. Prove that the segment of a tangent to a hyperbola cut
off by the asymptotes is equal in length to the diameter paral-
lel to it.

13. Prove that the tangents to a central conic at the extrem-

ities of a chord meet on the diameter bisecting the chord.
14. Show that a line through a focus of a central conic per-
pendicular to a diameter meets the conjugate diameter on a
directrix.
328 ANALYTIC GEOMETRY

15. Prove that, if P and P' are extremities of a pair of con-

jugate diameters of a central conic, the normals at P and P'
and the line through the center perpendicular to PP meet in
a point.

Poles and Polars

16. Find the polar of a focus of a central conic with respect
to the auxiliary circle.

17. Prove, for a central conic, that the line-segment joining

any point to the intersection of the polar of the point with a
directrix subtends a right angle at the corresponding focus.

18. The same for a parabola.

19. Show, for a central conic, that any chord through a focus
is perpendicular to the line joining the focus to the pole of
the chord.
20. The same for a parabola.

21. Two rectangular hyperbolas are so situated that the

axes of one are the asymptotes of the other. Prove that the
polars of a point with respect to the two hyperbolas are always
perpendicular.
22. The perpendicular from a point P on the polar of P
with respect to a central conic meets the transverse axis in A
and the conjugate axis in B. Show that PA PB = b^ a^. : :

23. The segment of the axis of a parabola intercepted by

the polars of two points is equal to the projection on the axis
of the line-segment joining the two points.

Locus Problems
24. A line is drawn through the focus of a central conic per-
pendicular to a variable diameter. Find the locus of the point
in which it intersects the conjugate diameter.

25. A point moves so that its polar with respect to an ellipse

forms a triangle of constant area with the axes of the ellipse.
What is its locus ?
 DIAMETERS. POLES AND POLARS 329

Ans. A pair of conjugate rectangular hyperbolas with the

axes of the ellipse as asymptotes.
26. Find the locus of the poles, with respect to a central
conic, of the tangents to a circle whose center is the center of
the conic. *

27. Find the locus of the poles, with respect to the circle
a;2 _|_ 2/2 = (x2, of the tangents to the parabola ?/- = 2 mx.

28. Find the locus- of the poles, with respect to the parabola
2/2 = 2 mx, of the tangents to the parabola y^ = — 2 mx.
29. Find the locus of the mid-point of a chord of an ellipse,
if the pole of the chord traces the auxiliary circle.
30. The same for a hyperbola. *
 CHAPTER XV
TRANSFORMATIONS OF THE PLANE. STRAIN

1. Translations. Definition. By a translation of a plane

region S is meant a displacement of S whereby each point of
S is carried in a given (fixed) direc-
tion by one and the same given dis-
tance. Thus, when a window is
raised, a pane of glass in the window
experiences a translation.
Fig. 1
It is not important what particular
region S is considered. Indeed, it is usually desirable to
consider the whole unbounded plane as S. The essential
thing is the above law which connects the initial position of
an arbitrary point of S with its final position.

Analytic Bepresentation. Let P : {x, y) be an arbitrary

point of the plane, and let P' : {x', y') be the point into which
P is carried by the translation. Let a
y P-Sx',v'^
and h be respectively the projections of
the directed line-segment PP' on the
axes of X and y. Then PSx,v)

\ x' =X -\- a, o
(1) Fig. 2

These formulas are the same as those which represent a

transformation of coordinates, the new axes being parallel to
the old and having the same respective directions. But the
interpretation of the formulas is wholly different. There, the
point P remained unchanged. It had new coordinates as-
signed to it by referring it to a new set of axes. Here, the
330
 : : : :

TRANSFORMATIONS OF THE PLANE 331

axes do not change. It is the point P that changes. The

point P is picked up and set down in a new place, namely,
at P'.

Example 1. Represent analytically the translation whereby

the plane is carried in the direction
V
of the positive a.'-axis a distance of 2
units X
o 1 2
Solution : x' = x-{-2, i/ y-
Fia. 3
Example 2. Let the curve C
(2) y = x^ — X -{•

be carried in the direction of the negative axis of i/ a distance

of f units. What will be the equation of the new curve, C?
The formulas representing the translation are
=
x' X, y'= y- 1-
Hence X—X y = y' + h
and equation (2) goes over into

2/' + I = x'^ -x' + ^

or y' = x'^ x'.

Pjq 4 The new is evidently symmetric

curve, 0',
in the origin. But the shape of is the C
same as the shape of C Hence C is symmetric in the point
A :
(0, 4), which corresponds to the origin.

Example 3. A freight train is running northwest at the

rate of 30 miles an hour. If {x, y) are, at noon, the coordi-
nates of an arbitrary point of the floor of one of the platform
cars, referred to axes directed east and north respectively, de-
termine the coordinates (cc', i/') of the same point t hours
later.
Here, the components of PP'^ after one hour has elapsed,
are clearly

a = 30 cos 135° = - 15 V2, h = 30 sin 135° = 15 V2.

332 ANALYTIC GEOMETRY

After t hours they are

Hence
x' = x- 15 V2?, y'= y + 15V2i.

EXERCISES
1.Express analytically the translation which carries the
origin into the point (2, — 1), and hence show that the point

(—1, 2) is carried into (1, 1). Draw a figure showing what

happens to the unit circle, aj^ -f ?/2 == 1.
2. Apply the translation of Ex. 1 to the curve

y = 2x^ + ^x + ^.
3. Determine a translation which will carry the curve

2/ = 4a;2 — 8a;4-3
into a parabola whose equation is in a normal form.
4. An aeroplane is flying at the rate of 120 miles an hour
on a straight, horizontal course having a direction 30° south of
east. If {x, y) are, at a given instant, its coordinates, referred
to axes directed east and north respectively, determine its co-

ordinates (ic', y') after t minutes have elapsed.

Prove analytically {i.e. by means of the representation (1)

of the text) the following theorems.
5. A translation carries a straight line, in general, into a
parallel straight line. What are the exceptions ?
6. A translation carries a circle into a circle of the same
radius.

7. A translation carriestwo mutually perpendicular right

lines into two mutually perpendicular right lines.

2. Rotations.Let the plane be rotated about the origin

through an angle What will be the coordinates, {x\ y'), of
6.

the point P' into which a given point P, with the coordinates
(x, y), is carried ?
 :

TRANSFORMATIONS OF THE PLANE 333

The solution can be read off at sight from the figure. We

have:
X = 0M= OM^ ; y = MP = M^P'-,
x'= OM', y' = M'P'.

Now, Proj OP' = Proj 03fi+ Proj 3f^P',

and if we take the projections first along the axis of x, and

then along the axis of y, we obtain P:(x:y')
y
immediately the desired relations :

i = X cos 6—y sin 6,

x' ,P:(x.y)
(1) <
yy' = x sin 6 -\-y cos 6.
It is easy to solve these equa-
tions algebraically for x and y or ;
-'"O
1 M' M
the formulas for x and y, in terms
Fig. 5
of a;' and y', can be written down
directly by projecting the broken line OM'P' along OM^ and
perpendicularly to OM^:

\ x= x' COS y' sin 6, -\-

(2)
\y = — sin ^ + y' cos 0.
£c'

Example. It is clear geometrically that a circle with its

center at the originmust be carried over into itself by any of
the above rotations. Let us see what the analytic effect on
its equation is if such a rotation is performed.
The equation of the given circle is

a:2 + 2/2 = p2.

Replacing x and y by their values from (2), we have
(x' cos 0-\-y' sin Oy--\-( — x' sin B -\- y' cos &)"-= p^^
or
cos2|9 x''^ -f- 2 sin 6 cos 6 xy + sin2 B y" = P'-
H- sin2 e — 2 sin ^ cos 6 4- cos2 e

Hence n -.
x'^ + y' p^.,

and we get the same circle, as we should.

334 ANALYTIC GEOMETRY

EXERCISES
1. Write down directly from a figure the formulas which

represent a rotation of 90° about the origin, and verify the re-
sult by substituting ^ =
90° in (1).

2. Show that, if the curve xy=2 a"- is rotated about

the origin through an angle 6 = — 45°, its equation goes
over into the usual form of the equation of an equilateral
hyperbola.

3. Rotate the parabola y^ = 2 mx through — 90° about the

origin.

4. Prove analytically that, if an arbitrary straight line be

rotated about the origin through 90°, the new line will be per-
pendicular to the old one.

5. Prove that, if an arbitrary line be rotated about the

origin through the angle 0, the angle from this line to the new
line will be B.

3. Transformations of Similitude. Let the plane be stretched,

like an elastic membrane, uniformly in all directions away from
the origin. This tranformation
is evidently represented analyti-
cally by the equations :

where A; is a constant greater

Fig. 6 than unity. If k is positive, but
less than unity, the transforma-
tion represents a shrinking toward the origin. The stretchings
and shrinkings defined by (1) are known as transformations of
similitude.
These transformations, like the translations and the rota-
tions, preserve the shapes of all figures ; but, unlike those
transformations, they alter the sizes of figures.
 TRANSFORMATIONS OF THE PLANE 335

Example. The equilateral hyperbola

3.2 _ 2^2
_ ^2
is carried by (1), if k is taken equal to -:

x' = -, 2/' = ^, or x=ax', y = ay\

a a
into the curve
a2a;'2 - a?y"^ = a%
or x'^ — = 1.
y'"^

Thus all equilateral hyperbolas are seen to be similar to one

another, since each can be transformed by (1) into the particu-
lar equilateral hyperbola
a;2 — ?/2 = 1.

Inverse of a Transformation. The transformation,

x'

(2) y'
y = —,

obtained by solving the formulas (1) for x, y, is called the in-

verse of the transformation (1). In general, if a given trans-
formation carries (x, y) into (x\ y'), the transformation carry-
ing (x', y') into (x, y) is known as the inverse of the given
transformation. Thus, the rotation (2), § 2, is the inverse of
the rotation (1), § 2.
It is clear that the effect of the inverse transformation, if
performed after the given one, is to nullify the given one.
Thus (1), § 2, rotates all figures through the angle 6, and then
(2), § 2, rotates them through the angle — 6, i.e. back into
their original positions.

EXERCISES
1. Show that the parabola y'^ = 2mx, < ?7i, can be trans-
formed by (1) into the parabola = x. What
y"^ value must be
taken for k ?
336 ANALYTIC GEOMETRY

2. Show tliat the effect of performing transformation (1)

and then transformation (2) is to leave the pla'ne unchanged.

4. Reflections in the Axes. Let the plane be reflected in

the axis of x. In other words, let it be rotated through 180°
about the axis of x. Let P: {x, y) be an
arbitrary point, and let P {x', y') be the ' :

point into which P is carried. Then,

obviously,

\y =-y-
Similarly, a reflection in the axis of y is

represented by the formulas :

Fig. 7
X = — X,
(2)
y' = y'
The condition that a curve be symmetric in one of the axes
(cf. Ch. V, § 2) is obtained at once from these transformations.
Thus the curve C will be symmetric in the axis of x if the
curve C, into which C is carried by (1), is the same curve as C;
and the test for this is, that the equation of C be essentially
unchanged when the transformation (1) is performed on it.
For example, if C is the curve

2^ -f- a:2 =2 2/2 + a^,

its equation is unchanged by (1), and hence is symmetric in

the axis of x. But it is changed by (2), and C is, therefore,
not symmetric in the axis of y.

Isogonal Transformations. A transformation is said to be

equiangular or isogonal if the angle which any two intersecting
curves, (7i and Co, make with each other is the same as the
angle which the transformed curves, C\ and C'2, make with
each other.
All of the transformations considered thus far are evidently
isogonal. We turn now to a transformation which is not.
 :

TRANSFORMATIONS OF THE PLANE 337

EXERCISE
Show
that the equations of the inverse of the reflection (1)
[or (2)] are precisely of the same form as the equations of the
reflection. A transformation for which this is true is said to
be involutory.

5. Simple Elongations and Compressions. Let the plane be

stretched directly away from the axis of ic, so that each point
is carried, along a parallel to the
axis of 2/, to twice its original dis-
tance from the axis of x (Fig. 8).
Evidently, the analytic condition
is that

tAJ — •X/«

More generally, if a point

P : (x, y) is to be carried to I times
Fig. 8
its from the axis
original distance
of X, where I may have any positive constant value, not unity,
the transformation will be given by the formulas

I
x^ = X,
(1)
\y' = iy-
When I is greater than unity, these formulas represent an
elongation ; when I is less than unity, they represent a com-
pression.
If the elongation is away from the axis of y or the compres-
sion is toward it, then

where k is greater than unity in the first case and less than
unity, but positive, in the second.
These transformations were discussed geometrically in Ch.
XIV, § 7. There we called them one-dimensional^ or simple,
elongations and compressions or, jointly, one-dimensional -,

strains. .
 :

338 ANALYTIC GEOMETRY

Example 1. Let the circle

i) a;2^ 2/2=1
be subjected to the transformation (1). Then it goes over into
7/ '2

a;'2 + ^=l.
Thus the circle i) is carried into the ellipse

ii) ''=''
f+i=^'
whose axis lying along the axis of x is identical with the cor-
responding diameter of the circle, but whose axis lying along
the axis of y is the corresponding diameter of the circle
stretched in the ratio 1:1.
Example 2. Let the ellipse ii) be subjected to the trans-
formation (2). Then
— + ^=1.
Tc^ 62

Thus the ellipse ii) is carried into the ellipse

iii)
S+I=i' «='^-

From these examples we see that the particular circle i) can be

carried by means of two one-dimensional strains into an arbi-
trary ellipse iii) whose axes lie along the axes of coordinates.
Exercise. Show that the circle i) can be carried into the
ellipse iii) by a single one-dimensional strain and a transforma-
tion of similitude.

Product of Two Transformations. The combined effect of

the two transformations of Examples 1 and 2 can be repre-
sented analytically as follows. Eirst, we have
(a) x' = x. y'=iy-
Next, the point (x', y') is carried into (ic", y") by the trans-
formation
(h) x"=kx', y" = y'-
Eliminating the intermediate stage (x', y'), we get

(c) wc" = kx, y" = ly.

 TRANSFORMATIONS OF THE PLANE 339

A transformation, as (c), which arises as the result of two

successive transformations, as (a) and (b), is called the pi^oduct of
these transformations. Similarly, (a) and (b) are spoken of as
the factors of (c) ; or (c) is said to be factored into (a) and (6).
Let the student verify the fact that, if the circle i) is sub-
jected to the transformation (c), it is carried over into the
same ellipse iii) into which
was carried by the successive
i)

applications of the transformations (a) and (5).

Properties of the Transformation. One of tlie most impor-

tant properties of one-dimensional strains is that, like the trans-
formations previously studied, they carry straight lines over into
straight lines.
This was proved geometrically on p. 304. The transforma-
tion considered there is given analytically by (1). It was
proved also that, if i is a line of slope A, the slope of the line
into which L is carried by (1) is
*

(3) X' = IX.

From the theorem contained 'in formula (3), it is seen that
a one-dimensional straiyi carries parallel lines into parallel lines.

Consider an arbitrary curve, C. Its

slope at any one of its points, P, is y

•
X = limM.
q^pPM
Perform the transformation (1) on C.
Then PM
remains unchanged in
length ; but MQ goes over into -»Ar

•X
o
Hence the slope. A', of C is Fig. 9

y^limM^ = nml^ = l} im
MQ
q-^P'P'M' Q^P PM q^pPM
or y = i\.
We have thus extended the validity of formulas (3).
* The proofs on p. 304 were given for compressions, but they are valid,
also, for elongations.

I
340 ANALYTIC GEOMETRY

It follows from two curves, Oi and C2,

this extension that, if
are tangent to each other, the transformed curves, C\ and C'2,
will also be tangent. For, if C^ and C2 are tangent, they have
the same slope \ at the point of tangency. Hence, at the cor-
responding point, G\ and G'2 will each have the slope A' = l\
and consequently will be tangent to each other.
Angles are not in general preserved by a one-dimensional
strain. It is true that right angles whose sides are parallel
respectively to the coordinate axes go over into right angles
satisfying the same condition. But consider, for example, the
angle between a line L
and the axis of x (p. 304, Fig. 13).
L is carried into L by the transformation (1), and the axis of x
remains fixed. It is clear, then, that the new angle is not
equal to the original one.
The areas of figures, also, are
changed, and changed in precisely

J
m
the ratio of

are
The
parallel

the areas
shown in the
of
I (or k)
ously true for rectangles whose sides
to the
A, of any other figure is the
area,
limit approached by the sum, B, of
rectangles
drawing
A = \\mB.
: 1.

:
This is

coordinate axes.

inscribed as
obvi-

By the transformation (1), A is carried

X into A' and B into B' evidently
o ;

Fig. 10 A' = lim B'.

Since the area of each rectangle represented in the sum B'
is I times the area of the rectangle from which it originated,

= IB.
B'
Hence A' = lim IB = lim I B,

or A' = iAy q. e. d.

Example. The area of the circle i) is tt. It follows, then,

that the area of the ellipse ii) is w b. Applying the method
 :

TRANSFORMATIONS OF THE PLANE 341

again to this ellipse, we obtain as the area of the ellipse iii)

the value tt a^. We have thus obtained the following result.

Tlie area of the ellijise

a^ b^~
is A= TT ah.

EXERCISES
1. Show that the circle x^ -f if = a^ can be carried into the
ellipse

a^ ¥
by a one-dimensional strain ; cf. p. 306.
Prove that the rectangular hyperbola
2. x^ — if = a^ can be
carried into the hyperbola

by a one-dimensional strain.

3. In ^Examples
2 of p. 338 can the order of the trans-
1,
formations be reversed ? Prove your answer.
4. A
one-dimensional strain changes, in general, the shapes
of curves. Is this true in all cases? For example, in the
case of a parabola ?
5. Prove analytically that the transformation- (1) carries a
straight line into a straight line.

6. Show analytically that, if a line L is carried by (1) into

a line L', the slopes of L and IJ are connected by formula (3).
Find the equations of the transformation which
7. is the
product of the two transformations

x' = x-l, y' = y + 2, x" = x' ,

y" = - y'.
8. The same for the rotation about the origin through 45°,
followed by the translation which carries the origin into
the point (3,-1).
342 ANALYTIC GEOMETRY

9. The same for the transformation of similitude which

doubles all the lengths, followed by a reflection in the axis of y.

10. Prove analytically, for the following transformations,

that the product of a transformation and its inverse is the
identical transformation^ x' = x, y' = y.
(a) translations ; (6) rotations ;

(c) reflections ;
{d) one-dimensional strains.

11. Factor the transformation

x' = 4.x, y' = '^yy
(a) into two one-dimensional strains ;

(6) into a one-dimensional strain and a transformation of

similitude.

12. Prove that the rotation about the origin through 180° —
also called the reflection in the origin — is the product of the
reflections in the axes.

13. Factor the transformation

x' = 4:X, y' = — 2y

into two one-dimensional strains g,nd the reflection in, the axis
of X. How else can it be factored ?

Express each of the following transformations as the prod-

uct of two or more simple transformations.

14.
 : : :

TRANSFORMATIONS OF THE PLANE 343

All of the foregoing transformations come under this type,

but there are transformations comprised under (1), for
example,
(2) x' = 2x-Sy-\-l, y' = -x-\-^y-2,
which are not of any of the above forms. We shall prove the
following theorem.

Theorem 1. Tlie transformation of the plane defined by

means of equations (1) can be generated by a succession of the
transformations studied in §§ 1-5. In other words, it can be
FACTORED into transformations of the type of those o/ §§ 1-5.

Proof If c and c' are not both 0, let the {x, ?/)-plane be
subjected to a translation

where i, rj are arbitrary and shall be determined presently.

Thus equations (1) are replaced by the following
lx'= axi-\- byi—{ai-{- b-q — c),
^^ y' = a% + b'y, - {a'i + b'rj - &).
\

We now determine ^ and rj so that both parentheses will

disappear. This is done by solving the simultaneous linear
equations
ai -{-by] — c = 0,

a'i + b'r) - c' = 0.

The solution is always possible and unique, since, by hypothe-
sis, A= ab' — a'5 =^ 0.
We thus have a simpler pair of equations to study, namely,

/gx x' = ax^ -f 6?/i,

I
[y' = a'x,-^b'y,.
It will be sufficient, then, if we can prove our theorem for
the case that c = and c' = 0, i.e. for the pair of equations

/£»\ {x'= ax -\- by,

^^ [y' = a'x+b'y,
 : : :

344 ANALYTIC GEOMETRY

for we have just seen that we can pass from equations (1) to
equations (5) by a translation, and this is one of the transfor-

mations admitted by the theorem.

Consider an arbitrary circle with its center at the origin

(7)
0^2
+ 2/2 = p2.
Let us see into what curve it is carried by (6).
To do this, solve equations (6) for x and y. The result is

the inverse of (6), namely

x = Ax' -\- By',
{

(8)
= A'x' + By,
1 y

where A-^' B=- A' = -

a'
B' = a-
"^"A'
and AB'-A'B = '^^^^^ = ^-^=^0.
A2 A
Next, substitute these values for x and y in (7)

(9) (^2 _^ ^/2)

^f2
^ 2{AB + A'B') x'y' + (^ + B'~) y'^ = p\
The locus of this equation is an ellipse with its center at
the origin. For, first, the equation has a locus, since all the
points {x', y') into which the points
{x, y) of the circle (7) are carried by

(6) lie on (9). Secondly, the locus

does not extend to infinity in any
direction.*
In general, this ellipse will not
be a circle. Let L' be the extremity
of an axis. Since the transforma-
tion (8) carries any straight line
through the origin into a straight
line through the origin, OL' will
correspond to a certain radius OL
of the circle (7).
Fig. 11 We now rotate the {x, 2/)-plane

*0r, the expression ^ - 4 ^ C (Ch. XII, §3), formed for (9), is

negative :

^AB + A'B'y - 4(^2 4. ^/2) (^2 + ^/2) ^_ 4(^J5/ _ A' By < 0.

 : ^

TRANSFORMATIONS OF THE PLANE 345

about the origin through an angle such that OL comes to lie

along the positive axis of x :

.^^. [Xi = x cos ^ — sin 6,

2/

^ ^ \y^ = x sin + y cos 6.

Furthermore, we rotate the (x', 2/')-plane about the origin

through such an angle 6' that OL' comes to lie along the
positive axis of x :

.... [Xi = x' cos 0' — y' sin B'

^ ^ \y{ — x' sin e' + y' cos 6'.

What is the final result? Obviously the following. An

arbitrary point (aji, y^ of the plane is carried by the inverse of
(10) into a point (x, y) this point is ;
carried by (6) into a
point (ic', y') and finally the point (x',
;
?/') thus obtained is

carried by (11) into To write out these transforma-

(ic/,?//).

tions explicitly would be a long piece of work but it is not ;

necessary to do so. For, first of all, each is linear and leaves

the origin unchanged. Hence the final transformation, carry-
ing the point (a^i, y^ directly into the point (a;/, 2//)> is also
linear,* and it leaves the origin unchanged. It is, then, of the
form

Consider next what we know about this transformation,

i) It carries the positive axis of x^ over into itself. Hence,
when ?/i
= 0, y^ must also vanish, no matter what value x^ may
nave ; tnus
 : : :

346 ANALYTIC GEOMETRY

To find the equation of the ellipse, solve (12) for Xi, yi, re-

membering that y = :

ft ceo

Thus (13) is seen to go into the ellipse

(15) '^IW + 2 %^x,'y,' + («' + ^^)>/2 = P^

Since the axes of the ellipse (15) lie along the coordinate
a;i'2/i' must disappear, and so we must have
axes, the term in

5l«= 0.
Now, 51 cannot be 0, for =
A 1/a. Hence ^ must vanish

^ = 0.
It follows, then, that

and thus (12) reduces to the transformation

Moreover, since the positive axis of Xi goes over into the

positive axis of x^, we see that

< a.

But 8 may In this case, a reflection in the axis

be negative.
of Xi will change the sign of 8, and hence the case is reduced
to the one in which 8 is positive

< 8.

Finally, the transformation (16) is the product of two one-

dimensional strains :

y' = y, \y' = ^y,

one along the axis of x and one along the axis of y.
Let us now recapitulate. The point (x, y) is carried into the
point (xi, 2/i) by the rotation (10) ;
{x^, y^) is carried into (a;/, yi)
 : .

TRANSFORMATIONS OF THE PLANE 347

by (16), which is a product of two one-dimensional strains and,

perhaps, the reflection in the axis of x ; finally, (cc/, y/) is

carried into (a;', ?/') by the inverse of another rotation.

(11),
The transforDiation (6) is, then, the product of these trans-
formations in the order enumerated. The proof of Theorem I
is thus complete.

Properties which all the component transformations of (1)

have in common are also properties of (1). Consequently, tlie

general affine transformation carries straight lines into straight

lines, 2Kirallel lines into parallel lines, and tangent curves into tan-
gent curves. It does not, in general, preserve angles or areas.

Isogonal Transformations. Of the component transformations

of (6), the rotations (10) and (11) always preserve angles.
This is true of the transformation (16) if and only if it is a
transformation of similarity, with or without a reflection in the
axis of X, i.e. if and only if 8 =
± a. Hence the most general
isogonal transformation of the form (6) is the product of the
rotation (10), the transformation

^\ = p^i, yi=±pyi, p = a>0,

and the inverse of the rotation (11). This product is easily
found to be
x' = p[x cos (0 TO')-y sin (0 T 0')^,
y' = ±p[xsm{eTO') -^yGos(eT B')].
But the angle 6—6' (or 6 + 6') is no more general than a
single angle, which we may denote by c^. Thus the result can
be written as
x' = p(x cos <f)
— y sin (f), y' =± p(x smcf> -\- y cos <f}).

Replacing x,y by Xi, y^, so that this transformation of the

form (6) reverts to the form (5), and then applying the trans-
lation (3), we obtain as the most general transformation of the
form (1) which is isogonal
^' = p{x eoscf>-y sin <^) + c,
I
\y' = ± p(x sin -^ y cos <^) c'
cf) -\-

Here
 348 ANALYTIC GEOMETRY

(18) a=p cos <fi, h= —psin<fi, a'=±psiii<f>, b'=±p cos <!>,

and hence,
(19) b' = ±a, a' = Tb.
Conversely, every transformation (1), for which (19) is true
for one set of signs, can be written in one of the forms (17)
and hence is isogonal. For, if a, b, a', b' are given, satisfying
(19) for one set of signs, values of p, cos <fi, and sin eft can be
found, so that equations (18) hold for the same set of signs.
These values are, namely,

p=Va2 + 62, cosc^ = —===,

Va2
sin<^=—
Va2
==.
-f b^ -|- b^

The following theorem summaries our results.

Theorem 2. TJie transformation (1) is isogonal when and

only when either
b' =a and a' =—b or b' =~ a and a' = b.
If it is isogonal, it can be written in one of the forms (17).

Homogeneous and Noyi-Homogeneous Transformations. A

polynomial in x and y is homogeneous, if its terms are all of
the same degree in x and y*^ Thus, the left-hand sides of for-
mulas (6) are homogeneous polynomials of the first degree. Ac-
cordingly, a transformation of the form (6) is called a homo-
geneous affine transformatiQn ; and, in distinction, a transforma-
tion of the form (1), where c and c' are not both zero, a non-
homogeneous affine transformation.
Since (1) can be reduced to (6) by means of a translation, we
have the theorem.
Theorem 3. A non-homogeneous affine transformation is the

product of a translation and the corresponding homogeneous

transformation.
* Only those terms with non-vanishing coefficients need be considered,
since a term whose coefficient vanishes has the value of 0, and is not de-
fined as having a degree. For example, if, in the polynomial x^ -\- 2xy
-f- ax, a has the value 0, the polynomial is homogeneous of the second
degree.
 TRANSFORMATIONS OF THE PLANE 349

EXERCISES
1. Prove analytically that the affine transformation (1)
carries a straight line into a straight line.

2. Show that, if X is a line of slope X, the line into which L

is carried by (1) has the slope
a'-\-b'X
A,, = •

a -\- b X

3. Using the result of Ex. 2, prove that (1) carries parallel

lines into parallel lines.

7. Factorization of Particular Transformations. We proceed

to illustrate the theory of the preceding section by carrying it

through, step by step, for a particular case.

Let the given transformation be

(1) x' =zx-{-3y, y' = — Sx—y.

The first step is to find the ellipse into which the circle,

(2) a;24-2/^=p^

is carried by (1). Solving equations (1) for x and y, and sub-

stituting the values obtained, namely,

x = -lx'- I/, y = ^x' + \y',

in (2), we have, finally,

(3) nx'^^ + Q x'y' + 5 2/'2 = 32 p\

This ellipse is as shown. One axis,
OL'y lies along the line x'-\-y' — 0.
By adding equations (1), we have
x'-^y'^-2x-{-2y,
and hence the radius OL of the circle
which is carried into OL' is along the
line X — y = 0. Furthermore, if L'
lies in the fourth quadrant, as in the
figure, then L lies in the first — not
in the third. Fig. 12
 : : :

350 ANALYTIC GEOMETRY

Consequently, a rotation of the (x, 2/)-plane about through

- 45°
(4) X, = Vi
V2 V2
brings OL to lie along the positive axis of x. And a rotation
of the (x'y 2/')-plane about through -|- 45°

(5) x,' = x' -

y )
, x' -\-y'

V2 V2
does the same for OL'.
The next step is to find the transformation carrying {x^, y^
directly into (ic/, yi). This we do by eliminating x, y and
x', y' from (1), (4), (5). Thus,

= ^(a^-f 32/ + 3a; + = 4^^ = 4a:i,

a^/ 2/)

V2 V2
2//=-L(a; + 32/~3a^-2/)=2:=^=22/i,
V2 V2
or
(6) x{ =4 a^i, 2// = 22/1.
Finally, we solve (5) for x^ and 2/'

(7) a;' = a?/ -f- 2/1'

2/
"
/_-^L + V
V2 '
V2
The transformation (1) is now seen to be the product of the
transformations (4), (6), and (7) ; (4)
carries (a;, 1/) into (cci, i/i) "^7 ^ rotation
about through — 45° ; (6) carries

(^i> 2/1) ill to (a;/, ?//) by two one-di-

mensional strains ; (7) carries (a;/, 2/1')
fx,i/)
*''^'^
into (it-', ?/') by another rotation about
^ through — 45°.
Fig. 13 Simplifications in Technique. In-
stead of seeking the ellipse in the
{x', y')-^\2tjiQ into which the circle (2) is carried, we might
equally well ask for the ellipse in the {x, ?/)-plane which is
 TRANSFORMATIONS OF THE PLANE 351

carried into the circle

(8) x"- + y" = p2.
The roles of the two planes are merely reversed. Adopting
this procedure obviates the necessity of solving (1) for the
values of x, y to be substituted in (2). For now (2) is replaced
by (8) and x'and y' are given by (1). Thus one step in the

process is eliminated. The others remain unchanged.

Another simplification arises in factoring a transformation
of the form
,Qx ]x'= + 3^ + 4,
a;

^^ \y' = -^x- y-2.

Instead of proceeding as in § 6, we set

I
_ and <; _
\y' = y-2, [y=:-Sx- y.

Thus (9) is the product of the transformation (1) and the

translation which carries the origin into the point (4, — 2).

EXERCISES

Factor the following transformations, using the simplified

method.
1. x' = x-{-Sy, y' = -3x-y.
2. x' = 3x-2y, y' = -2x-^3y.

3. x' = 5x-\-lly, y' = 10x-\-2y.

4. x' = 6x-^lSy-2, y' = 17x-h y S. -\-

5. x' = lllx-\-4:y-^l, y' = 52x-{-7Sy-7.

8. Simple Shears. In a rectangle with its center at the

origin and with its sides parallel to the coordinate axes, draw
the lines parallel to the axis of x. Twist this rectangle as

shown in the figure, leaving the line along the axis of x fixed
and sliding each parallel line ^long itself into a new position.
Thinking of the lines as representing the edges of a pack of
cards or of a block of paper is an aid in visualizing the motion.
 X

352 ANALYTIC GEOMETRY

If the line one unit above the cc-axis slides to the right
through the distance k, then the line y units above the a>axis
will evidently slide to the
right through the distance ky,
while a line which is, say, 2
units below the x-axis will
slide a distance 2/c to the
left. In other words, the
algebraic distance through
which each line slides is

— equal to k times the algebraic

distance of the line from the
ic-axis.

This is true, also, of the

Fig. 14
motion of each point of the
rectangle, since the lines slide as units. If, then, the whole

plane is twisted according to this law, an arbitrary point {x, y)

will be carried along a parallel to the axis of x through the

algebraic distance ky. Hence
x' = x-\- ky,
(1)
y' = y,
are the equations of the transformation.
Thus far we have assumed that k is positive. It may

equally well be negative. Then points above the axis of x are

shifted to the left, and points below it to the right.
If the sliding were along parallels to the axis of y, the trans-
formation would be
f
x' = X,
(2)
\y' = lx + y,
where I is any constant, not zero.
These transformations are known as simple shears, and the
motions which they generate are called shearing motions.

Examjyle 1. Subject the curve

(3) y = a^ + 2x
 TRANSFORMATIONS OF THE PLANE 353

to the shear

Here
 : :

354 ANALYTIC GEOMETRY

Simple shears do, however, preserve

areas. For, first, this is true for any
rectangle whose base is parallel to the
direction of shearing, since such a rec-
tangle is carried into a parallelogram
with the same lengths of base and alti-

tude ;
cf Fig. 14.
. Secondly, the area A
of any other figure can be considered
as the limit of the sum of the areas of
rectangles of the type just described,
which are inscribed in the figure as
shown. But this sum is equal always
to the sum of the areas of the corre-
sponding parallelograms, whose limit is
Fig. 17
the area A^ of the transformed figure.
Consequently, A = A', q. e. d.

EXERCISES

Construct the following curves.

2. y = 2x^ — ^x. 5. Sx=:y^ — 6y.

S. y = — a^-\-x. 6. 2x = — 4:y^ — Sy.

The same for the following curves, making use of a transla-

tion as well as a shear.

7. y= a^-\-x-2. 9. x= 2y^ - 'Sy -\-l.

8. 2y = 2afi-x-^3. 10. 3x = - 6y^ -2y -7.

11. Construct the curve

y=z2x^-6x'^ + 7x-l,
beginning by putting the equation into the form

y—h = 2{x — ay-\-k{x — a).

The same for the curves

12. y = x^ + 6x^-{-10x-\'7.
 :

TRANSFORMATIONS OF THE PLANE 355

13. 2a;= 22/3- 67/2 + 9?/ -3.

Factor the two following shears by the method of § § 6, 7.

14. x' = x-\-^^Sy, y' = y-

15. x' = x, y' = -^VSx-\-y.

9. Second Method of Factorization. Homogeneous Strains.

Theorem. The homogeneous affine transformatioi

x' = ax-\- b yy
\

A = aV - a'h ^ 0,
(1)
[
y' = o!^ + ^V,
can he factored into one-dimensional strains and simple shears,
ivith the addition, in certain cases, of a reflection in one or both

axes.

Case 1 a and b' not both 0. In proving the theorem we

:

begin with the case in which a and b' are not both zero, and
assume first that a=^0. A simple shear which suggests itself
as a probable component of (1) is

[ .
^

(2) a
\yi = y-
Eliminating x, y from (1) and (2), we obtain

x^ = axi,
(3) y = ci^i + - 2/1-
a

This transformation suggests as a factor the second shear

^ )
\y' = kXc^-\- 2/2,

where the value of k is to be determined. Elimination of

ic', y' from (3) and (4) gives

(y)
1
2/9 = (a' — ka) Xi-\ —a 2/i-

An obvious choice for k is that which makes a^ — ka= ; then

k — a'/a.
 :

356 ANALYTIC GEOMETRY

We have now factored. (1) into the transformations (2), (5),

(4), where k = a'/a, namely into :

«' = X2, \ X2 = axi, \xi = x-\--y,

[

[a [
a ^

The first and last of these transformations are simple shears.

The second can be factored into two one-dimensional strains,
or, in case a or A/a or both are negative, into these and a re-

flection in one or the other or both axes. Thus the theorem

is proved in this case.
The proof is similar in the case that 6' ^ 0. The factors of
(1) are

.^.
« =3^2+^2/2, ^2=^3^1, \Xi = X,

Case 2 : a = b' = 0. Here (1) becomes

(8)
\^' = by, A = -a'b^O.
\y' = wx.
This transformation can be factored into a rotation about the
origin through 90°
(9) x^ = -y, 2/1 = «>
and the transformation
x' = — hxi, y' = ayi-
It can be shown that the rotation (9) is the product of three
simple shears, namely,
\ X = X2, -
j 072 = — 2/b
ftq I ^ = ^j
!

i 2/
..I — ^. +
= 2^2
_l_ .,.
2/2,
]
i
„
2/2
— ^1,
= ./_ \
12/1 = ^ + 2/,

and this completes the proof of the theorem.

Homogeneous Strains. The extension to space of the trans-

formations (1) is given by the formulas
x' =a X+b y c -\- z,

. y' = a' X b' y -{- -{- c' z,

z' = a"x + b"y + c"z.

 TRANSFORMATIONS OF THE PLANE 357

The three-dimensional case admits a treatment similar

to the foregoing, and the results are like those obtained
above.
Transformations of the form (1) or (9) are known in
physics homogeneous strains.
as They are of particular
importance in the theory of elasticity. For, it can be shown
that, if an elastic body, such as a solid piece of rubber
or of steel, is slightly deformed from its normal shape, the
displacement of its points can be represented to a high
degree of approximation by a transformation of the form (1)
or (9).
It is a fact, which we shall not attempt to prove, that a
transformation (1) representing in the above sense a slight de-
formation cannot have a reflection in an axis as one of its com-
ponent transformations.
One-dimensional strains —
simple elongations and compres-
sions —and simple shears are often given the single name,
simple strains. Adopting this terminology, we can say Every :

homogeneous strain representing in the above sense a slight deforma-

tion can be generated by a succession of simple strains ivithout
refections.

EXERCISES
Factor the following homogeneous strains by the method of
this paragraph.

1. The strain of Ex. 1, § 7. 2. The strain of Ex. 3, § 7.

3. x' = 6y,y' = — 2x-\-y. 4. x'=ox — 5y, y' = 4:X + Sy.

The following homogeneous strains represent slight dis-

placements. Factor them and note that a reflection in an axis
never appears as a component transformation.

5. x' = 1.01x-\-.02y, y'^.03x-\- .9Sy.

6. x' = .9x- .ly, y'= .2x-^l.ly.
7. x'=(l-i-a)x-\-fty, y'= yx -\-(l -j- 8)y,
where a, ft, y, 8 are small quantities, such as .01 or — .08.
358 ANALYTIC GEOMETRY

Factor the following transformations.

8. The transformation of Ex. 4, § 7.

Q, x' = — 2x-\-y + 3, y' = Sx — 2y + 1.

EXERCISES ON CHAPTER XV
1. Prove that the product of any two affine transformations
is an affine transformation.

Definition. The general affine transformation (1), § 6, is

called non-singular, if

if A= 0, it is called singular. The expression A is known as

the determinant (cf. Ch. XVI) of the transformation.

2. The inverse of any non-singular affine transformation

is non-singular. This was proved incidentally on p. 344 in
the case of a homogeneous transformation. Prove it in the
general case.
3. Show that the product of two non-singular affine trans-
formations is non-singular.
Suggestion. Prove that the determinant of the product
transformation is the product of the determinants of the given
transformations.
4. The transformation,
x' = 2x — y, y' = 4:X — 2y,
is singular. Verify this and show that the transformation
carries all the points of the plane into points of the line
2x' — y' =: 0. Has the transformation any inverse ?
5. The product of the general rotation about the origin,
followed by the general translation, is a transformation known
as the general rigid motion. Find its equations.
Ans. x' = X cos 6 — y sin -\- a; y' = x sin -\- y cos 6 + b.
6. The product of the transformation of similitude of § 3
and the general rigid motion of the preceding exercise is

known as the general transformatioyi of similitude. Find its

 TRANSFORMATIONS OF THE PLANE 359

equations, and show that it is identical with the general isogonal

transformation for which 6' = a, a' =— 6 (§ 6, Th. 2).

7. A non-singular affine transformation carries the four

collinear points Pi, P2, Qi, Q2, into the four collinear points
P'l, P'2^ Q'b Q'2' Prove that, if Qi, Q2 divide Pj, P^ harmon-
ically, Q'l, Q'2 will divide P'iP'2 harmonically.
Suggestion. Prove the theorem first for the transformations
considered in §§ 1-5.
8. Find the equations of a rotation of the plane about the
point (aJoj 2/o) through the angle 0.
9. The plane is stretched uniformly in all directions away
from the point (.Tq, 2/0) • Pi^^^ ^^^ equations representing the
transformation.
10. Deduce the equations of the reflection in the line

Ax + By^ 0=0.
11. Deduce the formulas representing a one-dimensional
strain away from the line of Ex. 10.

12. Find the equations of the simple shear which leaves

each point of the line of Ex. 10 fixed.
13. Let the simple shear (1), § 8, be factored into the three
transformations of § 6, namely (10), (16), and the inverse of
(11). Prove that sin 2^' = sin 2^, but that the only allowable
solutions of this equation are ^' = 90°-^ and 6' =210° -6,
together with those equivalent to them. Show that, if the
first of these solutions is, chosen, a = tan B,h= cot 6, whereas,

if the second is taken, a = — tan ^, 8 = — cot 6. Prove that, in

either case, 2 cot 2 = k.
Suggestion. Form the product of the three transformations
and demand that it be identical with the transformation (1),
§8.
 CHAPTER XVI
DETERMINANTS AND THEIR APPLICATIONS

I. Determinants

1. Simultaneous Linear Equations. The solution of the

simultaneous equations,
/-j^N ciix + 61?/ =K
a^x + 622/ = h,
is

aj)2 — azbi aib2 — (X261

provided aib2 — ct2&i 0.* ^

If we have three simultaneous linear equations in three
unknowns,
aix + biy 4- CiZ = Zci,

(3) a^ 4- 622/ + C22: = A:2,

and first eliminate z, obtaining two equations in x and y, and

then from these equations eliminate y, we find, as the value of x,

«1&2C3 + «2^3Cl + «3^lC2 — «3^2Cl — a2^lC3 — ClACz

Similarly, we can find

the values of y and z. These will also be
in the form of quotients, with the same denominator as in (4),

*If 0162 — ciibi = (but ai and 61, and 02 and 62, are not both zero),
the two straight hnes represented by equations (1) are eitlier parallel
or coincident (Ch. 2, § 10, Ths. 3, 5) in the former case the equations
;

have no solution, in the latter, infinitely many solutions. Both cases are
exceptional to the general case, 0162 — a2&i 1^ 0, in which the solution
(2) is unique.
360
 DETERMINANTS 361

and the solution is valid subject to the condition that this

denominator is not zero.

Two- and Three-Rowed Determinants. The expressions

2.

in the numerators and denominators of the quotients in (2)

and (4) are of so great importance that they are given a name.
They are called determinants, — those in (2), determinants of
the second order, and those in (4), determinants of the third
order. A determinant, then, is a polynomial of the above type.
The determinant of the second order,

can easily be remembered by means of the diagram

(1)

in which the lines and the signs show how the terms of the
determinant are to be obtained.
The diagram

(2) ^HKAJ"

fulfills the same purpose for the determinant of the third order,

«l?>2^3 4- a2^3Cl + «3&lC2 — «3^2Cl — CL^^lCz — <^lh^2'

The four quantities «!, a-,, hi, 62? arranged in a square as in
(1), form what known
as a square array of the second order.
is

Similarly, the system of nine quantities, which forms the basis

of the diagram (2), is known as a square array of the third order.
The square array is not itself the determinant. It is merely
a convenient arrangement of the given four, or nine, quantities,
from which the value of the determinant can be written down.
However, it is common practice to use, as a symbol or nota-
 :

362 ANALYTIC GEOMETRY

tion for the determinant, the square array inclosed between

vertical bars, and to write, accordingly

(3) = tti^g — «2^1>

61 Ci

(4) 69 c.
2 = «1?>2C3 + «2^3Cl + a3^lC2 — «3^2Cl— a2^iC3 — ai63C2.
«3 C3

These symbols for the determinants are sometimes abbreviated

still further. Instead of the first, we often find \ai 62 1 or
merely |
a 6 |,
and for the second, \ai h^ Cg |
or |
a h c\.*
The solution of the equations (1), § 1, we can now write
in the form
K
X =
 DETERMINANTS 363

the 6's — the coefficients of y — by the A:'s ; and likewise

for z*
The four, or nine, quantities from which the determinant is
formed are known as the elements of the determinant. The
rows and columns in which they are arranged are called the
rows and columns of the determinant. The diagonal contain-
ing the elements cii, h^ (cg) is the principal diagonal ; the other,
the secondary diagonal. The determinants are often called two-
and three-rowed determinants, instead of determinants of the
second and third orders.

EXERCISES
Evaluate the following determinants.

2 3
1.
3 5

5.
364 ANALYTIC GEOMETRY

3. Determinants of the Fourth and Higher Orders. Given

sixteen quantities arranged in a square array :

tti '
bi Ci di

.. (h ^2 ^2 C?2

«3 h C3 dg

d^ O4 C4 (X4

What shall we mean by the determinant symbolized or de-

noted by
cii &i Ci di

^2 2 ^2 2

«3 ^3 C3 dg

flt^ O4 C4 U4

or, more simply, by |ai 62 Cg di |

or | a 6 c d |?

Ifwe were to proceed as before, we should write down four

simultaneous linear equations in four unknowns, with the
elements of (1) as the coefficients of the unknowns and ki, Ic^,

kg, A:4as the constant terms, and then solve the equations.
The value of each unknown would be a quotient, and all four
quotients would have the same denominator, which we should
then define as the determinant \a b c d\. As a matter of fact,
this denominator and each of the numerators contains 24 terms.
The prospect of solving the equations is, then, forbidding.
Why not form the products suggested by a diagram based on
(1), similar to the diagram for the three-rowed determinant, pre-
fix the proper signs, and call the result the determinant ? Un-
fortunately this method yields but 8 terms, whereas according
to our prediction the determinant, properly defined, contains 24.
We adopt here a new method of attack. Let us inspect
more between the square arrays of
closely the relationship
orders two and three and the corresponding determinants.
Consider a specimen term of the determinant (4), § 2. It con-
tains just one a, just one b, and just one c furthermore, each of ;

the subscripts 1, 2, 3 appears just once. In other words, the

term is the product of three elements, one from each row and one
 DETERMINANTS 365

from each column of the square array. Moreover, every product

of this typeis present as some term in the determinant, as can

be shown by writing down all such products and comparing

them with the terms of the determinant.
By analogy, then, to form the determinant \a h c d\, we
should write down all the products of elements of (1), each of
which contains just one factor from each row and just one
factor from each column of (1), that is, all the products of the
form afif^di, where i, j, k, I are the numbers 1, 2, 3, 4 in all
possible orders. There are 24 such products. For, we can
choose the first factor, say from the column of o's, in four ways
— from any one of the four rows and then the second factor, ;

say from the column of 6's, in three ways from any one of the —
three remaining rows and the third factor, in two ways the
; ;

fourth is then uniquely determined. The number of possible

products is, therefore, 4 • 3 • 2 • 1 = 4 = 24.
!

It remains to determine the signs to be given to the 24

products. Toward this end, let us write down the subscripts of
the terms of (4), § 2, in the order in which they occur, ivhen
the letters a, b, c are in their natural order. For the terms with
plus signs we have
12 3, 2 3 1, 3 12,

and for the terms with minus signs,

3 2 1, 2 13, 13 2.

The first set 1 is normal.

2 3 In the second set, 2 3 1, 2 and
3 each precede 1, and we say that there are two inversions from
the normal order. In 3 1 2, 3 precedes 1 and 2, again two —
inversions. In the three sets for the negative terms the num-
ber of inversions is respectively three, one, and one.
It appears, then, that the number of inversions in the set of sub-
scripts for a term with a plus sign is even (or zero), whereas for a
term luith a minus sign, this number is always odd.

Proceeding according to this rule, we should give to each of

the 24 products, afif^di, formed from (1) a plus sign or a minus
 ;

366 ANALYTIC GEOMETRY

sign, according as i j k I presents an even or an odd number of

inversions from the normal order 12 3 4. Thus the product
bidiG^c^ would be taken as plus, since, when the factors are
arranged in the order of the the num-
letters, viz. — a^b2Gidi,
ber of inversions in the subscripts 3 2 4 1 namely 4. is even,
The product aJ}2Cidi would be taken as minus; since the num-
ber of inversions in 4 2 3 1 is odd, namely 5.
We can now give a complete definition of the determinant of.

the fourth order.

Definition. Form all the products of elements of (1) ivhich

contain just one factor from each row and one factor from each
column of (1); to each product afi-c^di pre^x a }:)lus sign or a
minus sign, according as the number of inversions of i j k I from
the normal order 1 2 3 4 is even or odd. TJie sum of the

products, thus signed, is the determinant.

Determinants of the fifth, sixth, and higher orders are

similarly defined. Let the student think through the definition
for a five-rowed determinant, and let him show, also, that in the
case of two-and three-rowed determinants the definition yields
precisely the expressions which were defined as these determi-
nants in § 2.

The signed products which make up a determinant are known

as the terms of the determinant. Thus, -|- as 6204(^1 and
— a^^c^di are terms of |
a 5 c d\.

EXERCISES
1. What is the number of inversions of each of the following
orders, from the normal order ?
(a) 3 1 4 2 ;
(c) 2 5 3 1 4 ;
(e) 3 1 6 4 5 2
(5) 2 4 3 1 ;
(cZ) 4 3 5 2 1 ; (/) 6 5 4 3 2 1.

2. Write out all the terms oi\ab c d\. To how many pro-
ducts have you prefixed plus signs ? To how many, minus signs ?
3. How many terms has a determinant of the fifth order ?
Prove your answer.
 DETERMINANTS 367

4. The same for a determinant of the nth order.

5. Show that the sign to be prefixed to the product of the

elements of the principal diagonal is always the plus sign,
no matter what the order of the determinant.

4. Evaluation of a Determinant by Minors. Fix the atten-

tion on a particular element of a determinant A. Cross out
the row and column in which this element stands. There will
remain a determinant of order one less than that of A. This
determinant is known as the minor of the element chosen.
For example, the minor of a^ in the determinant
368 ANALYTIC GEOMETRY

tlie products of the type described and hence are all the
products occurring in M, q. e. d.
It will be shown later (§ 7) that the terms of mM as they
stand, or the terms of mM with all their signs changed, are
sum of the number of
precisely terms of A, according as the
the row and the number of the column in which m stands is
even or odd. Assuming this, we can now state the theorem :

Theorem 1. If m is the element in the i-th roiv and j-th column

of A, ayid M is its minor, -f '^M or — mM, according as
j i -{-

is even or odd, consists of all the terms of A ichich contain m.

Thus, in the case of the element a2 of (1), = 2, j = 1, and
i

i-\-j = S', accordingly, — a2^2 gi^^s all the terms of (1) con-
taining a2- For b2, i = 2, j = 2, and -\-j = 4, and so -f 62-^2
i

consists of all the terms of (1) containing fto-* Similarly, if

C2 is the minor of C2, — C2C2 consists of all the terms of (1)

containing C2.

The sum
(2) — a2^2 + &2A — C2C2

is precisely the value of the determinant (1). For, it consists

of all the terms of (1) containing 0,2 or 62 or C2, i.e. containing
an element of the second row, and every term of (1) contains
such an element. The student should also verify the state-
ment by comparing the terms of (2), when expanded, with
those of (1).
In (2) we have the sum of the products of the elements of
the second row by their minors, each product having the
proper sign according to Theorem 1. We say that (2) is the
evaluation or expansion of the determinaiit (1) hy the minors of
the second row.
Similarly, the sum,
(o) CjCi — C2C2 + C3G3,
of the products of the elements of the third column of (1) by
their minors, where the signs have been determined by Theorem
* Compare these results with those obtained directly at the beginning
of the section.
 DETERMINANTS 369

1, is precisely the determinant (1). We speak of (3) as the

evaluation of (1) by the minors of the third column.
The reasoning here is perfectly general, applying to a deter-
minant of any order and to any row or column of the deter-
minant. The result we summarize as follows :

Evaluation of a Determinant by the Minors of a

Row or a Column. Single out a row or a column of a
determinant. Multiply each element of it by the minor of
the element and prefix to the product the proper sign, as
determined by Tlieorem 1. The sum of the signed products is
the determinant.

We now have a feasible means of finding the values of de-

terminants of the fourth and higher orders. For example, the
determinant
2 5 -2 8
4 -9 3 7
(4)
-3 6 -4 4
-1 4 -3 5

evaluated by the minors of the first row, is equal to

-9 3
+ (-2)
370 ANALYTIC GEOMETRY

2. (Generalization of Ex. 1.) Given a determinant A of

the nth order and the n quantities ki, k2, •
•, A:„. Single out a
column (or row) of A, form the minors of its elements and
prefix to each the sign prescribed by Th. 1. Multiply each
signed minor by the corresponding k and take the sum of these
products. Prove that this sum is equal to the determinant
obtained from A by replacing the column (or row) in question
by A;i, Jto, •••, K-
By the method of this section, evaluate the following de-
terminants
3. That of Ex. 5, § 2. 4. That of Ex. 7, § 2.

2 3 15 3-2 5 1

5.
5 2-21 6.
4-3 7 2
3 4 6 2 -6 2 -3
-15 2 3 5 3-2 2

1 2
2 1 4 3 2
7. 5 2 2 -1
3 1 4 5
2 3 1 4

Simplified Evaluation by Minors. Given the determinant,

3 2 1
2-134
4 5 2
2 6 3

Three of the four elements in the third column are zero.

Accordingly, if we expand the determinant by the minors of

the third column, three of the four resulting products have

zero factors and drop out, so that there is left, merely,

3 2 1
-3452
2 6 3

Hence — 3 (7)= — 21 is the value of the determinant.

 :

DETERMINANTS 371

It is clear from this example that a determinant which has

the property that all but one of the elements in some row or
in some column are zero
is very simply evaluated. Conse-
quently, a determinant which has not this property can be
if

transformed into an equal determinant which has the prop-

erty, a simple method is at hand for the evaluation of all
determinants.
The transformation in question is always possible. It is
based on the following theorem.

Theorem 2. If the elements of a row {or column) of a de-

terminant are each 7nultipUed by the same quantity and are then
added to the corresponding elements of a second row {or column),
the value of the determinant is unchanged.

Let us first try to appreciate the value of the theorem,

postponing the proof until later. Consider the determinant
(4) of § 4, namely,
2

(1)

By application of the theorem we proceed to transform this

determinant into an equal determinant with the first three ele-
ments of the first column all zero.
Rewrite (1), putting in, to begin with, only the last row
372 ANALYTIC GEOMETRY
 DETERMINANTS 373

EXERCISES
Evaluate, by the above method, the following determinants.
8
374 ANALYTIC GEOMETRY

Consider, next, a determinant of the third order, with two

rows identical. Expand the determinant by the minors of
the third, 6t odd, row. Each of these minors has its two rows
identical and is, therefore, zero, since the theorem has been
proved for two-rowed determinants. Consequently, the given
determinant is zero.
Similarly, having proved the theorem for three-rowed de-
terminants, we can prove it for a four-rowed determinant.
For, we have but to expand the four-rowed determinant by the
minors of a row which is not one of the two identical rows.
This expansion will have the value zero, since each of the
minors in question is a three-rowed determinant with two
identical rows.
The process perpetuates itself. Hence the theorem is true
for a determinant of any order.
The method of proof used here is known as mathematical
induction.The fact that the theorem is true for a two-rowed
determinant leads up to its truth for a three-rowed deter-
minant, etc.

Corollary. If the elements of two rows {or columns) of a

determinant are proportional, the determinant has the value
zero.

Eor, each element of one of the two rows (or columns) in

question by hypothesis a multiple, m, of the corresponding
is

element of the other. Thus m can be taken out from the first
of the two rows (or columns) as a factor (Th. 3). The two
rows (or columns) are then identical, and Theorem 4 can be
applied.

Theorem 5. If each element of a row {or columyi) is the su7n

of two quantities, the determinant can he written as the sum of
two determinants.
Denote the determinant by A and the elements of the col-
umn (or row) in question by mi -f m/, m^ -f m^', Denote by ••••

A the determinant obtained by replacing all the m"s in A by

zeros, and by A' the determinant obtained by replacing all the
 DETERMINANTS 375

m's in A by zeros. We shall prove that

A = A + A'.
e
376 ANALYTIC GEOMETRY

tti bi + kai Ci

A' = ao ^2 + ^«2 ^2

ttg 63 + fcaa C3
 :

DETERMINANTS 377

Each pair of integers is in the same order in (2) as it

was in (1), except the pair I, I -{- 1. Consequently, if a pair
of integers in (3), not the pair I, I -\-l, presents an inversion
from the order (1), it also presents an inversion from the
order (2), and vice versa. But the pair I, I -\-l in (3) presents
an inversion from one of the orders, (1) and (2), and not from
the other. Hence, we conclude

Lemma 1. The total inimber of inversions from the order (2),

ivhich (3) presents, differs by one from, the total number of inver-
sions from the order (1), ivhich it presents.

For example, 2 3 14 5 has two inversions from the natural

order 12 3 4 5 and three from the order 12 4 3 5.
In the general determinant of the nth order.

«! &i Cj

0/2 O2 C2

«3 ^3 <h •

A=
• • •

• • • 4

(^n K Cn

.7 t
 :

378 ANALYTIC GEOMETRY

has one sign in A and the opposite sign in A'. Therefore,

^' = _ A. We have thus proved the theorem :

Theorem 6. If two adjacent rows of A are interchanged^ the

sign of A is changed.
Suppose, now, that we carry a row over m rows. This can
be effected by m interchanges of adjacent rows ; for example,

if the row is to be carried downward, by interchanging it with

the row just below it, then with the row just below its new
position, etc. Since each interchange of adjacent rows changes
the sign of A, the final determinant will be equal to A or — A,
according as m
is even or odd. This result we state in the
form of a theorem :

Theorem 7. If a row of A is carried over m rows, the result

is A or — A, according as m
is even or odd.

Finally, interchange If there are m rows

any two rows.
between the two, the interchange can be effected by carrying
one of the rows over these m and then by carrying the second
one over this one and the m, i.e. over m -h 1 rows. Thereby
the determinant experiences m-|-m + l=2m+l changes of
sign, i.e. an odd number. Thus we have the result

Theorem 8. If any two rows of A are interchanged, the sign

of A is changed.
NewRules for Determining the Sign of a Term. We first

state the following lemma :

Lemma 2. Take the first n integers in an arbitrary order:

(4) p q r ' ' i j ' ' ' t,

and in this order interchange two adjacent integers, i andj:

(5) p q r ' '

j i - • t.

Then the number of inversions of (5) from the natural order

differs from the number of inversions of (4) froin the natural
order by one.

The proof of this lemma is exactly like the proof of Lemma 1.

 DETERMINANTS 379

An arbitrary term of A, without its sign, can be written in

the form

(6) v^io^x^ z„
where
(7) V 10 X z

are the letters a h c A: in some order and

(8) p q r t

are the subscripts 12 3 ?i in some order. Let ^be

the number of inversions of the letters (7) from the natural
order and let M be the number of inversions of the subscripts
(8) from the natural order. ^+
Jf is the total number of
inversions in letters and subscripts.
If we interchange two adjacent factors in (6), the effect is to
interchange two adjacent letters in (7) and two adjacent sub-
scripts in (8). Hence, by Lemma 2, is changed by one * N
and M by one consequently, the sum JSf-^Mis changed by
;

2 or left unchanged. But any reordering of the factors in (6)

can be effected by a number of interchanges of adjacent fac-
tors. It follows, then, that any reordering of the factors of
(6) changes N
+ MhjSiJi even number or leaves it unchanged.
That is, the evenness or oddness of the total number of inver-
sions in letters and subscripts in a term of A is independent of
the order of the factors in the term.
We may, therefore, arrange the factors with the letters in
the natural order and count the inversions in the subscripts,
as in the definition, § 3, or Ave may arrange the factors with
the subscripts in the natural order and count the inversions
in the letters, or we may leave the factors unarranged and
count the inversions in both letters and subscripts. The re-
sult will always be even or always be odd, no matter which
of the three methods is used, and consequently the sign to
be given to the term will always turn out to be the same.

* Lemma 2 is stated in terrn^ of integers ; it liolds equally well for

letters.
380 ANALYTIC GEOMETRY

In the above methods of determining the sign of a term the

letters and subscripts (or the columns and rows)
enter sym-

metrically. The columns and rows also play the same roles
in the choice of the factors which constitute the term. In

other words, the formation of a determinant from its square

array bears equally on the rows and columns of the array.

We have, then, the following theorem.

Theorem 9. If the rows and columns of A are interchanged,

A is unchanged.

Consequently, Theorems 6, 7, and 8, which have been proved

for rows, are true also for columns.

Completion of the Proof of Theorem 1. If is the element m

in the ith row and jth column of A, we have to show that

+ mM or — givesmM
terms of A, according as i + j is even

or odd ; or more briefly, that (— ly+^mM always gives terms

of A.
If I
_ 1^ j _. 1^ I.e. if m is
the element in the upper left-hand
corner of A, the natural orders of letters and subscripts in M
are
5 c • • • •
fe and 2 3 • • • • n.

A term T oi M will present the same number of inversions in

letters and subscripts with respect to these orders as the cor-
responding term, a^T, of A presents with respect to the orders

a h c ' ' ' k and 12 3 n.

Hence it is -f mM which gives terms of A and, since i-\-j = 2,

this is in accordance with the theorem.
Consider, now, the general case m in the ith row and Jth
:

column. Carry the ith row over — 1 rows to the top of A

i

and then carry the jth column over j — 1 columns to the ex-
treme left of A. By Th. 7, the resulting determinant is

(9) (- 1)*'^^-^A,

and in it m is in the upper left-hand corner. It follows, then,

from the case first considered, that -fmitf gives terms of (9).
 —

DETERMINANTS 381

Hence (— lY'^'mM gives terms of

and therefore of A, since 2i + 2j — 2 is even, q. e. d.*

EXERCISES
1. Prove Lemma 2.

2. Determine, by each of the three methods above described,

the signs to be given to the following products :

(a) ^sCitta; (6) CzaiCkbi; (c) ^563610402.

8. Cramer's Rule. In § 2, we stated that the three simul-

taneous equations in three unknowns,

'
«i« + biy + Ciz = fci,

(1) oto^+ ^22/ + ^23 = Ajg,

(hX + 632/ + CsZ = ^^3,

have the solution

|A:?>c| _\a Ic c\ _l*^^^i

provided | a 6 c\=^0.

This rule for finding the solution of (1) is due to Gabriel

Cramer (1760). We proceed to prove it.

Assuming that equations (1) have a solution, we begin by-

multiplying them respectively by + Ai, — A2, H- A-^, i.e. by the

signed minors of ai, a^, a^ in the determinant \a b c\. Add-
ing the resulting equations, we have
(tti^i A + ^3^3) y
a^Az + ag^g) X + (&i4i — ?>2
—
+ {ciAi C2A2 + 03^3) = kiAi - kiAi + ^3^3-
z

The coefficient of x is the evaluation of |

a b c |
by the minors
of the a's. Similarly, the constant term is |
^' 6 c |
; of . § 4,

* All the theorems of this paragraph have been proved directly from
the definition of a determinant, without the use of any of the preceding
theorems, of §§4, 5, 6. So the paragraph could be inserted immediately
after § 3. Its importance, in comparison with that of §§ 4, 5, 6, is not,
however, sufficient to justify this.
382 ANALYTIC GEOMETRY

Ex. 1. The coefficients of y and z are |

6 6 c |
and \c h c |,

and these determinants are zero (Th. 4). Consequently,

\a h c\x=\'k h c\.

Since we are assuming that \

a h c\=^Of

^ = \a
\k b
^ J.
—
c\

c\ \

This is the value of x, as given by (2). Multiplying the

equations (1) respectively by Bi, — + B2, — B^ and adding,
we obtain the value of y. That of z is arrived at in a similar
manner.
What we have proved is this : If the equations (1) have a
solution, it is given by formulas (2). It follows, then, that
equations (1) have at most one solution, since formulas (2) give
unique values for x, y, z.
It remains to show that these values of x, y, z actually are a
solution, actually satisfy equations (1) in all cases.
i.e. This
can be done by direct substitution. Setting the values into
the first of equations (1) and multiplying through by |
a h c |,

we have
(3) ai\lc b c I
+ 61 1 a k c\ -\- Ci\ a- b A: |
— A:i |a b c\ = 0.
By proper rearrangement of columns in the first two deter-
minants (cf. Ths. 6-8), this becomes

ai\b c k\ —bi\a c ^'

|
+ Ci |
a b k\ — ki\ a b c |
= 0.
The left-hand side here is the evalua.tion of the determinant

ai
 DETERMINANTS 383

the first of the equations (1). In like manner it can be shown

that they satisfy the other two equations.
This completes the proof that equations
(1), provided \a b c\
=^ 0,have a unique solution, which is given by Cramer's rule.
Both the proof and the rule can be generalized to the case of
any number of simultaneous linear equations in the same num-
ber of unknowns. We state the result in general form.

Theorem 10. A nuinher of simultaneous linear equations in

the same number of unknowns, for ivJiich the deterininant of the
coefficients of the unknowns does not vanish, has one and only one

solution, which is given by Cramer^ s rule."*

EXERCISES

1. Deduce the value of y given by (2).

2. Prove that the values of x, y, z given by (2) actually

satisfy the third of equations (1).

Give Cramer's rule for four simultaneous linear equations

3.

in four unknowns. First write down the equations and then

the formulas analogous to formulas (2). No proof is required.

Solve the following systems of simultaneous equations. *

2x- + 3z+ t= 6,
2/ = 0,
Sy-4.z-\-2t- 4
-x + 2y-\-4.z-{-St = -6, 2x = 0,
-\-3z-4.t-\- 3
•
Sx-2y~ 2 + 4^ = -l, -4.x-\-2y +3^+11 = 0,
4.x-^3y-5z-4.t= 8. 3x-4:y + 2z - 5 = 0.
9. Three Equations in Two Unknowns. Compatibility. The
three linear equations,

aix + 6i?/ + ci = 0,
(1) a^x + b^ -f C2 = 0,
as^ + hy + C3 = 0,

* In case the determinant of the coefficients does vanish, the facts are
more complex. For two equations in two unknowns, they are given in
the footnote on p. 360. For a treatment of the general case, cf BOcher, .

Introduction to Higher Algebra, Ch. IV.

 :

384 ANALYTIC GEOMETRY

in the two unknowns

x, y are said to be compatible, or consist-

ent, if have a simultaneous solution. They will, in

they
general, be incompatible, since a solution of two of them will
not, in general, satisfy the third. It is important, then, to
determine the condition for their compatibility.
Two cases arise, according as the minors Ci, C2, C3 of the

elements Ci, C2, C3 in the determinant |

a 6 c |
are not, or are,
all zero. In case they are not all zero, at least one of them
must be different from zero. Suppose that 0^= \ai 62 is 1

not zero. Then the first two of the equations (1) have, by
Th. 10, one and only one solution, namely

(2) »=!^, ^=-r^-

This will be a solution of the third equation if and only if

«3 ^1 C2
I I
— 6.3 I
«! C2 I
+ C3
I
a^ ^2 I
= ^J
or, since the left-hand side here is the expansion of \a b c |
by
the minors of the third row, if and only if

(3)
•
\a b c\ = 0.
before formulating this result as a theorem, we give a
definition.

Definition. The members ai, a^, a^ and b^, b^, b^ are pro-
portional :

ai : a2 : ag = 5i :
62 : ^3?

if and only if there exist two numbers I and m, not both zero, such

that
(4) lai = mbi, la^ = m^g, = "^^s-
?«3

If bi, bo, bs are all zero and we take I = and m any number
=^ 0, equations (4) are satisfiedno matter what values ai, a^, a^
have. In other words, three arbitrary numbers a^, a^, a^, on
the one hand, and 0, 0, 0, on the other, are always proportional.
In particular, ai, 02, a-^ may also all be zero.
Suppose, now, that 61, bz, 63 are not all zero and let bi, for
example, be not zero. If, then, I were 0, we should have,
 DETERMINANTS 385

from the first of equations (4), m= ; but 1 = 0, m = is con-

trary to the definition. Consequently, in this case, I cannot
be 0. Hence we can divide each equation through by I. The
result is the equations

where k has the value m/l. Conversely, if in any given case

there exists a number k, zero or not zero, such that these equa-
tions hold, then ai, 02, a^ and bi, bo, 63 are proportional. Eor,
the equations are but a special case of equations (4), when
I = 1 and m = k. We have thus proved the following theorem :

If bi, bo, 63 are 7iot all zero, the numbers a^, ao, a^ and bi, 62? b^
are proportional if and only if there exists a number k, zero or
not zero, such that
(5) «! = kbi, a2 = kb2, a^ = kb^.
By application of the definition it is easy to show that the
minors d, C2, C3 in the above discussion are not all zero when
and only when ai, «2) <^3 ^iid bi, b2, 63 are not proportional cf. ;

Ex. 2. The foregoing result can be stated, then, as follows :

Theorem 11. If, in the equations (1),

tti : a2 :
03 ^ 61 : 62 :
63,

the three equations ivill be compatible ivhen and only ivhen the
determinant of their coefficients vanishes. TJiey then have one
and only one solution.
The case in which Ci = O2 = O3 = is left to the student j

cf. Ex. 3.

EXERCISES
1. Show that the equations,

aix -f 61 = 0, a2C» + 62 = 0, tti =5^ 0, a2 =^ 0,

are consistent if and only if |

a 6 1 = 0.
2. The proportion ai : ao : a^ = bi : b2 : b^ is valid if and only
if the three two-rowed determinants, which are formed from
the array •^ n r. ^
ai CI2 (X3

^1 ^2 ^3
386 ANALYTIC GEOMETRY

by dropping each column in turn, are all zero. Prove this

theorem.
3. If in the equations (1) a^, a2, aa, bi, b2, 63 are not all zero
and «!, a2, a^ are proportional to bi, b2, b^, the three equations
will be compatible if and only if Ci, C2, c^ are proportional to ai,

02, as and bi, ^2? ^3- They then have infinitely many solutions.
Prove this theorem.

In each of the following exercises determine whether or not

the given system of simultaneous equations is compatible. If
it is, find the solution.

4. Sx- +5 = 0, 4:X-\-3y-2 = 0, 5x-\-Ty-9 = 0.

2/

5. 2x-\- y-l = 0, Sx-2y-{-2 = 0, x-4.y-^ 2 = 0.

6. x-2y-\-3 = 0, -3x-\-6y-9 = 0, 2a; -4?/ + 6=0.

7. 4:X-\-2y-l = 0, 2x+ y-5 = 0, -6x - 3y -\-2 = 0.

8. Theorem. If, in the equations

aiX 4- b^y + CiZ di = 0,-f-

a2X + ^22/ + + d2 = 0,
<^22!

«3^' + hy + -h ^3 = 0,
C32;

a^x + 642/ + C4Z -f ^4 = 0,

the four minors Di, D2, 2^3, D^ of the elements di, ^2, d^, d^ in the
determinant \a b c d\ are not all zero, the equations are com-
patiblewhen and only when \a b c d\=0. TJiey then have one
and only one solution. Prove this theorem.

Determine in each case if the four given equations are com-

patible. If so, what is the common solution ?

2a; + 32/ 4- 4^-5 = 0, 4:X-2y -{-2z 5 = 0,---

5a;-22/ + 3z + 4 = 0, 2x- y+ 2 + 3 = 0,
^'
x + 3y-7z-\-5 = 0, 6x -3y i-3z-\-4. = 0,
3a; _ 42/ + 52 + 2 = 0. -2a; + y- z + 2 = 0.

11. State the generalization of Theorem 11 and the

theorem of Ex. 8 for the case of n +1 linear equations in n
unknowns.
 : :

DETERMINANTS 387

10. Homogeneous Linear Equations. The equations,

a^x + b^y = 0,
-f Ciz

(1) aoX + 622^ 4- ^2^ = 0)

a^x + b^y + c^z = 0,

form what system of homogeneous * linear equations.

is called a
In considering them, we assume that not all the coefficients
f'l) ^1) " '} ^3 ^^6 zero.
Let iCo, 2/0? ^0 be a simultaneous solution of the equations (1).
Then Icxq, kyQ, Jczq, where k is an arbitrary constant, is also a
solution of (1). For, if these values are substituted for x, y, z

in (1), we have
k {aiXQ -h bajo -h CiZq) = 0,
k {a^jXQ+ 622/0 + ^22^0) = 0,
k (a^xo + 637/0 + c-^Zq) = 0.
The three parentheses in these equations all have the value
zero, since Xq, ?/o,
Zq is a solution of (1). Hence the equations
are true, q. e. d.

This proof is applicable to the general case of n homogene-

ous linear equations in n unknowns. Hence we can state the
following theorem
Theorem 12. If Xq, ?/o, Zq, tQ is a simxiltaneous solution• •
•,

of n homogeneous linear equations in the n unknowns x, y, z, ••, •

t, then kxQ, kyo, kz^, •••, Mq, ivhere k is an arbitrary constant, is

also a solution.

An obvious solution of the equations (1) is 0, 0, 0. This

is the only solution, if the determinant |
a b c |
is not 0. For,
equations (1) are a special form of equations (1), § 8, when
ki =
k2 kz =0. If =
a 6 c 0, the latter equations have,
I I
^
by Th. 10, just one solution, given by formulas (2), § 8. But
this solution is 0, 0, 0, since each of the determinants in the
numerators in (2) now contains a column of zeros.
This result is also general
Theorem 13. If the determinant of the coefficients of n homo-
geneous linear equations in the n itnknowns x, y, z, •, t does • •

* Cf. p. 348.
 :

388 ANALYTIC GEOMETRY

not vanish, the only simultaneous solution of the equations is

x=0,y = 0,z = 0,"-,t = 0.
If, then, the equations are to have a solution other than the
obvious solution 0, 0, 0, • • •, 0, it is necessary that the determi-
nant of the coefficients vanish. It can be shown, conversely,
that if this determinant does vanish the equations will have
solutions other than the obvious solution. That is, the follow-
ing theorem is true.

Theorem 14. A system of n homogeneous linear equations in

n unknowns has a solution other than the obvious solution, 0, 0,
0, •••, 0, if and only if the determinant of the coefficients vanishes.

To complete the proof of this theorem in the case of equa-

tions (1),we must show that, if a b c = 0, the equations have | |

solutions other than 0, 0, 0. This we shall do by actually ex-

hibiting such solutions.
By hypothesis, \a b cl = 0. Then two cases arise, as
follows :

Case 1. Not all the mijiors in \a b c\ are zero. In this case

at least one minor in |
a 6 c |
does not vanish. Suppose that
the minor, |ai b2\, in the upper left-hand corner is not zero.*
We proceed, then, to show that equations (1) have a solution
X, y, z, in which z =1 ; i.e. that the equations,

a^x + biy ci = 0, -j-

(2) a2X + boy + Cg = 0,

ci3^ + hy + C3 = 0,

obtained from the equations (1) by setting z = 1, have a simul-

taneous solution for x and y.
Since, by hypothesis, \a b c|=0 and |ai 62 1 =7^ 0? equations
(2) have, according to Th. 11, just one solution, that given by
formulas (2), § 9, namely

(3) ^ = 1^^' 2/ = -|^-

* If in any particular case this minor were zero, the equations and the
terms in them could be rearranged, so that the minor would not be zero.
 DETERMINANTS 389

Consequently, equations (1) have the solution

(4) X = :LJ y
^ =- Cli C2
z = l,
I
«i 62 I I
«! &2

and hence, by Th. 12, the solution

(5) x=\b^ C2\, y = -\ai c^l z=\a^ 62 1,

or, finally, again by Th. 12, the solutions

(6) x = k\bi €2], 2/ =~ ^'1 <^i ^2 1> z = k\ai 62 1.

There are infinitely many solutions given by (6), since k

may have any value. Inasmuch as |
cii ^2 1 =5^ 0, only one of
these solutions is the solution 0, 0, 0, namely the one for which
k = 0. Hence the theorem is proved in this case.
Furthermore, (6) gives all the solutions of (1). To prove
this, let Xq, ?/o, Zq be an arbitrary solution of (1). If Zq = 0,
then 0^0 = 2/0 = 0, since for z = the first two of equations (1)
become
aix H- b^y = 0, a2X + 622/ = 0>
and the only solution of these equations is 0, 0, because
I
tti 621=5^^1 cf. Th. 13. If Zq=^0, then Xq/zq, 2/0 Ao? 1 is a solu-

tion of (1) and Xq/zq, 2/0/^0 is therefore a solution of (2).But

the only solution of (2) is given by formulas (3). Hence it
follows that
X,o_ 5i C2 «i C2 I

2^0 tti 62 ^0 «i ^2!'

or that
x^ = k\b^ C2I, 2/0
= K (X\ C2 ) 2o = ^'|«i ^2!,

where k has a definite value, not zero.

We may state the final result by saying that eveiy solution
of (1) is proportional to the solution (5), meaning, thereby, that
it is given by equations of the form (6) ; cf. § 9, eq. (5).

Case 2. All the minors in\a b c\ are zero. In this case it

follows, by § 9, Ex. 2, that

Ct] t 0\ '.
C\ ^ 0t2 • C?2 • C2 ^^ CI3 '.
O3 • C3.
 :

390 ANALYTIC GEOMETRY

This means that the left-hand sides of equations (1) are pro-
portional to one another. Consequently, all the solutions of
one of the equations are solutions of the other two, and hence
are all the solutions of the system (1).
The equation thus singled out must be one in which the
three coefficients are not all zero. This is true of at least one
of the equations (1), since, by hypothesis, not all the coefficients
in (1) are zero. Let it be true of, say, the first equation

(7) a^x -h biy -f c^z = 0,

and let ai, for example, be not zero.
In solving (7), the values of y and z can be chosen at
pleasure y = Jc, z = 1, and the value of x is then determined.
:

Consequently, all the solutions of (7), and hence of (1), are

given by

(8) x = -k^-l^, y = 7c, z = l.

Here there are tioo arbitrary constants, k and I. We say, then,

that the equations (1) have a two-parameter family of solutions
in this case ; and, in distinction, a one-parameter family of
solutions in Case 1.

The proof of Theorem 14, for n = 3, is now complete. In

the general case the facts and, consequently, the proof are
much more complicated.* See p. 403, footnote.

EXERCISES
1. Prove Theorem 14 for the case n = 2.
2. Prove the Theorem: If x^ y^, ti and X2, 2/2, •••, ^2 ^^^
•••,

two simultaneous solutions of n homogeneous linear equations in

the n unknowns x, y, z, •••, t, then Xi -\- a^, 2/1 + 2/2? ••-} h -h h *^

also a solution. Take first 71 = 3.

3. (Continuation of Ex. 2.) Show, further, that kxi -\- 1x2,

kyi 4- ly2, •", kti -f It 2 is a solution.

* Of. Bocher, Introduction to Higher Algebra, Ch. IV.

i
 DETERMINANTS 391

Solve the following systems of simultaneous equations, ob-

taining all the solutions in each case.
4. 4.x-2y=0, -6x-^3y =0.
5. 3x-^5yi-Sz=0, 4:X- y-^ z = 0, x-\-2y-2z=0.
6. 3x-{-2y-2z=0, 2x-{-Sy-z = 0, Sx-[-7y-5z=0.
7. x-2y-\-3z = 0, -Sx+6y-9z=0, 2x-4cy-\-6z=0.

11. Applications

11. The Straight Line. Equation of the Line through Two

Points, in Determinant Form. Let (x, y) be an arbitrary point
on the line determined by the two points (x'l, y^), (x2, 2/2), and let

(1) Ax + By + C=0
be the equation of the line. Since (x, y), (xi, yi), (x2, 2/2) lie

on the line, we must have

Ax -{-By -{-C^O,
(2) Ax, -\-By,+ C=0,
AX2 + 5^/2 + O = 0.
These equations are linear and homogeneous in the three un-
knowns A, B, C. They have a solution for A, B, C other than
the obvious solution 0, 0, 0, inasmuch as there is a line (1) on
which the three points (x^y), (x,, y,), {x^, 2/2) lie- Consequently,
by Th. 14, the determinant of the coefficients vanishes.
In other words, every point (^, y) or, on dropping the dashes,
every point {x, y) on the line satisfies the equation

X y 1

(3)
'
X, y, 1 =0.
X2 2/2 1

By a careful retracing of the steps, it can be shown, con-

versely, that every point {x, y) satisfying (3) lies on the line.
It would follow, then, that (3) is the equation of the line. We
shall adopt, however, a quite different method to prove this,
 :

392 ANALYTIC GEOMETRY

regarding the foregoing work as primarily of value in furnish-

ing us equation (3).
To show that equation (3) represents the line through
(^1) 2/i)j fe 2/2)? develop the determinant by the minors of the
first row. Equation (3) then takes on the usual form (1) of a
linear equation in x and y ; moreover, the values obtained for
A and B
^ = ^1 - 2/2 B = X2-Xi,
are not both zero, since the given points do not coincide.
Consequently, (3) represents some straight line.
This line is the required line, if the coordinates of the given
points satisfy (3). They do, for, if we replace x, y in the de-
terminant by Xi, yi, or by X2, 2/2? two rows of the determinant
will be identical and hence the value of the determinant will
be zero.

Three Points on a Line. Let the three points, which we

assume are distinct, be (x^, 2/1), {x^, y^), {^3, 2/3)- The equation
of the line through the second and third is, according to (3),

X y 1

(4) X2 y2 1 =0.
^3 2/3 1
The first point lies on this line if and only if (xi, 2/1) satisfies

(4), i.e. if and only if

^1 2/1

(5)
 :

DETERMINANTS 393

They are parallel, by Ch. II, § 10, Th. 3, if and only if

All Bi= A2 :Bo = A^. Bi,

i.e. if and only if

(7) ^1 ^2 ^3: : = ^1 : ^2 ^3. :

Suppose, now, that the three lines go through a point.

This means, analytically, that the equations (6) have a common
solution for x, y, i.e. are compatible. Hence, it follows, by
Th. 11, since in this case (7) cannot hold, that ^ B C = 0. | |

The determinant \
A
B C\ vanishes also when the three
lines are parallel,since then (7) is valid and the first two
columns in the determinant are proportional.
Conversely, if \A B (7 1 = 0, the lines (6) are parallel or
concurrent. For, if the determinant vanishes by virtue of the
first two columns being proportional, (7) holds and the lines are
parallel. On the other hand, if (7) does not hold, equations
(6), by Th. 11, are compatible and this means, geometrically,
that the three lines have a point in common.
We have thus proved the theorem
Theorem 16. Tlie three lines (6) are concurrent or parallel
if and only if the determinant of their coefficients vanishes:

\A B (7 1 = 0.

EXERCISES
Eind the equations of the following lines in determinant
form.
1. The line through (xi, 2/1) with intercept b on the axis
of 2/.

2. The line with intercepts a and b.

Find, in determinant form, the equations of the lines re-

quired in the following exercises of Chapter II. E-educe the

equation each time to the usual form.
3. Ex. 1, § 1. 4. Ex. 2, § 1. 5. Ex. 4, § 1. 6. Ex. 6, § 1.

7. Ex. 7, § 1. 8. Ex. 10, § 1. 9. Ex. 1, § 5. 10. Ex. 3, § 5.

394 ANALYTIC GEOMETRY

By the method of this paragraph, do the following exercises

at the end of Chapter III concerning three lines through a
point or three points on a line.

11. Ex. 1. 12. Ex. 2. 13. Ex. 3.

14. Ex. 4. 15. Ex. 5. 16. Ex. 6.

Are the lines given in the following exercises concurrent ?

parallel ?

17. Ex. 4, § 9. 18. Ex. 5, § 9. 19. Ex. 7, § 9.

12. The Circle and the Conies. Equation of the Circle through
Three Points. If the three points (x^^, yi), {x2, 2/2)) (^3> 2/3))

which we assume are not collinear, lie on the circle

(1) A{x^ + y'^)+Bx + Cy + D = 0,

it follows that
A(x^ + 2/1')+ Bx, 4- + i) = 0, (72/1

(2) A {x,^ + + Bx, +

2/2^) + i) = 0, (72/2

W
^ + 2/3-)+ ^^3 + Cy, + D = 0.
In (1) and (2) we have four homogeneous linear equations
in the four unknowns A, B, C, D, which have a solution other
than the obvious solution, 0, 0, 0, 0. Consequently, by Th. 14,
x^ +2/2

(3)
 DETERMINANTS 395

it has no locus ; cf. Ch. IV, § 2. That it represents a circle,

and, in particular, the required circle, is clear since the coor-
dinates of each of the three points satisfy it.

Condition that Four Points Lie on a Circle.

Theorem 17. The four points (iCi, y^), {x,, y^), (x^, y^), (x„ y^),
ofivhich we assume no three collinear, lie on a circle if and only if
xi" + yi' Xi 2/1 1

«2^ + 2/2^ X2 2/2 1

(4) = 0.
x^ 2/3 1

x^^2 + 2/4- X4 2/4 1

The proof is left to the student.

The general equation of the

Conic through Five Points.
straight line (1),11 contains three constants, A, B, (7, enter-
§
ing homogeneously —
one in each term and we can always —
pass just one line through two Also, the general points.
equation (1) of
the circle contains four homogeneous constants
A, B, C, B, and through three points (non-collinear) we can
always pass just one circle.
The general equation of a conic,

(5) Ax'^ -f Bxy -f Cy^ + Dx -\- Ey -f F= 0,

contains six homogeneous
constants, and accordingly we
should expect that through five points we can, in general pass
,

just one conic.

We prove this by writing down the equation of the conic
through the five points («i, y^), {x., y.,), {x^, 2/3), (^'4, 2/4), (a^s, 2/5)-
Proceeding as in the cases of the straight line and circle,
we find as the probable equation :

x'^ xy y2 X y 1
x,^ Xiyi 2/1^ Xl 2/1 1

X2' ^'22/2 2/2^ 1

(6)
2/2
= 0.
Xs'^ xm 2/3^ Xs 2/3 1

Xa^ ^42/4 2/4^ Xa 2/4 1

X. ^52/5 2/5' X. 2/5 1

396 ANALYTIC GEOMETRY

When the determinant is developed by the minors of the

first row, equation (6) takes on the form (5). Two cases then
arise, according as the values obtained for A, B, C are not, or
are, all zero.

Case 1. A, B, C not all zero. In this case it follows that

equation (6) represents some conic, in particular, a conic
through the five given points, since it is clear that the coordi-
nates of each of the points satisfy the equation.
We state, without proof, that this case occurs unless four,
or all five, of the given points are collinear.
If no three of the points are collinear the conic just found
must be non-degenerate. It is the only conic through the
five points. For, if there were a second conic through them,
the two conies (both non-degenerate) would intersect in five
points, and this is impossible.*
If three of the points are collinear, the conic found must
be degenerate ; f in particular it must consist of two straight
lines (Fig. 1). Clearly, these lines
are uniquely determined by the five
points and hence so is the conic.
„
FlG.
.
1
The results of this case we formu-
late as a theorem :
'

Theorem 18. Through Jive points, no four of which are

collinear, there passes one and only one conic. If three of the
points are collinear, the conic is degenerate ; otherwise, it is non-
degenerate.

Case 2. JL = 5 = C= 0. Then D = E = F= also, and

equation (6) reduces to the trivial equation = 0. Stated
:

without proof.
This case occurs if at least four of the five points are col-

* That two non-degenerate conies cannot intersect in more than four

points is geometrically evident ; an analytical proof is beyond the scope of
this book.

t If it were non -degenerate, we should have a non-degenerate conic

intersected by a line in three points —
an impossibility.
 : :

DETERMINANTS 397

linear. If just four are collinear, there are infinitely many

degenerate conies through the five points, each consisting of
the line of the four points and some line through the fifth.
If all five points are collinear, their line, taken with any
line in the plane, forms a degenerate conic through them, so
that here, too, there are infinitely many degenerate conies
through the five points.^

Parabolas through Four Points. Demanding that the conic

defined by equation (5) be a parabola puts one condition on
the coefficients in (5), namely,

(7) 52_ 4^(7^0.

Consequently, we cannot prescribe more than four points
through which a parabola must pass.
Let (0, 0}, (1, 1), (- 1, 1), (3, 9) be the four points. Then

,gx ^+ B+ C+ D+ E+F=:0,
^^ A- B+ C- D+ E + F=0,
9A+27B-\-81C-\-3D-\-9E + F=0.
To solve equations (7) and (8) simultaneously, find the values
of D, E, F in terms of A, B, C from the first three of equa-
tions (8)

(9) D = -B, E = -A-0, F=0,

and substitute them in the fourth equation. The result is

(10) B = -3a
Hence (7) becomes
9C2-4.4C=0,
and C=0 or C=iA.
From equations (9) and (10) Ave have, then

C=0, 5 = 0, D = 0, E = -A, F=^0,

otC=^A, B = -^A, D = ^A, E = --\^-A, F=0.
* There a one-parameter family of degenerate conies in the first
is

case, a two-parameter family in the second; cf. p. 390. Can the student
explain why ?
398 ANALYTIC GEOMETRY

Setting A=l in. the first case and ^=9 in the second, we
find as the resulting equations

x^ — y z= 0,
9a;2 _ 12xy + 4 12 - ISy = 0.
?/2 -f- cc

There are, then, two parabolas through the four given points.
We state without proof that this is, in general, true. Of
course, one or both of the parabolas may be degenerate, and
for special positions of the four points the two may coincide.
Finally, if the four points are collinear, there are an infinite
number of degenerate parabolas through them.

EXERCISES
1. State and prove the theorem giving the condition that
six points, no four of which are collinear, lie on a (non-degen-
erate or degenerate) conic. If four or more of the points are
collinear, is there a conic through the six ?

Find, in determinant form, the equations of the circles re-

quired in the following exercises of Chapter IV. Eeduce the

equation each time to the usual form.

2. Ex. 1, § 4. 3. Ex. 2, § 4. 4. Ex. 3, § 4.

In each of the following exercises determine whether or not

the four given points lie on a circle.

5. (0, 0), (3, 0), (0, 1), (2, - 1).

6. (2,0), (-3,0), (0,4), (-1,4).

7. (a, 0), (6, 0), (0, c), (O,

^|
Find, in each exercise that follows, the equation of the conic
through the given five points. Is the conic non-degenerate ?
8. (0, 0), (2, 0), (0, 2), (5, 2), (2, 5).
.4ns. 2a;2-3.T2/4-2?/2-4a;-4?/ = 0.
9. (1,0), (-1,0), (0,1), (0,-1), (1,1).
 DETERMINANTS 399

10. (1, - 1), (1, 1), (3, 11), (-3, - 11), (5, 19).
Ans. 15 a;2— 2/2 — 14
11. (0,0), (2,1), (3, -4), (0,2), (-2,0).
12. (1, 2), (0, 1), (6, - 1), (- 1, - 2), (3, 0).

Find the equations of the parabolas through each of the

following sets of four points.Are they degenerate or not ?
13. (0, 0), (1, 1), (1, - 1), (4, 2).
A71S. y^ — x = 0; (x — y){x — y — 2)=0.
14. (0, 0), (3, 1), (1, 3), (6, 3).

15. (2, 0), (0, 1), (- 1, 1), (5, - 2).

16. (2, 1), (7, 0), (4, 3), (5, -2).
In each of the following exercises determine whether or not
the six given points lie on a conic. If they do, find if the
conic is degenerate.

17. (0, 0), (1, - 1), (1, 3), (5, 5), (2, 4), (6, 3).
•

18. (- 1, - 1), (0, 2), (- 1, 0), (5, 2), (0, - 1), (9, 5).

19. (0, 1), (1, 0), (1, - 1), (3, 1), (-1, 3), (-3, -2).
20. (0, 0), (2, 0), (- 1, 1), (3, 1), (5, - 1), (- 4, 2).

EXERCISES ON CHAPTER XVI

Evaluate each of the following determinants, expressing the
result, if it is different from zero, in factored form.

1
1.

4.
400 ANALYTIC GEOMETRY

Prove that the following determinants have the value zero.

6. tto ^2 2 a, + 3 62
ag ^3 2 a-a -I- 3 63

1 3% -61 + 2
1 3 a2 — b2 — 2
7.
Q 303-63 + 4 «3 ?>3

1 3a, -54 + 2 a. ^>4

Definition. Let jh, Pi, -^l^n, Qi, Q2, -, Qn, and ri, rg, •-, r„ be
three columns (or rows) of a determinant. The third is said
to be a linear comhination of the first two, if two numbers,
k and Z, exist such that

ri = kpi + ?5i, r2 = /cpa + Iq-i, , ^n = 'kPn + ^^n-

In the determinant of Ex. 6, for example, the third column is
a linear combination of the first two and in that of Ex. 7 the ;

second column is a linear combination of the third, fourth, and

first.
*

8. Theorem. If one column, or roiv, of a determinant is a

linear combination of two others, the value of the determinant is
zero. Prove this theorem. How can it be extended ?

Solve the following equations for x.

x +1 4 2
9. aj-9 5 -3 = 0.
x-1 -1 1

ic —5
10.
 .

DETERMINANTS 401

kx-\-Sy-{- = 0,z Jcx— y-^ z = 0,

11. x-^4:y-3z = 0, 12. — 2y +
4:X kz = 0,
kx-^ y + 3z = 0. 6x-3y+(k + l)z = 0.

Determine k so that the following equations are compatible.

Find the common solution in each case.
2x — 3y — k = 0, X— ky + 2k=:0,
13. kx - y-k = 0, 14. kx— 4.y 5 =0, -\-

kx -^3y-5 = 0. x-\-2y-l =0.

Find all the solutions of the following equations.
Xy —z = 0,
-}- 2x-\-3y-\-6z = 0,
15. 16.
"""
2x-y -z=:0. 3x-6y-^2z = 0.
17. Show that all the solutions of the equations,

liX + niiy 4- ^iZ = 0,

I2X -\- mzy + n2Z = 0,
are given by x y
: -.
z = \m n\ :\n l\ : \l m\,
provided not all three of the determinants on the right are zero.

Applications
18. Show that the area of the triangle with vertices at the
points (xi, y{), {x^, 2/2), (^'3, 2/3) is

•'^'1
2/1 1

±^ ^2 2/2 1

^3 2/3 1

19. Prove that the equation of the line of slope X through

the point {xi, y^ can be written in the form

X y 1

^i 2/1 1=0.
1X0
20. Show that every equation of the form
X y 1

ai a^ rt3 = 0,
61 62 ^3
402 ANALYTIC GEOMETRY

where the minors of x and y are not both zero, represents a

straight line.

21. Show that the points (xi, yi), (xg, 2/2) are collinear with
the origin when and only when
^1 Vi
= 0.
X2 2/2

Prove that the distinct

22. lines Li, L2 of Ch. II, § 10, are

parallel if and only if

= 0.
A2 B2

Show that the lines JLi, L2 of Ch. II, § 10, are identical,
23.
ifand only if the three two-rowed determinants, which are
formed from the array
A, B, C,
^2 B2 C2

by dropping each column in turn, are all zero.

24. Show that the discriminant, A, of the quadratic equation

Ax^-\- Bx-{-C=0, A^O,

(cf. Ch. IX, § 5) can be written in the form

B
A= - 2AB 20'
25. Show that the discriminant. A, of the general equation
of the second degree in x and y (cf. Ch. XII, § 4) can be
written as *

2A B D
A^i B 20 E.
D E 2F
26. Prove that the polars of all points (having polars) with
respect to a degenerate conic are concurrent or parallel.
Suggestion. The conic can be represented either by
ax^ -|- 62/^ = or by y^ = c.
 DETERMINANTS 403

27. By applying Ex. 26, show that the general equation of

the second degree represents a degenerate conic when and
only when its discriminant, as given by the determinant in
Ex. 25, vanishes.
Suggestion. Demand that the polars of three non-collinear
points, as (0, 0), (1, 0), (0, 1), be concurrent or parallel. The
equation of the polar of (x^, y^ is that of Ex. 2, p. 188.

28. Prove that, if the general equation of the second degree

represents a non-degenerate conic, the line ax -{- by -\- c = ^ will
be tangent to the conic if and only if

2A B D a
B 2C E b
= 0.
D E 2F c

a b c

Suggestion. Apply the second method of Ch. IX, § 5. The

equation of the tangent at {Xi, y^) is given by Ex. 2, p. 188.

Note to p. 390. Theorem 14 leads to an important result concerning

the compatibility (cf. § 9) of equations (2), p. 388, namely :

Theorem. If equationfi (2) are compatible, the determinant of their

coefficients vanishes.

For, if equations (2) have a solution, Xq, yo, then equations (1) have
a solution, Xo, yo, 1, not the obvious solution, 0, 0, 0. Consequently, by
Th. 14, I
a & c I
= 0.
The extension of the theorem and the proof to the equations of § 9,
Ex. 8, and to the general case of § 9, Ex. 11, is immediate.
The determinant of the coefficients of the equations of § 9,
Ex. 7,
vanishes; the equations are, however, incompatible, they represent—
three parallel hues. In other words, the converse of the theorem is not
true ; cf. Th. 11.
 SOLID ANALYTIC GEOMETRY

CHAPTER XVII
PROJECTIONS. COORDINATES

1. Directed Line-Segments. In the Introduction to Plane

Analytic Geometry directed line-segments on a line L were
defined and discussed. Since L might be situated anywhere in
space, the theory there developed holds equally well for the
geometry of space. The student should review the details of
this theory. Of the formulas, let him recall in particular the
relation,

(1) MM, + M,M^ 4- ...

+ Jf„_o Jf„_i + M^_,N = MN,
which holds for any n -\-l points, M, M,, M2, •••
, M^_i, N,
lying on L.

2. Projection of a Broken Line. Given a point P and a line

L in space. The projection of P on i^ is defined as the foot, 31,
of the perpendicular dropped from P
on L, or as the point 3f in which the
plane p, passing through P perpen-
dicular to L, meets L. If P lies on
L, it is its own projection on L.
Let PQ be any directed line-seg-
ment in space and let and Mbe N
the projections of P and Q on L-
The projection of the directed line-
segment PQ on L is defined as the
directed line-segment MN. Fig. 1
405
 :

406 ANALYTIC GEOMETRY

If p and q are the planes through P

and Q perpendicular to
L, the projection, MN, of PQ
on L is equal to the directed line-
segment intercepted by the planes p and q on any parallel to L.
For example, it is equal to the directed line-segment PR in

Fig. 1*
Consider a broken line joining P
to Q and consisting of the

directed line-segments PPi, P1P2, -, P„-iQ, which do not

necessarily lie in a plane. The sum of the projections of these

directed line-segments is

JO/i -t- M^M^ + .- + M,_^N.

By (1), § 1, this sum is equal to MN, i.e. to the projection on
L of the directed line-segment PQ.
Thus Theorem 1 of the Introduction, § 3, is extended to the
geometry of space
Theorem. The sum of the ^projections, on any line L of space,
of PP^, P1P2, •-, P^-iQ, of any broken
the directed line-segments,
line joining a point P of space icith a second point Q is equal to
the projection on L of the directed line-segment PQ.

Theorem 2 of the Introduction, § 3, may be extended in a

similar manner. Let the student state and prove the result.

The projection P on a plane is defined as the

of a point K
foot of the perpendicular dropped from P on K. If P lies in
K, it is its own projection on K.
Let a plane K
and a line L be given. If L is not perpen-
dicular to K, the projection of i on ^is defined as the line in
which the plane through L perpendicular to intersects K. K
If L is perpendicular to K, the projection of L on /f is merely
a point, the point in which it meets K.

i 3.yihe Angle between Two Directed Lines. Given any two

indefinite straight lines in space and on each of them a sense ;

to define the angle between these two directed lines.

*In drawing this figure, we have placed ourselves in space so that the
plane through L and P appears to us as a vertical plane.
 PROJECTIONS. COORDINATES 407

If the lines meet, they lie in a plane. The angle between

them shall be defined as the angle between the half-lines, or

rays, issuing from their point of intersection

in the given directions (Fig. 2).

If the lines do not meet, choose an arbitrary

point A of space, and draw from A two rays
'^
^^^ ^
respectively parallel to and having the same
senses as the given lines. The angle between the given lines
shall be defined as the angle between these rays.*

Remarks. The angle is the angle between the directed lines,

not the angle from one to the other. It has always a positive
or zero value, i.e. a numerical, and not an algebraic, value.
It is futile, in the geometry of space, to try to distinguish
between positive and negative angles. For instance, suppose
that, in an attempt to define the angle from one of two directed
lines lying in a plane to the other, we should agree that angles
measured in the counter-clockwise sense are to be considered
positive and those measured in the clockwise sense, negative.
Then the angle /rom the one directed line to the other, if viewed
from a certain side of the plane, would appear positive but, ;

viewed from the other side of the plane, the same angle would
be negative. Viewing the angle from one side of the plane is
as justifiable as viewing it from the other, since the plane is im-

mersed in space and not displayed on a blackboard or on the

page of a book. Consequently, we should still be at a loss as
to whether the angle is positive or negative.
There are two angles between the rays shown in Fig. 2, namely,
and 360° — 6. One of these is necessarily less than or equal
to 180°. It is this angle which we agree to take as the angle be-
tween the directed lines.

* For example, in Fig. 1, the line of PQ, directed from P to Q, and i,

two directed lines. The angle between them is

directed to the right, are
the angle 6 constructed by choosing A on the first line, at P, and by draw-
ing through P the line L' parallel to and having the same sense as L.
408 ANALYTIC GEOMETRY

EXERCISES

1. Fasten a sheet of paper to the floor with one edge against

the wall, and tack a second sheet to the wall with one edge
along the floor. Draw on each sheet a directed line so that the
two lines meet. (It is wise to draw the lines before fixing the
sheets in position.) Crease a third sheet of paper so as to form
an angle which will just fit between the two directed rays. By
measuring this angle with a protractor determine the angle be-
tween the two directed lines.
2. Repeat Ex. 1 with two directed lines differing widely in
position from the first two chosen.
3. What is the angle between the two lines of Ex. 1, if the
sense on one of them is reversed ?

4. By the method of Ex. 1 find the angle between two

directed lines, one on the floor and one on the wall, if the two

lines do not meet.

5. that, if L and L' are any two lines in space and any
Prove
plane F is
passed through L, there will be a plane through W
L' perpendicular to F. That is, show that the above method is
applicable to the problem of determining the angle between
any two directed lines.

(4.
X 4. Value
yj of the a Directed Line-Segment.
Projection of
Asst^ to a line L space a
of sense and adopt a unit of
length for all measurements in space. Then a directed
line-segment AB
on L is represented by an algebraic num-
ber, equal numerically to the length of AB
and positive or
negative according as the direction from to ^ 5
is the same

as, or opposite to, the direction given to X; cf. Introduc-

tion, § 2.

In particular, to the projection MN

on i of a directed line-
segment PQ corresponds a certain algebraic value or number,
which we can, without confusion, denote also by MN. Clearly,
Proj. PQ= -Proj. qP.
 PROJECTIONS. COORDINATES 409

Let the length PQ of the directed line-segment PQ be given

| |

and also the angle which the line of PQ, directed from P to
Q, makes with the directed line L. If (

PQ lies in a plane with i, we know from

Plane Trigonometry that

(1) MN= Proj.^ PQ=\PQ\ cos $.

The general case, in which PQ is not m p^^ 3

a plane with L, is shown in Fig. 1. Con-
sider the projection, PE, of PQ on the line L' through P,
parallel to and having the same sense as L. Since PQ lies in
a plane with L', we have the previous case. Consequently,

PR = \PQ\ cos 0.

But PR = MN and thus formula (1) is established in the gen-

eral case.

EXERCISES
1. Draw Fig. 1 for various positions of P and Q and in each

case verify formula (1).

2. By application of (1) verify that Proj. PQ = - Proj. QP,

3. If P and Q lie in a plane perpendicular to L, M and N
coincide and MJSf= 0. Prove this by applying formula (1).

4. Prove that the directed line-segments 3fN and M'N',

which are the projections on L of two directed line-segments
PQ and P'Q' on the same line, are proportional to PQ and P'Q' :

MN ^ PQ
M'N' P'Q''

5.Prove that the theorem of the preceding exercise is true

if PQ and P'Q' are on parallel lines.

^^5^ Coordinates. Three directed lines drawn through a point

of space, so that each is perpendicular to the other two,
form a sy stern of rectangular coordinate axes. The coordinates
of a point P of space with respect to the system of axes are
 .

410 ANALYTIC GEOMETRY

defined as the numbers which represent algebraically the

projections of the directed line-segment OP on the three
directed lines. Thus, if Ox, Oy, Oz
denote the three directed lines and a;, y, z
stand for the coordinates of P,

X = Pro j Ox OP, y = Vm].oy OP,

z = Proj.o^ OP.

The projections of OP on the three

-^ directed lines can be constructed by pass-
'*
F 4 ^^ planes through P perpendicular to
the three lines (Fig. 4). These form with
the planes of the lines a rectangular parallelepiped, or box.
The directed edges of the box which issue from are the
projections of OP.
Every point P of space has unique coordinates (x, y, z). Con-
versely, if any three numbers x, y, z are given, there is a
unique point P having these numbers as its coordinates. This
point can be located, either by constructing a box or, more
simply, by laying off OM
= a; on the axis of x, then =y MN
on a parallel through M
to the axis of y, and finally NP = z on
a parallel through J^ to the axis of z, as shown in Fig. 4. It
is to be remembered that OM, MN, and NP are directed line-

segments. The direction of OM, for example, is the same

as, or opposite to, that of Ox, according as the number x
is positive or negative. The figure is drawn for the case that
X, y, z are all positive.
The point is the origin of coordinates, the directed lines
Ox, Oy, Oz are the coordinate axes, and the planes xOy, yOz,
zOx are the coordinate planes. The origin has the coordinates
(0, 0, 0), a point on a coordinate axis always has two of its

coordinates zero, and a point in a coordinate plane always has

one zero coordinate. Thus the point on the axis of y three units
distant from in the positive direction has the coordinates

(0, 3, 0), and the point in the {y, 2;)-plane, whose coordinates
in that plane are y = 2, z — 2>, has the coordinates (0, 2, 3).
 PROJECTIONS. COORDINATES 411

Octants. Bounded by the coordinate planes there are eight

regions, called octants. It is clear that, if the a>co6rdinate of
one point of an octant is, for example, negative, the a>coordi-
nates of all points of the octant are negative ;
similarly, for
the y- and 2-coordinates. Thus we can speak of the octant

( — ,
-f-, -[-), and mean thereby that octant in which the
aj-coordinate of every point is negative and the y- and 2;-coor-
dinates are both positive. The octant (+, -h, +) is known as
the first octant ; we make no attempt to number the others.

Figures. In drawing a figure in a plane to represent a

figure in space,we make use of what is known as a parallel
projection. The axes of y and z are represented by two per-
pendicular lines and the axis of a? by a line drawn in a con-
venient direction. All distances in the (y, 2)-plane or in any
parallel plane are drawn to scale, so that a figure in such
a plane appears as it actually is in space. Distances on or
parallel to the axis of x are foreshortened a convenient amount.
The direction of the line representing the axis of x and the
amount of foreshortening along this axis depend largely on
the figure in space which is to be represented. In general,
however, we shall draw the line representing the axis of x at
an angle of 120° with that representing the axis of y and take
as the unit distance on the axis of x three-fourths the unit
distance on the other axes.

Bight-Handed and Left-Handed Coordinate Systems. The

system of axes in Fig. 4 is the one we shall employ. Another
system in common use is shown in Fig. 5. The
essential difference between the two is this :

If,from a point on the negative axis of a;, we

view the (2/, z)-plane, the direction of rotation
from the positive i/-axis to the positive 2;-axis
is that of a right-handed screw in case of the
first system and that of a left-handed screw in
case of the second. Accordingly, the system we are using is

called a right-handed system ; the other, a left-handed system.

 ; ;; ; ;

412 ANALYTIC GEOMETRY

Any other rectangular system of axes is essentially the

same as one or the other of these that is, it is either right- ;

handed or left-handed, by the above test (cf. Ex. 7).

EXERCISES
1. Plot the following points, drawing the line representing

the axis of x at an angle of 120° with that representing the

axis of y, and taking ^ in. as the unit on the axes of y and z

and I in. as the unit on the axis of x.

(a) (0, 3, 0) ; .
(b) (0, 1, 3) ;
(c) (2, 5, 0) ;

(d) (4, 0, 0) ;
(e) (0, - 2, 0) (/) (4, 1,3);
(5, - 2, 4) (h) (3, 2, - 5) \i) (- 2, 3, U)
(g)

(i) (1,-1, -3); (fc)(-2,4, -3); (l) (-1,-1, -2).

2. Determine the coordinates of the point P in Fig. 4 when
the units on the axes are taken as in Ex. 1.

3. The same for the point marked by the period in " Fig. 4,"
if this point is i a unit above the (x, 2/)-plane.

4. What are the coordinates of the projections of each of

the following points on the coordinate axes ? On the coordi-
nate planes ?

(a) (3, 5, 2) ; (p) {- 3, 2, - 1) (c) (x, y, z).

5. What equation is satisfied by the coordinates of those

points and only those points which lie in the (y, 2;)-plane?
In the (z, ic)-plane ? In the (ic, 2/)-plane ?
6. What two equations are satisfied by the coordinates of
those points and only those points which lie on the a>axis ?

On the 2/-axis ? On the 2;-axis ?

7. Through a point draw three mutually perpendicular

lines, which, when directed, are to serve respectively as the
axes of x^ y, and z. Show thg,t there are eight possible com-
binations of the directions which can be given to the lines and
that, of the eight resulting systems of axes, four are right-
handed and four, left-handed.
 —
\ • PROJECTIONS. COORDINATES 413

I 6. Projections of a Directed Line-Segment on the Axes. Given

tW^poii^ts Pi and P2 ^^'ith the coordinates (xi, yi, Zi) and

(x2, 2/2) ^2)' To determine the projections of the directed line-
segment P1P2 on the coordinate axes, project the broken line
P1OP2 on each of the axes in turn. Since always

Proj. P1P2 = Proj. PiO -f Proj. OP2,

it follows that
Proj. PiPo = Proj. OP2 - Proj. OP^.

But the projections of OP2 and OPi on the three axes are, by
definition, the coordinates of P2 and Pi. Consequently, the
projections of the directed line-segment PiPo on the three
axes are, respectively,

(1) X2-X1 2/2-2/1 ^2-2^1-

Bypassing planes through the points Pi and Po perpen-

dicular to the three axes, we obtain on the axes the actual
projections, XiXo, ^1^2) ^1-^2?
of the directed line-segment
Po.
P1P2.* The planes also deter-
mine a rectangular parallele-
piped, or box, whose three
R— > I

dimensions are equal to the V

/il /Y^
numerical values of the three 1

projections. Accordingly, the rf

edges of the box, when properly
directed, are precisely equal to Fig. 6
the projections. In particular,
the three edges emanating from Pi, i.e. the directed line-
segments PiP, PiS, PiT, are equal respectively to the three
projections X1X2, Y1T2, Z1Z2.

EXERCISES
1. Plot PiPo when Pi the point (h) of Ex. 1, § 5, and Pg
is

is (c). Determine the projections from the figure and verify

by applying formulas (1).
* To keep the figure simple, only XxX^ and ^1^2 are shown.
 : ; ;

414 ANALYTIC GEOMETRY

2. The same when

i) Pi is (c) and P^ is (/) ii) Pi is (/) and P^ is {g).

3. The projections of P1P2 on the axes are 2, — 5, 3 and

those of P2P3 are — 3, 2, 1. What are the projections of P1P3 ?

Justify your answer.
4. If the points Pi, P, S, T of Fig. 6 have, respectively,
the coordinates (2, 1, 3), (5, 1, 3), (2, 4, 3), (2, 1, 6), what are
the coordinates of P2 ?
5. If the projections of P1P2 on the axes are 3,-5,-2
and Pi has the coordinates (2, 1, 3), what are the coordinates

of P, ? .

istance between Two Points. Let the two points be

'ikft-^ints Pi, P2 of § 6. Then the segment P^P^ \& a diagonal
of the box in Fig. 6. It is a simple matter to show tha,t the

square of the length of a diagonal equals the sum of the

squares of the lengtliE of the edges :

p^p^^ = p^m + PiS^ + A^'.

Hence i)'= (x^ - x^y - ViY + (^2 - ^ify
-f- (2/2

and D = V(a:2 - x^f + - + - ^if- (2/2 2/1)' (2^2

Inasmuch as it is the squares of the quantities (1), §6,

which appear here under the radical, it is immaterial that

these quantities have algebraic values, i.e. may in some cases

be positive and in others, negative cf. Ch. ; I, § 3.

EXERCISES
1. Find the distances between the following pairs of points,
expressing the results correct to three significant figures.

(a) (5, 1, 4), (4, 3, 2) (h) (2, -1, 3), (-1, 1, -3) ;

(c) (2, -1, 8), (-2, -3, 5); {d) (3, 6, -2), (5, -1, 4) ;

(e) (2, - 3, 5), (- 1, 4, 5) ; (/) (1, 2, 4), (1,

- 3, 4).

2. Find the distances of each of the following points from

the origin
(a) (4, 2, 8) ; (6) (3,
- 5, - 2) ;
(c) (x, y, z).
 PROJECTIONS. COORDINATES 415

3. Find the distances of each of the points of Ex. 2 from

the coordinate axes.
4. Find the lengths of the projections on the coordinate
planes of the line-segment joining the points (2, 3, 5) and

(5, 6, 7). Draw a figure showing the projections, or line-

segments equal to them.
5. What equation is satisfied by the coordinates of those
points and only those points which on the unit sphere,
lie —
the sphere whose center is at the origin and whose radius is

one unit ?

<^'^\9. Mid-Point of a Line-Segment. Let Pi (xi, y^, Zi) and :

^2 '
(^2) ?/2) ^2) be the extremities of the line-segment PiP2- If
P : (x, y, z) is the mid-point of P1P2, the directed line-seg-
ments PiP and PP2 are equal and have, therefore, equal pro-
jections on the coordinate axes. Thus we have, by (1), § 6,
X "~- tt/j — X2 ~~~ X,

and similar equations in the y- and ^-coordinates. Hence,

\ J
2 2 2

This result can be stated in words as follows : Tlie coordi-

nates of the mid-point of a line-segment are, respectively, the

averages of the corresponding coordinates of the end-points of
the segment.

EXERCISES
'
1. Determine the coordinates of the mid-point of each of
the line-segments given by the pairs of points in Ex. 1, § 7.

Draw figures and check your answers.

2. Show that the sum of the squares of the diagonals of the
quadrilateral whose successive vertices are at the four points
(5, 0, 0), (0, 6, 0), (1, 2, 3), (3, - 2, 8) is double the sum of
the squares of the line-segments joining the mid-points of the
opposite sides. N. B. The four points do not lie in a plane.
 =

416 ANALYTIC GEOMETRY

3. Show that the line-segments joining the mid-points of

the opposite sides of the quadrilateral of Ex. 2 intersect and

bisect each other.

9. Division of a Line-Segment in a Given Ratio. Let it be

required to find the coordinates {x, y, z) of the point divid- P
ing the line-segment PiP^ in the given ratio mi/m2. Since
the directed line-segments P^P and PP2 are to be in the ratio
mi/m2, this must also be the ratio of their projections on any
one of the axes (§4, Ex. 4). Accordingly, we obtain the
equation
a; — Xi _ mi

and similar equations involving the y- and 2;-coordinates.

Solving these equations respectively for x, y, and z gives, as

the required coordinates of the point P,

7712 + mi mg + mi mg + 'm>x

External Division. It is sometimes of value to have at

hand formulas giving the coordinates (x, y, z) of a point P
which lies on the line of Pi and Pg, but exterior to the seg-
ment P1P2, and whose distances to Pi and P2 are in a given
ratio 7?ii/m2, not equal to unity. Let the student show that
in this case it is the directed line-segments PiP and P2P
which are in the given ratio mi/mg, and that the type of
equation now obtained is
^^

x— Xi_mi
X — X2 m2

so that the required coordinates of P are

xoN
iZ)' x
WoiCi — miXo
=— „
y
_ m22/i — ^1^2
— ^
^
_ mgZi — mig2
— .

— m2 — mj m2 — mi
J )
^ mg mi
The point P is said to divide the segment P1P2 internally,
in the first case ; externally, in the second. The numbers mi
and m2 entering into the ratio of division do not have to be the
 PROJECTIONS. COORDINATES 417

actual lengths of the corresponding line-segments, but may be

any numbers proportional to these lengths.*

EXERCISES
Find the coordinates of the point on the line-segment
1.

joining (2, — 3, 6) with (5, 4, — 2), which is twice as far from

the first point as from the second. Ans. (4, 4, %).
Find the point on the line through the two points of Ex.
2.

1,which is outside the line-segment bounded by them and is

twice as far from the first point as from the second.
3. Find the point which divides internally the line-segment
from (2, 3, 4) to (o, - 3, 0) in the ratio 3 : 4.

4. The preceding exercise for external division.

EXERCISES ON CHAPTER XVII t

1. Show that the points (2, 4, 3), (4, 1, 9), (10, - 1, 6) are
the vertices of an isosceles right triangle.
2. Prove that the tetrahedron with vertices at the points
(0, 0, 0), (0, 1, 1), (1, 0, 1), (1, 1, 0) is a regular tetrahedron.
3. Show that the points (0, 0, V2), (1, 1, 0), (0, 0, - V2),
(—1, — 1, 0), (2, — 2, 0) are the vertices of a regular pyramid
with a square base.
4. Given the points A, B, C with coordinates (2, — 3, 5),
(4, 2, 3), (6, 7, 1). By proving that AB^BC==AC, show that
the three points lie on a line.

5. Show that the three points of Ex. 4

lie on a line by
proving that their projections on each of two coordinate
planes lie on a line. Justify this method of proof.
6. Determine the point on the axis of y which is equidis-
tant from the two points (3, — 2, 4), (—2, 6, 5).

* Thus, in the case of internal division, if PiP = 100 cm. and PP2 =
25 cm., wi and m^. might be properly and wisely chosen as 4 and 1.
t Exercises 1-6, 14-18 of Ch. XIX, § 1, and Exercises 1-8, 18-22 of
Ch. XX, § 1, may be introduced here, if it seems desirable.
418 ANALYTIC GEOMETRY

7. Determine the point in the (?/, 2:)-plane which is equi-

distant from the three points (3, 0, 2), (2, 3, 0), (1, 0, 0).

8. Two vertices of a regular tetrahedron are at the points

(0, 0, 2V2), (0, 2, 0). If the other two vertices lie in the
(x, 2/)-plane, hnd their coordinates.

.9. A regular pyramid, of altitude h, has a square base

whose vertices lie on the axes of x and y and whose edges are
of length a. What are the coordinates of the vertices of the
pyramid ?
10. If P is the mid-point of the line-segment P1P2, and P
and P2 have the coordinates (3, — 2, 5) and ( — 2, 4, 3) respec-
tively, what are the coordinates of Pj ?

11. If Pdivides the line-segment P1P2 internally in the

ratio 2 3,
: and Pi and P have respectively the coordinates
(1, 4, 3) and (3, 2, — 1), determine the coordinates of P2.
Ans. (6, - 1, - 7).

12. Find the ratio in which the point B of Ex. 4 divides the
segment AC of that exercise. Ans. 1 : 1.

13. A point with a^-coordinate 6 lies on the line joining

the two points (2,-3, 4), (8, 0, 10). Find its other two
coordinates.
Suggestion. Determine the ratio in which the point divides
the line-segment bounded by the two given points.
14. Find the point in which the line joining the two
points (2, —3, 1), (5, 4, 6) meets the (z, a5)-plane.

Ans. (3|, 0, 3i).

15. If the length of the line-segment PiP2 is D and the

lengths of its projections on the coordinate planes are Z)i, Z>2,
i>3, show that

16. Show that the lines joining the mid-points of the

opposite sides of any quadrilateral ABCD intersect and bisect
each other. JSf.B. The points A, B, C, D do not necessarily lie
in a plane.
 PROJECTIONS. COORDINATES 419

17. Show that the sum of the squares of the diagonals of any
quadrilateral is twice the sum of the squares of the line-seg-
ments joining the mid-points of the opposite sides.

18. Prove that the center of gravity (intersection of the

medians) of the triangle with vertices at {x^, y^, z^), (ajj, 2/2, Z2),
(^j 2/3) Z3) lia-s the coordinates

K^i + ^2 + xs), i(yi -f 2/2 + 2/3), i(^i + Z2 + h)'

19. Prove that the lines joining the vertices of a tetrahedron
with the centers of gravity of the opposite faces all go through
a point P, which divides each of them in the ratio 3:1.
20. Prove that the lines joining the mid-points of opposite
edges of a tetrahedron all go through a point, which bisects
each of them. Show that this point is identical with the
point P of Ex. 19.
 CHAPTER XVIII

ECTION COSINES. DIRECTION COMPONENTS

VL^ Direction Cosines of a Directed Line. Given a directed

line L in space ; to find a means of determining or fixing its
direction.
The directed line L makes definite angles, a, y8, y, with, the
positive axes of x, y, z, respectively. If L does not go through
the origin, 0, draw L' through O
parallel to L and agreeing with it in
sense. Then a, ^, y are equal respec-
tively to the angles which L' makes
with the axes (Ch. XVII, § 3). The
angles /?, y are called the direction
a,

angles of the directed line L.

Direction Cosines. The cosines of

Fig. 1 the angles a, /3, y, namely cos «, cos /3,

cos y, are known as the directioyi

cosines of L. Since a, /8, y are, by definition (Ch. XVII, § 3),

angles between 0° and 180° inclusive, they are uniquely deter-
mined when their cosines are given, and conversely. Accord-
ingly, we can use either the direction angles or the direction
cosines to fix the direction of L. We choose the direction
cosines.
Evidently, two directed lines which are parallel and have
the same sense have the same direction angles and the same
direction cosines.

Exercise. If two lines are parallel but have opposite senses,

show that the direction angles of one are the supplements of
420
 : ; ;

DIRECTION COSINES 421

the direction angles of the other, and that the direction cosines
of one are the negatives of the direction cosines of the other.

Example 1. What are the direction cosines of the positive

axis of y ?
Here, a = 90°, ^9=0°, y = 90°

and cos a = 0, cos /8 = 1, cos y = 0.

Examjjle 2. Find the direction cosines of the line bi-
secting the angle between the negative axis
of y and the positive axis of z, and directed
upward.
In this case,

a = 90°, = 135°,
p y = 45° ;

cos a = 0, cos ^ = — i V2, cos y =| V2. Fig. 2

Theorem 1. The sum of the squares of the direction cosines

of a directed line is equal to unity

(1) cos2 a -f- cos2 ^ -h cos2 y = 1.

To prove this theorem, take a point P : (x^, y^, z^ on U
(Fig. 1) so that the direction from Oto P will be the direction
of L\ and consider the projections of the directed line-segment
OP on the axes. These are equal, on the one hand, to the coor-
dinates cCq,
2/oj ^0 P (Oh. XYII, § 5), and on the other, to the
of
quantities OP cos a, OP cos^, OP cosy (Ch. XYII, § 4).
Hence

(2) Xq = OP cos «, 2/o

= OP cos ^, Zq = OP cos y
(3) cos a = Q- = -^, cos/8 COSy = ^0-.
OP' OP ^ OP
R + cos2 y^ = ^o' + .Vo'+
^0'
Thus cos2 a + cos^ ^ .

0P2
But OP"" = x,^ + 2/0' + ^0',

and the theorem is proved.

We have shown, then, that every directed line has definite
direction cosines, the sum of whose squares is unity. The con-
422 ANALYTIC GEOMETRY

verse is also true : Any three numbers, the sum of whose squares
is unity, are the direction cosines of some directed line.
Preliminary to proving this, we revert
to the proof of Theorem 1 and choose
the point P in particular as the point
in which the ray issuing from in the
direction of V
meets the unit sphere
(Fig. 3). Then 0P= 1, or

and (2) becomes

(4) Xq = cos a, yo = cos /3, Zq = cos y.
That is, the direction cosines of a ray
issuing from are equal to the coordinates of the point in which
the ray pierces the unit sphere.

The desired proof is now simple. If there are given any

three numbers, x^, y^, Zq, the sum of whose squares is unity,
they will be the coordinates of some point P of the unit sphere,
and hence they will also be the direction cosines of a certain
directed line, namely, the line passing through V
and P and
directed from O to P, q. e. d.

Example 3. The three numbers -|, -|^, —| are the direction

cosines of some directed line, for

(i)^+(F+(-i?=i+i+i=i-
The direction angles of the line are, respectively, 48° 11',
70° 32', 131° 49'.

Example 4. A directed line makes angles of 60° and 45°

with the axes of x and y, respectively. What angle does it

make with the axis of 2 ?

Here,
cos a = cos 60° = i, cos /3 = cos 45° = ^ V2.
Hence
(i.)2
_^ (i V2)2 + cos2 y = 1 and cos y = ± 1.
 :

DIRECTION COSINES 423

Thus y = 60° or 120°. There are, then, two directed lines

making the given angles with the x- and y-3ixes. The one
makes an angle of 60° with the 2:-axis the other, an angle of
;

120°.

Direction Cosines of the Line through Two Points. Let P1P2

be a directed line-segment lying on the directed line L and
having the same sense as L. If X1X2, Y1Y2, Z^Z^ are the
projections on the axes of P^P^ (Ch. XVII, Fig. 6), we have,
by Ch. XYII, § 4,

X1X2 = Z> cos a, Y^Y^ — D cos /?, ZiZ>i = D cos y,

where D is the length of the segment P1P2'
Hence
r5)cosa=^l^, cos^ = -?iS, cosy = ^?^2.
The content of these equations can be stated in words, as
follows

Theorem 2. If a directed line L is given and on L any

directed line-segment P1P2 having the same sense as L is chosen^
the dii'ection cosines of L are equal to the projections of P1P2 on
the axes, each divided by the length of PiPo-

If Pi and P2 have the coordinates (xi, yi, Zi) and (x2, 2/2, Z2),

formulas (5) become, by Ch. XVII, §§ 6, 7,

(6) cosa = ^2jz_^ cos/? = fc:^, cosy = ^^~^S

where D = ^{x2 - x^f + (2/2 - Vif + {^2 - ^lY-
These are the formulas giving the direction cosines of the line
passing through Pi (xi, 2/1, ^1) and Pg (0^2, 2/2> ^2)) ^'^d. directed
: :

from Pi to P2.

EXERCISES

1. What are the direction cosines of a line parallel to the

axis of z and having the same sense ? Having the opposite
sense ?
 ;: ; ;

424 ANALYTIC GEOMETRY

J
> 2. A line bisects the angle between the positive axes of y
and z and is directed upward. What are its direction cosines ?

\ .' 3. What are the direction cosines of a directed line which

lies in the (2;, x)-plane and makes an angle of 30° with the
positive z-axis ? Two answers.
-^
j
4. A line in the (x, ?/)-plane has the slope V3. What are
'
its direction angles and direction cosines, if it is directed for-
') wards?
5. Construct the directed lines through the origin having the
following direction cosines. What are the direction angles ?

(a) -1,0,0; _ (6)i,iV3,0; (c)-iV3, 0,i;

id) 0, - iV2, W2; (e) - J,

-i, ;
(/) jVS, - ^VS, iV3.

^1 6. Find the direction angles and the direction cosines of a

^
line if
(a) cos a = 1 (6) cos ^=— i cos y = i V3
c (c) cos a = 1, cos ;8 = — I V2.
;

7. Find the direction angles and the direction cosines of a

directed line if

(a) a = 120°, y8 = 60° (b) a = 135°, y = 120°

W ^ = |'V = ^5 (cZ) a = 4o°, /3 = y;
(e) « = ^ = y; (/) «=y=180°-^.
Find the direction cosines of the line passing through the
8.

origin and each of the following points, and directed from the
origin to the point

-1, -4, -1).

o^ (a) (2, 3, 6) ;
(b) (4, 8);

Find the direction cosines of the lines determined by the

9.
(c) (3, 0); (d) (5, 8,

pairs of points in Ex. 1 of Ch. XVII, § 7, if each line is directed

from the first of the given points to the second.
10. A line-segment P1P2 lias the length 6 and the line of Pi
and P2, directed from Pj to Pg, has the direction cosines — |,
I, |. If the coordinates of Pi are ( — 3, 2, 5), what are those
ofP,?
 ;

yy'

DIRECTION COSINES 425

^y
Lgle between Two Directed Lines. Let it be required to
llie angle Q between the two directed lines, L^ and Z2,
whose direction angles are «!, ySi, yi
and ttg, /82, 72-
We can assume without loss of
generality that L^ and L^ pass
through the origin.^ Take any
point P (a'o, ?/o, ^o) o^ ^i? so that the
:

direction from to P is that of Xi,

and draw the broken line OMNP,
whose directed segments OM, MN, M^
NP are, respectively, the coordi-
Fig. 4
nates 0^0, 2/oj ^0 of P' The projection
of this broken line on L2 equals the projection of OP on L^ :

(1) Proj .,^0P= Proj .r^^OM^ Pro j .^^ MN + Proj .^^ ^P.
By Ch. Xyil, § 4,

Proj.x^ 0P= OP cose.

Similarly, Proj.^^ 0M= \0M\ cos ^ /rOxlf,

where K is a point on L2, such that OA" has the direction of

\ L2. I£ the directed line-segment 031 has the direction of the
positive axis of x, as is the case in the figure, we have
\0M\ = OM, IWM= ^ a.,
and therefore,
Proj.^^ 0M= OM cos Oo-

If OM has the direction of the negative axis of x,

0M\ = - OM
1
and ^ KOM= 180° - ag
in this case, then,

Proj.£^ 0M= - OM cos (180° - Wg) = 03/ cos a^.

Consequently, in either case, we have, since 0M= x^,

Proj.x 0M= Xq cos 02-

* For, if they did not, we could consider equally well the angle between
the two parallel lines through the origin having respectively the same
senses as the given Unes.
426 ANALYTIC GEOMETRY

Similarly,

Proj.i^ MN= 2/0 cos y82, P^oj.^^ NP = z^ cos yg.

Thus (1) becomes

(2) OF cos = Xq cos a2 + 2/o

cos jSg + ^0 cos yg.

By (2), § 1,

Xq = OP COS «!, 2/0

= OP COS (3iy Zq = OP cos yi.
Substituting these values in (2) and dividing through by OP,
we obtain, tinally,

(3) cos 6 = cos «! cos a2 + cos (3i cos ^82 + cos yi cos yg.
We have, then, the result : The cosine of the angle between

two directed lines equals the sum of the products of the corre-
sponding direction cosines of the lines.

Example. Find the angle between the two directed lines

whose direction cosines are, respectively, |, —\, — f and f, -J.
-|-,

Here cos ^ = 2. 3 + (-1) -2 +

^^-^ ^--
(-2).
^ ^
6
— "^
~2i'
8

whence is found to be 112° 24'.*

Parallel and Perpendicidar Directed Lines. Li and L2 are

perpendicular if and only if ^ = 90° or cos ^ = 0, that is, if and
only if

(4) cos «! cos CC2 + cos ySi cos (^2 + cos yi cos y2 = 0.
In words Two directed lines are perpendicular, if and only if
:

the sum of the products of the corresponding direction cosines of

the lines is equal to zero. Thus the directed lines which have
the direction cosines J, — |, f and ^, i|, j^^ are perpendicular,
since
2.2-1.14 + 2.5 ^^
3 . 15

* It is to be remembered that the angle between two directed angles is

an angle between 0° and 180° inclusive cf. Ch. XVII, § 3. ;

 DIRECTION COMPONENTS 427

We repeat here the results concerning parallelism obtained

in § 1. The directed lines Li and L2 are parallel and have
the same sense if and only if they have equal direction cosines :

(5) cos «! = cos 02? cos fii = cos 182, cos yi = cos y2.
On the other hand, they are parallel, but with opposite senses,
when and only when the corresponding direction cosines are
negatives of each other :

(6) cos «! =— cos a2, cos (3i =— cos jSg? cos yi = — cos y2.

EXERCISES
In each of the following exercises find the angle between
two directed lines with the given direction cosines.
1 2 3. 6.. 6. 2. 3 9 1 i_ 2. 2 1 2
*• T' 7? 7 ' 7' 7? 7* ^' 3' 3' 3 ' 3' 3> 3*

2 _23? H ± 4._3_ 12 _4. 9_6

11
2
ft I . 1_
^13'
13
**•
3' 3 ' 9' 9' 9' 1 3' 1 3 ' 11' 11? TT*
2 2
5.
V14' V14' V14' ve' ve' V6
6.
4 2 12 1 3
V2l' V21' V21' VI4' VI4' Vl4
7. Show that three directed lines with the direction cosines

Iji
13?
a_
13'
4_
13'
_4_12
13'13J
3
13J
3_4
13' TS"'
12
T3'

are mutually perpendicular.

8. Find the angle subtended at the point (5, 2, 3) by the

points (2, 0, - 3), (- 9, 7, 5). Ans. 79° V.
9. Determine the angles of the triangle with vertices at the
points (1, 0, 0), (0, 2, 0), (0, 0, 3).

Direction Components of an Undirected Line. The quan-

tifies ^, —^^, |- are the direction cosines of some line, properly
directed, and the quantities — y, f , — -f-
are the direction co-
sines of this line, oppositely directed.
428 ANALYTIC GEOMETRY

Both sets of direction cosines are proportional to the quan-

tities 2, — 3, 6. Consequently, these quantities pertain, not to
the line directed in the one sense or the other, but to the line
bare of sense, i.e. to the undirected line. call them direc- We
tion components of the undirected line.
It is clear that instead of 2, — 3, 6 we might have taken
equally well - 2, 3, -6, or 4, - 6, 12, or 200, - 300, 600, since
the two sets of direction cosines are proportional to the quan-
tities in any one of these triples. In other words, the direc-
tion components of the undirected line are not uniquely deter-
mined. There are infinitely many sets of direction components ;

if one set is 2, — 3, 6, all are given by the quantities 2p, — 3p,

6p, where p an arbitrary number, not zero.
is

Conversely, we have given the set of direction components,

if

2, — 3, 6, of the undirected line, and divide each by the square

root of the sum of the squares of the three, i.e. by

V(2)2 +(- 3)2 -f (6)2 = 7,

we obtain the direction cosines, ^, — f, f of the line directed
,

in one sense. Those of the line directed in the opposite sense

are the negatives, — 2, 3, — 6, of the given direction com-
ponents, each divided by the above square root.

Tlie General Case. Let a line L be given and on it the arbi-

trary directed line-segment PiP^ whose projections on the
axes are XjXo, Y^ F,, Z^Z. (Ch. XVII, Fig. 6). The direction
cosines of L, when directed in the sense of P1P2, are, by (5),
§1,

^^^
~D~' ~D^' T^'
if, however, L is oppositely directed, in the sense of P^P^^ they
are

^^ D '
D ' D '

where, in each case,

(3) D = VXiXs^ + Fi F22 + Z^Z^

 DIRECTION COMPONENTS 429

The two sets of direction cosines are proportional to the

quantities *

(4) -X'iX2, ^li^2) -^1-^2-

These quantities pertain merely to the undirected line L. We

call them a set of direction components of L.
Since the quantities (4) are the projections of PiP^ on the
axes, this definition can be stated as follows.

Defixitiox. a set of direction components of an undirected

line L are the projections on the axes of a directed line-segment
on L.
Instead of XiXi, Fi Y2, Z^Zo, we might have taken, as direc-
tion components of L, X2X1, Y^Yi, Z2Z1, i.e. the projections of
P2P1 on the axes; or 8X1X2, 3 YiYo, S Z1Z2, i.e. the projec-
tions of a directed line-segment on L having the same sense as
P1P2 but three times the length.
There are, then, infinitely many sets of direction components
for L. Any two sets are, however, proportioned. For, two arbi-
trary sets consist of the projections on the axes, X^Xs, Fi5^,
Z1Z2 and Xi'X2', Y1Y2, Z^Z.^, of two arbitrary directed line-
segments, P1P2 and P1P2, on L. But the projections of P1P2
and P1P2 on any line are in the same ratio as P1P2 and PiP^
(Ch. XVII, § 4, Ex. 4), and, therefore,

1 -A.2 -t
]
x.y 1 I ±2 -* 1 -* 2 -^1 -^2 -* 1 -^ ?

X1X2 P1P2 Y1Y2 P1P2 Z1Z2 P1P2

or

(5) Xi'X2' = p X1X2, Yi Yo' = pYi Y2, Zi'Zo' = p Z1Z2,

where the factor of proportionality, p, is P/Pa'/^i A? Q.- ©• d.

Not all three direction components can he zero. For, if X^Xo,

Y1Y2, Z^Zo were all zero, then, by (3), D= \

P^P2 \
would be
zero. But this is absurd, since Pi and Po are distinct points.
We summarize our results in the form of a theorem.

* The factor of proportionality is, in the first case, l/I) ; in the second,
- 1/D.
430 ANALYTIC GEOMETRY

Theorem 1. A71 undirected line L has infinitely many sets of

direction components ; if I, m, n is one set, all the sets are given by
pi, pm, pu, ivhere p is an arbitrary number, not zero; moreover,
I, m, n are not all zero. Any line parallel to L has the same sets
of direction components as L.
The last statement has not been explicitly proved. We leave
the proof to the student ; cf. Ch. XVII, § 4, Ex. 5.

Example 1. Find the direction components of a line parallel

to the axis of z.

Take any directed line-segment P1P2 on the line. Its pro-

jections on the axes are 0, 0, P1P2., or, if a is the number
representing PiPo, they are 0, 0, a. One set of direction com-
ponents is, then, 0, 0, a (a:=^0); a simpler set, and the one
generally used, is 0, 0, 1.

Example 2. A line bisects the angle between the positive

axes of y and z. What are its direction components ?
The projection on the aj-axis of any directed
line-segment P1P2 on the given line is zero,
and the projections on the axes of y and z are
equal. If the number representing both the
latter projections is a, a set of direction com-

Fig. 5
ponents for the line is 0, a, a. A simpler set
is 0, 1, 1.

Geometrical Representation of Direction Components. The

directed line-segments X1X2, YiY^, Z^Z^, which are the pro-
jections of P1P2 on the axes, rep-
resent geometrically the set of
direction components (4) of L.
Instead of them we prefer to use
the equal directed line-segments
P^R, P,S, P^T, issuing from Pi
(Fig. 6). These form what we
shall call a directed trihedral;
P^R, PijS, PiT are its directed
edges, and Pi, its vertex. ' /« Fig. 6
 DIRECTION COMPONENTS 431

The directed trihedral Pi-EST represents the set of direc-

tion components (4). Any second set, consisting of the pro-
jections on the axes, X^'Xz', Y1Y2, Z1Z.2, of any second
directed line-segment Pi'PJ on L is represented by the directed
trihedral P^'-R'S'T. For the two directed trihedrals we have,
from (5),

(6) P,'E' = pP,E, P,'S' = pP,S, Pi'T' = pP,T,

Because of this relationship we call them similar. That is,

tico directed trihedrals are similar, if homologous directed edges

are proportional, i.e. if the directions of the three edges of
one trihedral are all the same as, or all opposite to, the direc-
tions of the three edges of the other, and if the lengths of
homologous edges are proportional.
Since any two sets of direction components of L are in the
relation (5), the directed trihedrals representing them are in
the relation (6) and are, therefore, similar. Consequently,
the directed trihedrals representing the infinitely many sets
of direction components of L are all similar.

Construction of a Line with Given Direction Components.

Let it be required to construct the line L passing through a
given point P^ in space and having
the direction components 4, 3, 2. —
Construct a directed trihedral
Pi-RST with Pi as vertex and with
edges PiP, P^S, P^T defined, both
in length and direction, by the
numbers 4, — 3, 2. Complete the
box determined by the trihedral
and draw the diagonal P1P2 issuing
from Pj. The line of this diagonal
is the required line L. For, the
projections of P1P2 on the axes Fig. 7
have the values 4, — 3, 2.

Incidentally, we have shown that the triple 4, —3 2 is

actually a set of direction components of some line, L. We

432 ANALYTIC GEOMETRY

proceed to show, further, that L and the lines parallel to it

are the only lines having this triple or, more generally, the
triples 4p, — 3/o, 2p, p ^
0, as direction components.

Evidently L is the only line through Pi with the direction

components 4, — 3, 2. For, these components determine the
trihedral at Pi uniquely, the trihedral determines the box
uniquely, and the box the line.

If we took 8, — 6, 4 instead of 4, — 3, 2 as the given direc-

tion components, the resulting trihedral would have edges with
the same directions, but twice as long, as the edges of the
original trihedral, would be similar to the original trihe-
i.e. it

dral. The diagonal new box which issues from Pj

of the
'would be on a line with the diagonal P1P2 of the old box
and so the same line L would be determined. Similarly,
if any multiple, 4p, — 3p, 2p, of 4, —3, 2, where p is any

positive or negative number, were taken as the direction com-

ponents.
Finally, if we start from a new point P/, it is clear
that the line L' through
with the given direction com-
it

ponents will be parallel to L or, in case P/ lies on L, the

same as L.
The reasoning here is perfectly general, applying to any
triple of numbers, I, m, n, not all zero. The result is the fol-

lowing converse of Theorem 1.

Theorem 2. If I, m, n are any three numbers, not all zero,

the triples of mimbers pm, pn, where p is arbitral^ but not

pi,

zero, are sets of direction components of some undirected line L

and of the lines parallel to L, and of these lines only.
Remark. If one direction component is zero, the corre-
sponding edge of the directed trihedral disappears, and the
box becomes a rectangle, with L along its diagonal. If two
direction components are zero, the directed trihedral becomes
a directed line and L lies along this line.
 DIRECTION COMPONENTS 433

EXERCISES

In each of the following exercises find all the sets of direc-

tion components of the given line and then choose from them
a simple set.
1. A line parallel to the axis of x.

2. The line bisecting the angle between the positive axis

of X and the negative axis of z.

3. A line in the (oj, ?/)-plane having the slope 2.

Ans. p, 2p, 0; 1,2,0.
4. A line in the {y, 2;)-plane making an angle of 60° with
the 2/-axis. Two answers.
5. A line making equal angles with the three coordinate
axes.

6. The line through the origin and the point (2, 1, 3).

7. The line through the points (2, 3, 5), (4, 7, 8).

8. What can you say of the position of a line if one of its

direction components is zero ? If two are zero ?

In each of the following exercises construct the line through

the given point with the given direction components.

Point Direction Components

9. Origin, 3, 5, 2.

10. Origin, 2, — 3, 6.

11. (2,4,3), p,p,2p(p^0).

12. (5, - 4, 6), 3, 0, - 1.
13. (2, 5, - 3), 0, 1, 0. ^A
f 47 Formulas for the Use of Direction Components. Direction
Two Points. Let the two points
cfmnponents of the Line through
Pi (xi, 2/i, Zi), P2 {x2, .V2) ^2) ^6 given. Since the projections of
: :

P1P2 on the axes are (Ch. XVII, § 6),

(1) ^2, - iCi, 2/2

- 2/1, Z2 - Zi,

L
 —
434 ANALYTIC GEOMETRY
•

these three quantities are a set of direction components of the

undirected line passing through Pi and Pg.

Relationship between Direction Comjjonents and Direction

Cosines. We saw in § 3 that the direction components of an
undirected line are any three numbers, not all zero, propor-
tional to the direction cosines of the line, directed in one sense
or the other.
Conversely, starting with the arbitrary set of direction
components, X1X2, F1F2, ZA, of an undirected line, and
dividing each component by the square root of the sum
of the squares of the three, i.e. by the quantity D given by
(3), § 3, we obtain the direction cosines (1), § 3 of the
line, directed in one sense. And dividing the negatives
of the direction components by the same square root, we
get the direction cosines (2), § 3 of the line, directed
oppositely.
We state this result as a theorem.

Theorem. Ifl,m,n are a set of direction components of an

undirected liyie L, the direction cosiyies of L, when given a sense,
are
cosa = ^
±1 — ,

VZ2 -h m2 4- n^

± m
(2) cos p=

cos y = ±n

where either all the upper signs, or all the lower sigiis, are to be
chosen according to the sense which has been given to L.

If we had used, as the direction components of L, the arbi-

trary set pm, pn pi, (p ^ 0) instead of the particular set I, m, n,

the same formulas would have resulted. For if in (1) I,

(2)
m, n are replaced by pl, pm, pn, p comes out as a factor from
the square root in the denominator of each fraction and can-
 DIRECTION COMPONENTS 435

eels the p in the numerator, so that the fractions are left

unchanged.*

ExaTnple 1. A directed line has the direction cosines ^,

~ \y — f . What are the direction components of the line,
undirected ?

Obviously, 2, — 1, — 2 are one set of direction components,

the one generally used ; all sets are given by 2p, — p, — 2p,
where p =^0.

Example 2. An undirected line has the direction compo-

nents 4, — 3, 12. What are the direction cosines of the line
when directed?
The sum of the squares of the given direction components
is 169. Hence the direction cosines of the directed line are
either y^g-, — -^, i|, or ^3, 3^, — — i|, depending on the sense
in which the line is directed.

Angles between Two Undirected Lines. Between two di-

rected lines there but one angle 6 such that < ^ < 180°.
is

Between two undirected lines L^ and Zo there are, in general,

two such angles, as can be seen readily from a figure. The
two angles are supplementary and have, therefore, cosines
which are negatives of each other.
If the direction components of Li and L^ are ^1, mi, Ui and

hi ^2) ^^2> these cosines are given by the formula

(3) cos ^ =+ + ^^^ + ^1^

^J^ ^

V/r + mi2 V/2^ + ^2^ + ^2^

-I- 7ii2

To establish this formula, write down by use of (2) the direc-

tion cosines of each line, directed in either sense, and then
apply formula (3), § 2.
It follows from (3) that the two lines Li and Xo are perpen-
dicular when and only when
(4) Z1Z2 + 77iim2 + n-ji.^ = 0.
* If /o is negative, it is — p which comes out as a factor from each
square root and hence each sign ± becomes ^.
 436 ANALYTIC GEOMETRY

The lines Li and L2 are parallel if and only if their direction

components are proportional, i.e. if and only if

where /a is a number not 0. This follows directly from

Theorems 1, 2, § 3.
/

EXERCISES

]
1. From the direction cosines of the directed lines of
Exs. 1, 4, 6, § 2, find the direction components of the lines,

undirected.
M 2. From the direction components of the undirected lines

^N^ of Exs. 9-13, § 3, find the direction cosines of the lines, directed
^ first in one sense and then in the other.

'^ij^ 3. A line has the direction components 2, 8, 9. What are

^ its direction cosines, if it is directed upwards ?

^O 4. Find the direction components of the lines joining the
origin with the points (c), (/), (i), Q) of Ex. 1 of Ch. XVII,
§5.
'-.J^ 5. Find the direction components of the lines determined
by the pairs of points in Ex. 1 of Ch. XVII, § 7.

In each of the following cases determine the angles between

two lines with the given direction components. First test the
lines for parallelism or perpendicularity.

6. 3, 4, -1 ; 5, - 2, 7. 7. 4, - 2, 6 ;
- 6, 3, - 9.

8. 2, - 1, 3 ; 2, 1, - 1. 9. - 3, 4, 2; 5, 8, 1.

10. Show that the line joining the origin to the point
(2, 1, 1) is perpendicular to the line determined by the
points (3, 5, - 1), (4, 3, - 1).

^. Line Perpendicular to Two Given Lines. If two lines,

intersectingin a point P, are given, there is a single line
through P perpendicular to each of them, namely, the line
through P perpendicular to the plane determined by them.
 DIRECTION COMPONENTS 437

More generally, let any two non-parallel lines,* Li and 7>2,

with the direction components Zj, m^ Wi and I2, tiu, ?i2 be
given. Let X be a line perpendicular to each of them f
and let it be required to find for it a set of direction com-
ponents, I, m, n.

We Let 2, 3, 1 and 1, 4, 2 be the

begin with a special case.
direction components of Li and Lo, respectively. Since L,
with the direction components I, m, n, is perpendicular to Li
and also to L2, we have, by (4), § 4,

^ ^
Z4-4m + 2n = 0.
From these tiuo homogeneous linear equations it is impossible
to determine uniquely the three unknowns I, m, n. But this
was to be expected. For, there is not a unique set of direction
components, I, m, n, of L, but infinitely many sets.
In general, then, there will be a set for which n = 1. To
determine the values of m and n for this set, we must solve
simultaneously the equations

^
^
?-f-4m-f 2 = 0.
The solutions are I = ^, m = — f Consequently, one set of
.

direction components of i is — |, 1. A simpler set is 2,

-f,

-3,5.

In the general case, since L is perpendicular to both Li and

L2, it follows that

III + m^m -f Win = 0,

I2I + m2m 4- ^271 = 0.
* From now oil we drop the qualifying adjective "undirected," and
speak merely of lines and directed lines, as usual.
t There are infinitely many common perpendiculars to ii and L2.
They are, however, all parallel to one another and hence the direction
components of any one of them will be the direction components of all
the others.
 :

438 ANALYTIC GEOMETRY

Here, too, we try to find a set of direction components I, m, n

of Ly for which n = 1. Then equations (3) become

lit-\-m]m =— rii,

(4)
IJ. + mom = — 712.

The solutions of these equations are

— ^2^1
Z = Wi?i2 — ^2^1 m= yj ]^t2

These values for I and m, together with n = 1, form a set of

direction components of i.. 'A simpler set

is

or, in determinant form,

mi ?ii ^1 «1 rrii

(6)
m2 »l2 ^2 ^2 m2

For the special case first treated, these determinants have

the values
3 1 2 3
4 2 1 4

or 6-4 = 2, 1-4 = - 3, 8-3 = 5. But 2, - 3, 5 were the

direction components found, and thus the work in the special

case is checked.

Rule of Tliunib. To obtain the determinants in (6) easily,

write the two given sets of direction components under one
another
l\ '>ni Til,

I2 m-2 ^2*

The first determinant in (6) is formed from the second and

third columns of this array the second is formed ;

O
from the third and Jirst columns not the first
and third — and the third from the first and second
columns. Thus the sets of numbers, 2 3, 3 1,
1 2, represent the columns used in the three de-
Fig. 8 terminants. The first set, 2 3, is all that need
 .

DIRECTION COMPONENTS 439

be remembered. For, by advancing the numbers of this set ac-

cording to the cyclic order 1 2 3 1 2 • • •, this set 2 3 becomes
3 1, i.e. the second set ; and advancing the second set 3 1
cyclicly, we get the third set, 1 2.

Critique. Not all the determinants (6) are zero, for if

lim^ — l^fJii = — m^rii = nil2 —

9711722 712?! = 0,
then ^1 : Zg = wii m2 = Wi ^2 : :

or li : rrii = : rii I2 : 7)12 : 712

and hence the lines Li and L2 would be the same or parallel,

which is contrary to hypothesis.
In obtaining the solution (5) of equations (4) we assumed,
tacitly, that Zi??i2 — 12^1 ^ ; there was, then, a set of com-
ponents I, 771, 71, for which n=l. If Zim2 — /2^i = 0, at least
one of the two remaining determinants cannot be zero. If,
for example, nih — 712^1 =5^ 0, there will be a set of components
I, m, n, for which tti = 1, and we can find this set by putting
m = 1 in (3) and solving the resulting equations for / and n.

EXERCISES
In each of the following exeo^cises determine the direction
components of a line which is perpendicular to each of two lines
having the given direction components. Actually solve the
equations and then check the result by the rule of thumb.

|3, 4, 2; ^ f5, 6, -3; ^ 1, 2, -1

1. / o
2,
o
3.
-
2. {
^'
-4,
7 :'
-1.
3.
"•
f

-1, 9
)

[1, [2, 1 3,

|2, 1, -1; fO, 1, 0; |3, 0, 2

""' ""'
•
[4, 2, 3. jo, 0, 1. |1, 0, -1.

7. The two
directed lines Li and Zg passing through the
origin and having respectively the direction cosines iV2,
-I-V2, and — \ V2, ^ V2, are perpendicular to each other.
Find the direction cosines of a third line Z3 through the origin,
perpendicular to both L^ and L^, and so directed that Li, Z2, L^
 :

440 ANALYTIC GEOMETRY

form a directed trihedral (with edges of indefinite length)

which is right-handed by the test of Ch. XVII, § 5.
Ans. 0, 0, 1.

8. The above problem, if the direction cosines of Li and L2

are, respectively, -f, f, -f , and f, f, f Ans. f,
- f, f.

6. Three Lines Parallel to a Plane. Given three lines, with

the direction components ^1, mi, n^, U, mg, /I2, and l^, m^, n^.
The lines will be parallel to a plane or will lie in a plane,
if and only if there is a line, with the direction components

/ m n, which is perpendicular to each of them, i.e. if and

only if
III + miTTi + ni)i = 0,
(1) I2I + n^m + 712^ = 0,
l^l + m^m + n^n = 0.
But, by Ch. XYI, § 10, these three homogeneous linear equa-
tions have a solution for I, m, n, other than the obvious solu-
tion 0, 0, 0, when and only when the determinant of their
coeffi-cients vanishes
li mi ni

(2)
^ h ^2 ^2 =0.

We have, therefore, the theorem * :

Theorem. Three lines are parallel to a pZa^ie or lie in a

plane, if and only if the determinant of their direction components
has the value zero.
EXERCISES
1. Show that three lines through the origin with the
direction components 2, —1, 5, 3, 2, — 4, 7, 0, 6 lie in a
plane.
* The proof covers not only the general case, when the given lines
have but one common perpendicular direction and the equations (1) a
one-parameter family of solutions, but also the special case in which the
given lines are parallel, when the lines have infinitely many common
perpendicular directions and the equations (1) a two-parameter family
of solutions.
 DIRECTION COMPONENTS 441

In each of the following exercises show that three lines with

the given direction components are parallel to a plane.

2. 3,1,2; 5,-4,3; 1,6,1.

3. -1,1,2; 2,-1,1; 1,1,8.

EXERCISES ON CHAPTER XVIII

1. Show that the triangle with vertices at the points

(1, 3, - 5), (3, 4, - 7), (2, 5, - 3) is a right triangle.
2. Prove that the points (2, - 1, 5), (3, 4, - 2), (6, 2, 2),

(5, —3, 9) are the vertices of a parallelogram.

3. Show that (2, 3, 0), (4, 5, - 1), (3, 7, 1), (1, 5, 2) are
the vertices of a square.
4. Prove that the three points A, B, C with the coordinates
(5, -2, 3), (2, 0, 2), (11,-6, 5) lie on a line by showing that
the line AB has the same direction as the line AC.
5. Show that the two points (4, —2, —6), (—6, 3, 9) lie

on a line with the origin.

6. Show that two points (xi, y^, z^, (iCo, 2/2? ^2) li^ on a line
with the origin when and only when their coordinates are
proportional : Xi :
?/i :
^i = 0:2 :
1(2 '•
^2-

7. Show that the four points JL, B, C, D, with the co-

ordinates (3, 4, 2), (1, 6, 2), (3, 5, 1), (4, 5, 0), lie in a plane by
proving that the sum of the angles which BC and CD subtend
at angle A equals the angle which BD subtends at A.
8. Find the projection, on a directed line having -J,
— f, -f-

as its direction cosines, of the directed line-segment joining the

origin to the point (5, 2, 4). Ans. 4.

Suggestion. Use the method of § 2 or employ formula (2)

of Ch. XYII, § 4.

9. Show that the projection, on a directed line having

cos a, cos (3, cos y as its direction cosines, of the directed line-
segment joining the origin to the point {x, y, z) is

X cos a -f- 2/ ^^^ P -\-^ cos y.

442 ANALYTIC GEOMETRY

10. Find the projection, on a directed line with the direc-

tion cosines |, f, — ^, of the directed line-segment joining the
point (3, —2, —5) to the point (8, 0,-2). Ans. 5.

11. The previous problem, in the general case.

12. Two lines, Li and L2, have the direction components 1,

1, and 0, 1, — 1, respectively. Find the direction com-
ponents of a line which is perpendicular to Li and makes an
angle of 30° with L^. ^ns. 1, - 1, 2.

13. Prove that each two opposite edges of the tetrahedron,

with vertices at the points (0, 0, 0), (1, 1, 0), (0, 1, —1),
(1, 0,-1), are perpendicular.
14. A tetrahedron has three pairs of opposite edges.
Prove each of two pairs are perpendicular,
that, if the edges of
the edges of the third pair are also perpendicular. Choose the
coordinate axes skillfully.
15. Prove the identity

(/XlVo — /W'2Vl)^ 4- (V1X2 — V2A1)- -f (X1/X2 — ^2/Al)^

= (V + Ml' + vr-)(V + t^2'' + V2') - (A1A2 -I- fliflz + ViV2)2.
In the following exercises Ai, fxi, V], A2, />t2> ^2 (and A3,

/A3,V3) denote the direction cosines of directed lines, L^, L2

(and Z3), which we can assume go through the origin. In

solving the exercises, the identity of Ex. 15 will be found
useful.

16. If ^ is the angle between Li and L2, show that

sin^ = (/X1V2 — iJ-oviY H- (V1X2 — ^2X1)^ + (Ai^2 — hf^iY-

17. Prove that if Li and L2 are perpendicular, the direction
cosines of their common perpendicular X3 are

± (/"'IV2 — /^2n)> ('^1^2 — V2X1), ± (A.1/X2 — A.2^l)-

18. Show that, if the plus signs are taken in the above
formulas, Z3 will be so directed that the lines Li, L^, L^ will
 DIRECTION COMPONENTS 443

form a directed trihedral which is right-handed by the test of

Ch. XVII, § 5.

19. Prove that, if Li, Lg, and Zg are mutually perpendicular,

the determinant, \\ /x v\, of their direction cosines has the
value -f 1 or —
1, according as the directed trihedral consist-
ing of Li, L2, X3 is right-handed or left-handed.

I
 CHAPTER XIX

GTHE irfaces and Equations.

PLANE

Example
x=5
1. The equation

is satisfied by the coordinates of those points and only those

points which lie in the plane parallel to the (?/, 2) -plane and
5 units in front of it. We say that the equation represents
this plane.

Example 2. Consider the equation

x= y.

The points in the {x, whose coordinates satisfy it are

y)-plane
the points of the line L bisecting the angle between the posi-
tive iP- and ?/-axes. Since z is unrestricted by the equation,
the points in space whose coordinates satisfy it are the points
which lie directly above or below L, or are on L, i.e. the
points of the vertical plane through L. The equation, then,
represents this plane.

Example 3. The equation

represents, in the (z, x)-plane, a circle, C, with its center at the

origin and of radius 5. But the equation does not restrict in
any way the value of y. Consequently, it represents in space
the circular cylinder formed by drawing through each point
of the circle C
a line parallel to the axis of y and extending
indefinitely in both directions.

Surfaces. The planes and the cylinder represented by the

three equations considered are known as surfaces ; the cylinder
444
 ^ : :

THE PLANE 445

is a curved surface and the planes are plane surfaces. In gen-

eralizing the foregoing discussions we should, then, say
An equation in x, y, z represents, usually,* a surface. The
surface consists of all those points and only those points whose

coordinates, ivhen substituted for x, y, z in the equation,

satisfy it.

Shifting the point of view, we assume now that it is a sur-

face, and not an equation, which is given. Then we should
say:

The equation of a given surface is an equation in x, y, z which

is satisfied by the coordinates of every point of the surface and by

the coordinates of no other point

Problem Find the equation of the sphere whose center

1.

is at the origin and whose radius is a.

A point {x, y, z) lies on this sphere if and only if the square

of its distance from the origin is equal to a^

a;2 -f 2/2 -H 2^ = a^.

Therefore, this is the required equation.

Problem 2. Find the equation of the plane which passes

through the axis of x and makes an angle of
30° with the {x, 2/)-plane, as shown in Fig. 1.
This plane intersects the {y, 2;) -plane in the
line whose equation iw the (y, z)-plane is

2; = tan 30° 2/ or z = \^^y.

Fig. 1
But this equation, considered as an equation in
X, y, z, leaves x unrestricted ; consequently, it represents in
space the given plane, i.e. it is the equation of the given
plane.

* An equation in JC, ?/, z does not always represent a surface. For

example, the equation x^ + y"^ = represents a line, namely, the z-axis ;

the equation x2 + ?/2 4. ^2 _ represents just one point, the origin ;

and
the equation x^ + y"^ + z- -\- 1 = represents no point whatsoever.
446 ANALYTIC GEOMETRY

EXERCISES
What does each of the following equations represent ? Draw
a figure in each case. ^^ ^^

1. 2 = 0. jfcy22/ + 32; = 6. 'V^ + 2i/2 = 4.

.
3. x-^y = 0. 7. = a2. ^^a;2 - 9 ^2 = _ 9.
a-2^2/2

4. 2-2a; = 0. <3)22 = 2a;. 12. + + ^2 ,= 4.aj2 i/2

13. Which of the surfaces rex3resented by the above equa-

tions pass through the origin? Which contain a coordinate
axis ?

Find the equations of the following surfaces.

14. The (y, ;s)-plane.

15. The plane parallel to the (x, ?/)-plane and 3 units above
it.

16. The plane parallel to the (z, iK)-plane and 2 units to the

left of it.

17. The plane bisecting the angle between the (a;, y)- and
(V)
«)-plaiies and passing through the first octant.

18. The plane perpendicular whose trace *

to the (x, 2/)-plane

on that plane has the slope 3 and the intercept 2 on the axis
oiy.
19. The circular cylinder whose radius is 3 and whose axis
is parallel to the ic-axis and passes through the point (0, 1, 2).
20. Theparabolic cylinder whose rulings are parallel to the
?/-axis and whose trace on the (z, cw) -plane is a parabola with
its vertex at the origin and its focus at the point (2, 0, 0).
The elliptic cylinder whose rulings are parallel to the
21.
2-axis and whose trace on the {x, ?/)-plane is an ellipse which
has its center at the origin, its foci on the x--axis, and axes of
lengths 6 and 4.
* The trace of a surface on a plane is the line, or curve, of intersec-
tion of the surface with the plane.
 THE PLANE 447

22. The hyperbolic cylinder whose rulings are parallel to

whose trace on the {x, y)-plane is a rectangular
the z-axis and
hyperbola with its center at the origin and foci at the points
(0, ± 2, 0).

23. The sphere whose center is at the point (1, 0, 0) and

whose radius is unity.

24. The sphere whose center is at the point (1, 2, — 3) and

'
whose radius is 5.

(^. Elane through a Point with Given Direction of its Nor-

mals: Let there be given a point Pq with the coordinates
(^0) 2/oj ^o) and a
line L with the direction
components I, m, n. Through Pq perpen-
We
J/CK ^
dicular to L there is just one plane.* y^\
j \^ y
propose to find
Let P : (x, y, z)
its equation.
be any point of the plane
/^

y
/
^V/y
other than Pq. Then it determines with Pq l\/
a line, PqP, which is perpendicular to the fig. 2

line L. Since, by Ch. XYIII, § 4,

are the direction components of P^P and I, m, n are the direc-

tion components of L, it follows, by (4), Ch. XVIII, § 4, that

(1) l{x-XQ)-\-m{y-yQ)^n{z-ZQ)=0.
Conversely, if P: (x, y, z) be any point other than Pq whose
coordinates satisfy equation (1), this equation says that the line

PqP is perpendicular to L and hence that P lies in the plane.

The coordinates Xq, 2/oj ^q of ^^^ excepted point, Pq, obviously

satisfy equation (1). We have shown, then, that this equa-

tion is by the coordinates

satisfied of those points and only
those points which lie in the plane. Hence it is the equation
of the plane.
There are infinitely many lines L perpendicular or, as we
say, normal to the plane, and they are all parallel to one an-
* The figure is drawn for the special case in which L passes through Pq.
Direction
 THE PLANE 449

Let (a^o, 2/o> ^o) be a point whose coordinates satisfy equation

'-

(1)

(2) Axo H- Byo -{-Czo + D = 0.

Subtracting equation (2) from equation (1) we obtain

A{x - Xq)-\- B(y - yo)-{- C{z - Zo)=0.

But this equation is of the form (1), § 2, where

I : m : n =A B : : C.

Therefore it represents a plane which has A, B, C as the

direction components of its* normals. This result we state as a
theorem.
Theorem. Tlie general linear equation (1) always represents
a plane. The coefficients A, B, C are the direction components
of the normals to the plane.

Example 1. The equation

2a; -32/ + 42! -6 =
represents a plane whose normals have 2, — 3, 4 as direction
components. The point (2, 2, 2), for example,
lies in the plane, since when we set ic = 2,
2/ =2 in the equation, we find 2 = 2 as the
value of z.

We obtain a rough plot of the plane by-

constructing the point (2, 2, 2) and the line
through with the direction components 2,
it

— 3, 4, and by drawing then, as accurately

as possible, the plane through the point perpendicular to
the line.

Example 2. The equation

x = 2, or x-{-0y-\-0z = 2,
is the equation of a plane having the lines with the direction
components as normals.
1, 0, But these lines are parallel
to the axis of x and hence the plane is parallel to the {y, z)-
plane. In particular, it is two units in front of that plane.
 450 ANALYTIC GEOMETRY

Remark. In the proof of the theorem it was assumed that

there is always a point whose coordinates satisfy equation (1).

This assumption is easily justified. By hypothesis, at least

one of the three coefficients A, B, C is not zero. Suppose that
C=^0. Then equation (1) can be solved for z:

^_ Ax-\-By-\-D
z- .

^
Giving to x and y definite values, Xq and ?/o,
we obtain for z from
this equation a definite value, Zq. Then the point (o^o, yo, Zq)

has coordinates which satisfy (1). For example, if a^o = and

?/o
= 0, then Zo = — D/C, and the point is (0, 0, — D/C).
EXERCISES
In each of the following exercises determine the direction
components of the normals to the given plane and the coordi-
nates of a point lying in it. Construct the plane by the method
of ExaB3.ple 1.

f 1. 3.T + 5^ + 6^-5 = 0. 5. 2x-^Sy-5 = 0.

^-^'2 x-y-[-z-3 = 0. 6. 3 - 2 2 - 4 = 0.
a;

3. + 2y-3z+6 = 0.
4.x 7. 5 + 8 = 0.
?/

^,^4. 5x-2y-3z + = 0. 'i 8. 22-7 = 0.

Intercepts.
4. Let a plane be given by means of its equation.
A simple method of plotting the plane, in case it cuts the axes
^ in three distinct points (one on each axis), consists in determin-
V'ing from the equation the coordinates of these three points
and then in plotting the points and constructing the plane
through them.
The point of intersection of a plane, for example,

(1) 2x-3y-\-4.z-6 = 0,

with the axis of x has its y- and 2-coordinates both equal to

zero. Consequently, to "find the ic-coordinate of the point, we
have merely to set y = and 2 = in the equation of the plane
and to solve for x. Thus, in this case, we have
2a;— 6 = 0, or a; = 3.
 THE PLANE 451

The point of intersection of the plane (1) with the axis of x

is, then, (3, 0, 0). In a similar manner we find (0, 2, 0) and —
(0, 0, 4) as the points of intersection of the plane with the axes
of y and z respectively. By plotting these
three points and joining them by lines, we
obtain a good representation of the plane.
The numbers 3,-2, | are known as the
intercepts of the plane (1) on the axes of x, y, z,

respectively. That is, the intercept of a plane

on the axis of x is the a:-coordinate of the
point in which the plane meets the axis of x.
The intercepts on the axes of y and z are
similarly defined.
A
plane which passes through an axis or is parallel to an axis
has no intercept on that axis. Every other plane has definite
intercepts on all three axes and these intercepts determine the
position of the plane unless they are all zero, that is, unless the
plane goes through the origin.

EXERCISES
Determine the intercepts of the following planes on the
coordin^f^ axes, so far as they exist, and construct the

^2a;+3?/ + 42-12 = 0. 6. x + ^y — z = 0.
2. 3rc-2?/ + ^-6 = 0. 7. 2 - 3 4- 12 = 0.
a; 1/

3. X y — z — 2 = 0.
-{- 8. 32/ + J:2-6 = 0.

4. 2a;H-5?/-3^ + 8 = 0. 9. ox + 2z = 0.
5. x-\-2y -\-z-\-^ = 0. 10. 3 + 5 = 0.
a;

// 5. Intercept Form of the Equation of a Plane. Given a

plane whose position is determined by its intercepts. Let
these intercepts, on the axes of x, y, z, be respectively a, 6, c.

To find the equation of the plane in terms of a, b, c.

We have the problem of finding the equation of a plane

 :

452 ANALYTIC GEOMETRY

through the three points (a, 0, 0), (0, b, 0), (0, 0, c). Let this
equation be

(1) Ax-^By-^Cz-^D = 0,

where the values of A, B, 0, D are to be determined. Since

the plane does not go through the origin, D ^0. Since it

contains each of the given points, the following equations must

hold
Aa-{-D = 0, Bb-\-D = 0, Cc + D = 0.
Hence A = - D/a, B= - D/h, C = - D/c. Substituting
these values for A, B, C in (1) and dividing through by Z), we
obtain

(2) ^ + ^ + ? = l.
a b c

That this is the desired equation can easily be checked by

substituting successively in it the coordinates of the three
points in question.
Only planes which intersect the axes in three points that are
distinct can have their equations written in the form (2). A
plane through the origin is an exception, because at least one of
its intercepts is zero and division by zero is impossible. A
plane parallel to an axis is also an exception, since it has no in-
tercept on that axis.

EXERCISES
FjifS the equations of the planes with the following inter-
ts.

.J2,
3, 4. 3. - 2, .4, 5. 5. - 4, - 6, - 2.
2. 2,-3,-1. 4. -5,-3,2. 6. 2, -.8,-6.

Find the equations of the following planes.

7. With intercepts on the x- and ^/-axes equal to 2 and 3 and
parallel to the axis of z.

8. With intercept —3 on the 2;-axis and parallel to the

(Xj 2/)-plane.
 THE PLANE 453

9. A regular quadrangular pyramid has its vertices at the

points (0, 0, 6), (2, 0, 0), (0, 2, 0), (- 2, 0, 0), (0, - 2, 0).

Find the equation of its faces.

10. The same, if the vertices are at the points (0, 0, c),

(a, 0, 0), (0, a, 0), (- a, 0, 0), (0, - a, 0).

6. Plane through Three Points. Three points, not lying in a

straight line, determine a plane. In any particular case the
equation of the plane can be found by the method of the pre-
ceding paragraph. In the general case, when the points are
arbitrary and have the coordinates (x^, y^, z^, (a-2, 2/2? ^2)^
It is, however,
fe, 2/3J 2:3), this method could
still be applied.

simpler to write the equation in determinant form, by analogy

to the equation in that form of the straight line through two
o
points (Ch. XVI, § 11). We have, namely,

X y z 1

x^ 1
(1)
2/1
= 0.
2/2
1

X^ 2/3 2^3 1

To show that this equation actually represents the plane

through the three points, develop the determinant by the minors
of the first row. The equation then takes on the usual form,
Ax-\-By-\-Cz-hD = 0,
of a linear equation ; moreover, the values obtained for A, B, C:

2/1 2!l
1 X'l

A= 2/2 ^2 1 B=-
2/3 ^3 1
 454 ANALYTIC GEOMETRY

This is the plane through the three points, since the sub-
stitution of the coordinates of any one of the points for x^ y,
z in (1) makes two rows of the determinant identical and hence
causes the determinant to vanish and the equation to be
satisfied.

EXERCISES

/Find the equations of the following planes by applying

formula (1) and simplifying the result.
Through (1, 2, 0), (- 2, 3, 3), (3, - 1, - 3).
Through (2, 5, - 3), (- 2, - 3, 5), (5, 3, - 3).

3. Through (1, 1, 0), (0, 1, 1), (1, 0, 1).

4. Through (1, 1, - 1), (6, 4, - 5), (- 4, - 2, 3).

5. Through (4, 5, 2), (- 3, - 2, - 5), the origin.
6. Through {x^, y^, z^), (x^, y^, z^), the origin. Keep the
equation in determinant form, but simplify it.

7. Establish the intercept form of the equation of a plane

by applying formula (1).

8. By the method of the preceding paragraph find the

equation of the plane of

(a) Exercise 1 ;
(h) Exercise 3 ;
(o) Exercise 5.

9. Find the equations of the faces of the tetrahedron whose

vertices are at the points (0, 0, 0), (0, 3, 0), (2, 1, 0), (1, 1, 2).
Ans. 2 = 0, 2x = z, 2x-4.y-^z = 0,2x-{-2y-\-z = 6.

Perpendicular, Parallel, and Identical Planes. Angle be-

7.

tween Two Planes. The normals to the two planes,

(1) A,x + B,y + C,z -f A= 0,

(2) A2X + B2y 4- C2Z -f A= 0,

have Ai, A> Ci and A2, A, Co, respectively, as direction com-

ponents.
The planes are perpendicular if and only if their normals

are perpendicular; and parallel (or identical), if and only if

 :

THE PLANE 455

the normals of one are also the normals of the other. Conse-
quently, we have, by Ch. XVIII, § 4, the following theorems.

Theorem 1. The planes (1) and (2) are perpendicular when

and only when
(3) ^1.42 + B.Bo + C1C2 = 0.

Theorem 2. The planes (1) and (2) are parallel

* ivhen and
only ivhen
(4) AiiBi'. Ci =^2 '•
B2 : C2.

The condition that the two planes be identical is analogous

cf. Ch. II,
to the condition that two straight lines be identical ;

§ 10. We can state, then, the theorem

Theorem 3. TJie planes (1) and (2) are identical luhen and
only when
(5) Ai 5i: :
d : i>i = ^2 : ^2 : C'o : A-
The proof of the theorem is left to the student.

Angle between Two Planes. Between the two

planes (1) and (2) there are, in general, two
different angles having values between 0° and
180° inclusive, and these angles are supple-
mentary. They are equal to the angles be-
tween the normals to the two planes. Since
Ai, Bi, Ci and A2, B2, C2 are the direction com-
ponents of the normals, the cosines of the
angles are given, according to (3), Ch. XVIII, Fig. 5

§4, by

(6) COS =± A1A2 -h B1B2 -^ CA

V^i^ 4- ^i' + Ci2 V^2' 4- ^2^ H- Ci

EXERCISES

In each of the following exercises determine whether the

given planes are parallel or perpendicular, and in case they
are neither, find the angles between them.
* Or, in a single case, identical. Cf. Th. 3.
456 ANALYTIC GEOMETRY

1. 5a;-3?/ 4-22 + 4 = 0, 3x - y -9z -^2 = 0.

2. 2x-y-^Sz-l = 0, 2x-y-\-Sz + 3 = 0.
3. x + y + z = 0, 3a;+ 6?/ -22 + 12 = 0.
4. 2x-2y + 4.z-\-o = 0, -
3x St/ + 62 - 1 = 0.
5. 7a; + 52/ + 62 + 30 = 0, 3a; - - IO2 + 4 = 0.
?/

6. 2a; + + 2-3 = 0,
2/
a;+ ?/-32+4 = 0.
7. a;+ + 22 = 0,
?/ + 2 = 0.
a;

8. 4a; + 8?/ + 2-8 = 0, + 2 - 3 = 0.

3/

9. 2a; + + 32- 2 = 0,
2/
x-2y-\-5 = 0.
10. Show that two planes are parallel when and only when
their equations can be written in the forms

Ax + By+Cz = D, Ax + By-\-Cz = D\

Planes Parallel or Perpendicular to a Given Plane.

8.

1. Find the equation of the plane which passes

Example
through the point (5, 2, — 4) and is parallel to the plane

(1) •
2a; + 4?/-62-7 = 0.
The normals to the plane (1) have the direction components
2, 4, — 6 or, more simply, 1, 2, — 3. The required plane has
the same normals and passes through the point (5, 2, —4).
By (1), § 2, its equation is

l(a;_5)+2(2/-2)-3(2 + 4)=0,
or a^ + 22/-32-21 = 0.

Through a given point and parallel to a given plane there is

but one plane its equation can always be found by the above
;

method. But through a given point and perpendicular to a

given plane there is not just one plane, but infinitely many,

namely, the planes which pass through that normal to the

all

given plane which goes through the given point. To single

out one of these planes we must impose a further condition.
We might demand, for instance, that the required plane pass
through a second given point or, again, we might specify that
 THE PLANE 457

it be perpendicular to a second given plane. We proceed to

consider illustrative examples of these two types.

Example 2. Find the equation of the plane passing through

the two points (3, — 2, 9), (— 6, 0, —4) and perpendicular to
the plane
(2) 2a;-2/ + 4z-8 = 0.
First Method. Let the equation of the plane in question be

(3) Ax-^By-{-Cz-\-D = 0.

Since the plane contains the two given points, we must have
(4) 3A-2B-{-9C+D = 0,
(5) -6^ _4C-fD = 0.
Since it is perpendicular to the plane (1), it is necessary that

(6) 2^.-5 + 40=0.

In (4), (5), (6) we have three simultaneous linear equations
in the foitr unknowns A, B, C, D. But from three linear
equations it is impossible to determine uniquely the values of
four unknowns. It may be possible, however, to determine
the values of three of the unknowns in terms of the fourth,
say the values of A, B, C in terms of D.
Accordingly, we rewrite the equations in the form
SA-2B + 9C=-D,
6A 4-4 0- A
2A- B + 4:C= 0,

and solve A, B, C, either directly or by determinants.

for We
do obtain a solution, namely,
A = iD, B = -D, C = -iD.
Hence (3) becomes
^Dx -Dy - \Dz -h D = 0.
The plane represented by this equation is always the same, no
matter what value, other than zero, is given to D. A simple
choice is : D=2. We obtain, then, as the equation of the
required plane,
x-2y-z-\-2=0.
 :

458 ANALYTIC GEOMETRY

Second Method. Let N

be any normal to the required plane.
Since the plane contains the two given points, is perpen- N
dicular to the line L^ joining these points. Since the plane is
perpendicular to the given plane (2), is perpendicular to N
any L2 normal to (2). Thus
line is a common perpendicular ^
to the lines Li and ^2.
The direction components of L^ are, by Ch. XVIII, §4,
3_(_6), -2-0, 9-(-4), t.e.

9, -2, 13;
those of Z2 are 2, — 1, 4.

Consequently, by (6), Ch. XVIII, § 5, the direction components

of N are
-2 13 13 9 9 2
-1 4 4 2 2 1

i.e. 5, - 10, - 5, or 1,-2,-1. .

Our problem is now reduced to that of finding the equation

of the plane which passes through one of the given points,
say (3, — 2, 9), and has 1, — 2, — 1 as the direction com-
ponents of its normals. This equation is

l(a;_ 3)- 2)- 1(2 - 9)= 0,

2(2/ -I-

or x-2y-z-\-2=(i.
Example 3. Find the equation of the plane passing through
the point (2, 5, — 8) and perpendicular to each of the
planes
2a;-32/ + 4;3H-l = 0, 4.x + y -2z + = 0. Q>

Either of the methods employed in the previous example is

applicable. We choose the latter. A normal to the re- N
quired plane is perpendicular to the normals to both the given
planes. These have, respectively, the direction components
2, 4 and 4, 1, — 2. Consequently, the direction compo-
— 3,
nents of N
are 2, 20, 14 or 1, 10, 7.
The equation of the plane through (2, 5, — 8) with 1, 10, 7
as the direction components of its normals is
 THE PLANE 459

(x - 2)+ lOOy -5) 4- 7(2 + 8) = 0,

a;+ 10?/ + 7 -1-4 = 0.
'

or 2;

This is the required equation.

EXERCISES

In each of the following exercises find the equation of the

plane which is parallel -to the given plane and passes through

the given point. In Exs. 5, 6 find the equation directly

by inspection of a , figure.

Plane Point
1. 5a;-2^ + 3z-4 = 0, (2,4,3).
2. 3a;-h42/-82-2 = 0, (0,0,0).
3. 4x-22/-62=9, (2,-1,0).
4. 3:c- = 0,4:^ (5,2, -3).
5. 3a; + 8 = 0, (1, -2,5).
6. 2?/- 5 = 0, (4,0,3).

7. Find the equation of the plane passing through the

points (3, 1, 2), (3, 4, 4) and perpendicular to the plane
5a;_|_iy-|-42;=0. Apply both methods, checking the result of
one by that of the other. Ans. 2x-{-2y — 3z — 2 = 0.
The previous problem, if the given points and the given
plane are as specified. Use either method in Exs. 8-10 in ;

Exs. 11, 12 solve the problem directly by inspection of a

figure.
Points Plane
8. (3, 4, 1), (2, 6, - 2), 2a;- 3?/ + 4z- 2 = 0.
9. (0,0,0), (4,3,2), x-\-y-hz=0.
10. (3, 2, - 4), (5, - 1, 3), 4.x-5y = S.
11. (1, 0, 0), (1, 2, 5), 32/-7 = 0.
12. (0, 2, 0), (2, 0, 0), 2;2-f-5 = 0.
13. There are infinitely many planes which pass through
the two points (2, — 3, 4), ( — 2, 3, — 6) and are perpendicular
 :: ?

460 ANALYTIC GEOMETRY

to the plane whose equation is2a; — SiZ + Sz — 10 = 0. Why

Justify your answer.
14. What is the equation of the plane which passes through
the point (1, — 2, 1) and is perpendicular to each of the
planes
Sx-\-y-\-z-2 = 0, x-2y-\-z-\-4: = 0?
Apply both methods, checking the result of one by that of the
other. A71S. 3x—2y — 7z = 0.

The previous problem for the following given planes and

given point. In Exs. 15-17 use either method ; in Exs. 18, 19
obtain the result directly from a figure.

Planes Point

[6a;+2?/ — o2-|-4 = 0,
16. ^ + ^ + ^7^' . (1,-1,1).
[x-y + + 2 = z 0,

17. l:+^^^+r^=°' (0,0,0).

18. x = 2,y = 3, (2, - 5, 3).

19. 2a; + 2 = 0, 3aj-2; = 6, (2,1,-3).

20. There are infinitely many planes which pass through

the point (2, — 5, 0) and are perpendicular to each of the
planes

4a;- 2?/ -6^ + 3 = 0, - 6aj + 3^/ + 92 + 10 = 0.

Why ? Justify your answer.

9. Distance of a Point from a Plane. To find the distance

A of the point P : {xq, yo, Zq) from the plane

Ax + By-\-Cz + D = 0,

draw a line through P perpendicular to the (x, ?/)-plane and

mark the point Q in which this line cuts the given plane.
 THE PLANE 461

Then, as the figure shows, A is the numerical value of the

product QF cos 6, i.e.
A= I
QPgosBI,

where is the acute angle between the line QP and the normal
F'P to the plane.
The normal P'P has the direction
components A, B, C and the line QP
has the direction components 0, 0, 1.
Consequently,
C
cos 6 =± ,
•

VA^+ B^+ C2

It is immaterial to us which of the

two signs is the proper one, for we are interested only in
numerical values.
It remains to find QP. The a> and ^/-coordinates of Q are
the same as those of P, namely, Xq, 2/0 ; denote the 2-coordinate
of Q by Zq. Since Q : (a^o, yo, Zg) lies in the given plane, it fol-

lows that
Axq + Byo -{-Cz^ + D = 0,
462 ANALYTIC GEOMETRY

where, in the second formula, the sign to be taken is that

which gives a positive result.

In the above deduction we assumed that the line through P

perpendicular to the {x, 2/)-plane meets the given plane, i.e. that

the given plane is not perpendicular to the {x, ?/)-plane. This

means, analytically, that we have assumed that C^O.
Since we know that at least one of the three coefficients
Ay B, C is not zero and since the final formula (1) bears equally
on A, B, and C, it is immaterial which one of the three coeffi-
cients we assume not zero. The result would have been the
same ifwe had assumed, say, A=^ 0, instead of G ^0.

EXERCISES
1. Establish formula (1) on the assumption that A:^0.
Find the distance of each of the given points from the cor-
responding given plane. In Exs. 6, 7 check the result by in-
spection of a figure.

Point Plane
2. (3,-2,1), 2x-y + 2z-\-3 = 0. Ans. 41
3. (2,5,-3), 6x-3y + 2z-4. = 0.
4. (0,2, 1), 4a: + 3?/ + 9 = 0.

5. Origin, Sx-\-y — 4:Z — 6 = 0.

6. (3, 8,-6), y-5 = 0.
7. (-2,3,4), 2^ + 7 = 0.
8. Find the lengths of the altitudes of the tetrahedron of
Ex. 9, § 6.

10. Point of Intersection of Three Planes. Let there be

given three planes which intersect in a point, i.e. three planes
which have just one point in common, as, for example, the
planes of the ceiling and two intersecting walls of a room, or
the planes of three faces of a tetrahedron.
The point of intersection of the planes is that point whose
coordinates satisfy each of the three equations of the planes.
 THE PLANE 463

In other words, it is the point whose coordinates form the

simultaneous solution of the three equations. Consequently,

to find its coordinates we have but to solve the three equations
simultaneously.
Consider, for example, the three planes represented by the
equations
Sx-\-4y -5z = —llf
2x-\- y-\-6z = IS,
X — 3y -\- z = 6.
The simultaneous solution of these three equations is most
simply effected by the use of determinants (Ch. XVI, §§2, 8).
The result is y — 1, z = 2. Accordingly, the point of
x=l, =
intersection of the three planes is (1, — 1, 2).

Intersections of Three Surfaces. The method to be used

in finding the point (or points) of intersection of any three
surfaces, given by their equations, is now obvious. The equa-
tions are to be regarded as simultaneous equations in. the un-
known quantities, x, y, and z, and solved as such.

TJiree Arbitrary Planes. Let the equations,

Aix + B^y + C,z -\-D,= 0,

(1)' A2X + B,y + C2Z + A = 0,
A^x + B^y + C\z 4- A = 0,

represent three distinct planes. If the determinant, \A B C\,

of the coefiicients of x, y, and z is not zero, the three equations
have a unique solution (Ch. XVI, § 8, Th. 10) and hence the
three planes intersect in a single point.
Conversely, if the three planes have just one point in com-
mon, \A B C\=^0. For, \i \A B C\ vanished, the normals to
the three planes would all be parallel to a plane, M, by
Ch. XVIII, § 6. Consequently, the lines of intersection of
the three planes, taken in pairs, would be perpendicular to M.
If there were just one such line of intersection, the three
planes would have all the points of this line in common ; if

there were no line of intersection or more than one, the three

 ;

464 ANALYTIC GEOMETRY

planes would have no point in common. In either case the

hypothesis is contradicted. Hence we may state the theorem :

Theorem. Tlie three planes (1) intersect in a single pointy if

and only if A B C\ 4-0* \

If I
^ jB C = I
0, two or all three of the planes may be parallel
these cases are easily detected by inspection of the equa.tions.
Or, the three planes, taken in pairs, may intersect in three
distinct parallel lines. Or, finally, they may have a line in
common. We shall learn later, Ch. XXI, § 2, how to dis-
tinguish, from the equations of the planes, between these last

two cases.

EXERCISES
In each of the following exercises show that the three given
planes intersect in a single point, and find the coordinates of
the point.

1. The planes of Ch. XVI, § 2, Ex. 10.

2. The planes of Ch. XVI, § 2, Ex. 11.
3. The planes of Ch. XVI, § 2, Ex. 12.
4. The planes of Ch. XVI, § 2, Ex. 13.

5. Find the coordinates of the vertices of the tetrahedron

whose faces lie in the planes

z = (i, 2?/ -32 = 0, a; -2/ + 3 = 0, 5a; - 2?/ + 3^ = 0.

Find the points of intersections of the following surfaces.
Draw a figure in each case.
6. a; = 4, 2 =- 2, x"^ -\-
y"^
= 25.
7. X -\- y = 2, X —y= 0, a;^ + ^^ — 1 = 0.
8. a;2 + 2/2 _|_ ^2 _ 9^ 5x-\-y — Sz = 5, x = z.
Ans. (2, 1, 2), (I, i, i).

* Or, the three equations (1) have a unique solution, if and only if
\A B C\^0. This is the converse of Th. 10, Ch. XVI, § 8. have We
thus completed, by geometric methods, the proof of an important fact in
the theory of Hnear equations.
 THE PLANE 465

In each of the exercises that follow give all the information

you can concerning the relative positions of the three given
planes.
Sx-4:y-^z + = 0, l 9x+6y-3z + 7 = 0,
9. —2x-\- 2/H- 2 + 5 = 0, 10. x — 2y-\- 2 + 3 = 0,
6x-8y-3z-2 = 0. 6a; + 4?/-2z-l = 0.

11. + 12 = 0, 3x-\-5y-l = 0,
2a;-3?/ 5aj+2?/+ll = 0.
12. 4aj-32-5 = 0, 3a; + 5:^-11 = 0, 7a; + 22; + 3 = 0.

EXERCISES ON CHAPTER XIX

1. When will the plane Ax-\-By-^Cz-\-D = pass through
a coordinate axis, e.g. the axis of z ? When will it be parallel
to a coordinate axis, e.g. the axis of ic ?

2. Find the equation of the plane through the axis of z

and the point (1, 2, 0).

3. Find the equation of the plane through the axis of y and
the point (2, 3, 1).

4. What is the equation of the plane whose intercepts are

one half those of the plane 2x — Sy-\-4:Z — 12 = 0?
5. A perpendicular from the origin meets a plane in the
point (2, — 3, 4). What is the equation of the plane ?

6. A line through the point (2, 3, 7) meets a plane in the

point (5, — 1, 2). Find the equation of the plane.
7. Find the equation of the plane which bisects perpendic-
ularly the line joining the points (4, 3, — 1), (2, 5, 3).

8. Determine the point on the axis of y which is equidis-

tant from the points (3, 7, 4), (— 1, 1, 2).

9. One vertex of a box is at the origin and the edges

issuing from along the positive coordinate axes. Prove
lie

that the intercepts of the plane which bisects perpendicularly

the diagonal through are inversely proportional to the
lengths of the edges.
466 ANALYTIC GEOMETRY

10. For what value of m will the two planes,

2x-\-my — z = 6x — 5y — Sz = S,
4:,

(a) be perpendicular ? (6) be parallel ?

11. m will the two equations

For what value of
— =
mx y-hz-\-S 0, 4.x- my-[-mz + 6 =

represent the same plane ?

Find the angle which the line through the points
12.

(3, 2, —
1), (0, 4, 1) makes with the
plane 2x — y — z-\-S = 0.
Suggestion. Find first the angle between the line and a
normal to the plane.
13. What angle does the plane Sx — y — z = B make (a) with
the (x, 2/) -plane ? (6) with the y-axis ?

14. Find the distance between the two parallel planes

2x-3y + 6z + 6=0, 2 x- Sy -\- 6z -1 = 0,

Suggestion. Find the distance of a chosen point of the first

plane from the second.

15. Show that the distance between the two parallel planes

Ax + By-{-Cz-\-D=0, Ax-\- By + Cz + D' =0

'

.
\I>'-D\ '

IS •

There are two points on the axis of z which are distant

16.
four units from the plane 2 x — y -\-2 z -\-3 = (i. Find their
coordinates. Ans. (0, 0, 41), (0, 0, - 7|-).
17. Show that the equation of any plane parallel to the
plane
Ax + By + Cz-\- D=^0
can be written in the form
Ax -\- By -\- Cz = k.

18. Using the method of Ch. II, § 11, work Exs. 1-4, § 8, of
the present chapter.
 THE PLANE 467

19. There are two planes parallel to the plane 2x—6y-{-Sz

=4 and distant 3 units from the origin. Find their
equations.

20. Find the equation of the plane parallel to the plane

given in Ex. 19and so located that the point (3, 2, 8) is mid-
way between the two planes. Ans. 2 x — 6 y -\- S z — 32 = 0.
21. Three faces of a box lie in the planes 2x — y = 6,

x-\-2y = S, z = S and a vertex is at the point (9, 5, 2). Find

the equations of the planes of the other three faces.
22. Find the equation of the plane which passes through
the point (2, — 1, 8) and is parallel to each of two lines hav-

ing 2, — 3, 4 and 5, — 7, 8 as their direction components.

Ans. 4:X-{-4:y-\-z — 12 = 0.

23. Show that the equation of the plane passing through

the points (xi, yi, z^), (iCo, ?/2> ^2) and perpendicular to the plane
Ax-^ By + Cz -\- D= can be written in the form

X y z 1

^1 Vi ^i 1
= 0.
X2 2/2 Z2 1

A B
24. Show that the four planes,

2x-y-z-3 = 0, x-y-[-2z-3 = 0,
x — 2y-\-z = 0,^ a;H-2/ + 2 — 6 = 0,
meet in a point.
25. The six planes, each of which passes through the mid-
point of an edge of a tetrahedron and is perpendicular to the
opposite edge, go through a point.
Prove this theorem for the tetrahedron of Ex. 9, § 6.

26. Prove the theorem of Ex. 25 for the general tetrahedron,

choosing the coordinate axes skillfully.
27. Show that the plane

^x-\-^y + ^z = 5
468 ANALYTIC GEOMETRY

is 5 units distant from the origin and that f ,

f, -J
are the di-
rection cosines of a normal to it, directed away from the
origin.

28. State and prove for the plane

(1) X cos a-\-y cos /? + 2; cos y = p, P^O,

the results corresponding to those given in the preceding exer-
cise. Show that the equation of every plane can be written in
the form (1). Prove, also, that the distance of the point
(xq, 2/0, Zq) from the plane is

I
Xq cos a + 2/0 cos p -\-Zo cos y —p \.

29. Prove that, if a plane has the intercepts a, b, c and is

distant p units from the origin,

1+1+1=1.
a2 52 c2 p^

Symmetry
30. A surface is symmetric in the {x, y)-plane if the substitu-
tion of — z for z in its equation leaves the equation essentially
unchanged. Prove this theorem and state the corresponding
theorems for symmetry in the {y, z)- and (2, a;)-planes.
31. A surface is symmetric in the axis of z if the substitution

of — and of — y for y, leaves the equation esseritially un-

X for X,

changed. Prove this theorem and state the corresponding

theorems for symmetry in the axes of x and y.

32. Prove that a surface is symmetric in the origin if the

substitution of —x for x, of — y for y, and of — z for z, leaves
the equation essentially unchanged.

Test the surfaces of the following exercises of § 1 for

33.
symmetry in each coordinate plane, in each coordinate axis, and
in the origin.

(a) Ex. 8 ; (6) Ex. 9 ;

(c) Ex. 10 ;
(d) Ex. 12.

34. Prove the following theorems :

 THE PLANE 469

(a) If a surface is symmetric in each of two coordinate

planes, it is symmetric in the coordinate axis in which the two

planes meet.
(b) If a surface is symmetric in each coordinate plane, it is

symmetric in the origin.

(c) If a surface is symmetric in a coordinate plane and in

the coordinate axis perpendicular to this plane, it is symmetric

in the origin.
 CHAPTER XX .

THE STRAIGHT LINE

1. Equations of a Curve. Example 1. Given the two equa-

tions

(1) x = 0, 2/ = 0.
The points whose coordinates satisfy both equations simultane-
ously are the points on the axis of z, and no other points. We
say that the two equations represent the axis of z.

Example 2. Consider the two equations

(2) 3a;-4?/-2 + 6 = 0, 5a; + 3?/ + 2^-8 = 0.

A point whose coordinates satisfy both equations at once must
lie in each of the two planes represented by the equations, i.e.
it must be a point on the line of intersection of these planes.
Conversely, the coordinates of any point on this line satisfy
both equations. Thus the two equations, considered simidtane-
ously, represent a line, the line of intersection of the two planes
which the two equations, taken individually, define.

Example 3. Take, now, the pair of equations

(3) x^-\-y'' + z'^ = 4:, x-y = 0.

By reasoning similar to that of Example 2, it follows that
these equations, taken together, represent the curve of inter-
section of the two surfaces which are defined by the two equa-
tions considered individually. The first equation is that of the
sphere whose center is at the origin and whose radius is two
units long. The second equation represents the plane through
the axis of z bisecting the angle between the positive oc- and y-
470
 : ;

THE STRAIGHT LINE 3c_ 471

axes. Consequently, the two equations, considered simultane-

ously, represent the circle in which the plane intersects the
sphere.

Example 4. Consider, lastly, the pair of equations

(4) a;2 -f ^2 -f-

^2 = 4^ aj2 + - 1)^ = 1.
(?/

The first represents the sphere of Example 3. The second is

the equation of the circular cylinder erected vertically on the

circle in the (a;, ?/)-plane whose center
is at the point (0, 1, 0) and whose
radius is unity. The curve of inter-
section of the two surfaces, i.e. the —
curve represented by the two equa-
tions taken simultaneously, is shown —
in Fig. 1. It does not lie in a plane
to distinguish it from curves which
do, we call it a twisted curve.

Space Curves. The straight lines

(1) and (2), the circle (3), and the

twisted curve (4) are all called space
curves. There are, then, three types
of space curves : straight lines, plane
curves other than straight lines, and
twisted curves.
We now put into definitive form
what we have learned from the fore-
Fig. 1
going examples
Two equations in x, y, z, considered simultaneously, re^yresent
usually * a space curve. The curve consists of all those points and
only those points ivhose coordinates satisfy simidtaneously both
equatio7is. It is the total intersection of the two surfaces which
are defined by the two equations when taken individually.
* Two equations do not always represent a curve. For example, the
pair of equations x2 + ?/2 = 0, x"^ + z"^ = represents just one point, the
origin ; and the equations of two parallel planes, asx — y + z = 2 and
X —y -\- z = S, represent no point at all when considered simultaneously.
 :

472 ANALYTIC GEOMETRY

If on the other hand it is a curve, and not a pair of equations,

which is given, we should say:
A space curve can be represented by two simultaneous equations.
TJiese can be axy two equations which are satisfied simultaneously
by the coordinates of every point of the curve and by those of no
other point. This means, geometrically, that the curve can be
considered as the intersection of any two surfaces on which it
provided the two surfaces have no other point in common.
lies,

For example, a straight line can be considered as the inter-

section ofany two planes through it. Consequently, it can be
represented by the equations of any two of these planes. In
other words, the two equations of the line are not unique. Thus,
as equations of the axis of z we might take the equations (1) or
we might take, equally well, any other two equations represent-
ing planes through the axis of z, as

ic-2?/ = 0, 17 a? 4- 56 2/ = 0.
Equations however, the simplest choice, and naturally
(1) are,
we shall find it of aid in analytical work to choose always that
pair of equations representing the curve under consideration
which seems most simple.
Problem 1. What are the equations of the straight line L
through the origin with the direction components 2, 3, — 1 ?
If P: {x, y, z) is any point on the line,
other than the origin 0, then x, y, z, con-
sidered as the projections of OP on the axes,
are also direction components of L. Hence,
X, y, z are proportional to 2, 3, —1
FIG. 2
(5) ^. = 2p, y = 3p, = -p,
z

where the factor of proportionality, p, is not constant, but vari-

able, depending for its value on the position of P on the line.*

* From (5),
OP2 = x2 + ?/2-|- 2;2=:14p2 qT
OP
= ± -^'
R
Vl4
That is, p is equal numerically to the distance of P from O, divided by
Vl4. Its sign depends on the side of on which the point P lies.
 : :

THE STRAIGHT LINE 473

Conversely, if P : (x, y, z) is any point other than 0, whose

coordinates satisfy equations (5) for some value of p, these
equations say that the direction of is that of L and hence OP
that P lies on L.
The coordinates (0, 0, 0) of the excepted point, 0, obviously
satisfy (5), when p is given the value 0. Consequently, equa-
tions (5) represent those points and only those points which
lie on iy, i.e. they represent L.
Instead of equations (5) it is more convenient to write

E = ^ = _i_.
(6)^
^ 2 3-1
This continued equality yields the three equations

(7) 3a;-2y = 0, 2/ + 3^ = 0, a; + 22 = 0.
One of these equations must be superfluous, since we know
that two equations are all that are necessary to represent a
line. As a matter of fact, the three planes defined by the
three individual equations all pass through L and hence one of
them is superfluous in determining L. We prove this analyti-
cally by showing that a simultaneous solution of any two of
the three equations always satisfies the third. Thus, if a^o, ^o?

^o are any values of ic, 2/, z which satisfy the first two equations,
I.e. if
3 a^o -2 2/0 = 0, 2/0 + 3 = 0,
2!o

elimination of y^ gives the relation

a^o + 2 20 = 0,
which says that these values also satisfy the third equation,
q. e. d.
Since one of the equations (7) is superfluous, we might take
any two of these equations, as
3a;-22/ = 0, 2/ + 32 = 0,
to represent L. more convenient, however, to consider
It is
the continued inequality (6) as defining Z/, and to call this
continued inequality tlie equations of L, remembering always,
that one of the equations which follow from it is superfluous.
 :

474 ANALYTIC GEOMETRY

Problem 2. Find the equations of the curve of intersection

C of two circular cylinders of the same radius a, whose axes
are respectively the axes of x and y.
The equations of the two cylinders are

(8) 2/^ + 22=a2, a;2 -f 2;2 = a2,

and these two equations, taken together, represent the curve C.
They are not, however, the simplest pair of equations possi-
ble. If Xq, 2/o, Zq are any set of values of x, y, z satisfying them
simultaneously, i.e. if

elimination of Zq gives the relation

a^o^ - 2/o' = 0,
which says that Xq, ?/o, Zq also satisfy the equation

(9) x^-y'^ = 0.
That is, the curve of intersection C of the cylinders (8) lies on
the surface (9).
Conversely, the surface (9) intersects each of the cylinders
(8) in the curve C and in no other points. For, if Xq, y^, Zq are
any values of x, y, z satisfying equation (9) and the first, say,
of equations (8), we have
^o' - 2/0" = 0, 2/o' + 2o' = o?'
Elimination of yo gives the relation

Xq -\- Zq~ =a y

which says that Xq, ^/o, Zq satisfy also the second equation of

(8), q. e. d.

We have proved, then, that the total intersection of any two

of the surfaces (8) and (9) is the curve C. Hence any two of
the equations (8) (9) define C. A simpler pair than the
and
pair (8) is the combination of one of the equations (8) with the
equation (9), for example

(10) a;2 + 22 = ^2 ^2_2^2^o.

 THE STRAIGHT LINE 475

The surface (9) consists of the two planes,

passing through the z-axis and

bisecting the angles between
the (y, z)- and (z, a;)-planes.

Each of these planes intersects

either cylinder in an ellipse ; cf.

Ch. XII, § 6. Consequently,

the curve C consists of two el-

lipses. We have, then, not only

obtained the simpler equations Fig. 3

(10) to represent the curve, but

we have also, in the process, succeeded in determining the
nature of the curve.

EXERCISES
In each of the following exercises determine what the given
equations, considered simultaneously, represent. Draw a
figure.

1. 2/ = 0, = 0. 2; 2. ic = 4, = 0.2/

3. a; += 0, - 3 = 0.
4 ;? 4. 2?/+ 3 = 0,3:2-5=0.
5. 2/
— = 0, = 3.
2; a; 6. 3 a; -f- 2 2/ = - 4 = 0.
0, 2;

7. 2/ -f- = 2, 2 a =5.
2; 8. ic — = 0,
2/ X — 2 = 0.

9.
^'
= ^ = ?-. 10. x = ^y = z.
3 2 6
11. x2 + + ;22_9^0, = 2.
2/^ 2/

12. a:2 ^ + — 16 = 0, X + = 4.
2/^ 2;2 2;

13. a;^ + = 4, 2 + 2 = 3.
2/2 2/

14. a;2 4- + 22 _ 25 = 0, x^ + y^^ 16.

2/2

15. a;2 + + ^2 _ 1(3 ^ 0, 4 + - 2)^ = 4.

2/2 a;2 (2/

16. a;2 4- r = «% z ay.

17. Which of the curves represented by the above pairs of

equations pass through the origin ?
476 ANALYTIC GEOMETRY

Find the equations of the following curves :

18. The axis of y.

19. The line in the (a;, ?/)-plane 3 units in front of the

(?/, 2;)-plane.

20. The line 2 units to the left of the {z, ic)-plane and 3
units above the (a;, ?/)-plane.

21. The line | units behind the (2/,

2!)-plane and | units to
the right of the (2, a;)-plane.

22. The line which lies in the plane passing through the
ic-axis and bisecting the angle between the positive y- and
;3-axes and is 4 units above the {x, i/)-plane.
23. The line through the origin with the direction com-
ponents 1,-1,-1.
24. The line through the origin with the direction com-
ponents 2, 0, 3.

25. The circle of radius 3 whose center is on the axis of z

and whose plane is 4 units above the {x, 2/)-plane.

26. The radius 2 whose center is at the origin

circle of

and whose plane passes through the ^/-axis and bisects the
angle between the positive axis of x and the negative axis of z.

Pind a simpler pair of equations to represent the curve

given in each of the following exercises and then identify the

curve. Draw a figure.
27. 2 - 3 2 = 0, 5 + 16 2 = 0.
2/ 2/

28. x^y + z=z2,x + y — 6 = 0.

29. The curve of Ex. 14.

30. x^ + y^ + z^^ a2, 2 2/2 + 2^2 = a\

2. Line of Intersection of Two Planes. Let the plane

Aa; + -Bi2/4-a^ + A =
and the plane
A2X + B^y + C22; + i>2 =
 THE STRAIGHT LINE 477

be two planes which meet. According to the theory of the

preceding paragraph, their line of intersection is represented
by the two simultaneous equations

(1)

i.e. by these equations, where the only sets of values of x, y,

z considered are those which satisfy both equations.
Since the line lies in each plane, it is perpendicular to the
normals to each plane. That is, it is a common perpendicular
to the normals to the two planes. From this fact its direc-
tion components can easily be determined, by the method of
Ch. XVIII, § 5.

Consider, for example, the line (2) of § 1. The normals to

the two planes determining this line have 3, — 4, — 1 and 5,

3, 2 as direction components. Consequently,

4 -1 -1 3 3 -4
3 2 2 5 5 3

^1
478 ANALYTIC GEOMETRY

EXERCISES

Find the direction components of each of the following

lines.Determine also a point on the line and hence construct
the line.

1. 2x-\-y-\-3z = 0, .
x— 2y -\-4:Z= 0.
2. Sx-4.y-6z-\-7 =0, 2x-y-2z + l = 0.
3. x-j-Sy-3z + 5 = 0,
4. 9 — y — Z = Oy
'^x
^- x+ y = 5.
5. 3x By -2-1 = 0, - 4 = 0.
a:

x + 2y = 0, y-3z = 0.
7. What are the equations of the edges of the tetrahedron
of Ch. XIX, § 6, Ex. 9 ? Draw a figure and label the edges
and the pairs of equations to correspond.

8. The same for the tetrahedron of Ch. XIX, § 10, Ex. 5.

9. Show that the lines of Exs. 1

and 4 are perpendicular.
10. Show that the lines of Exs. 3
and 6 are parallel.
Fig. 4
11. Each side of a hip roof (Eig. 4)

makes an angle with the horizontal whose tangent is ^ What

angle do the edges of the roof make with the horizontal ?

3. Line through a Point with Given Direction Components.

Let it be required to find the equations of
the line L which goes through the point
Pq:{xq, 2/0, 2o) and has the direction com-
ponents I, m, n.

Let P (x, y, z) be any point on i, other

:

than Po- Then the direction components

of PqP, namely, -
P^^ 5

(1) X — Xi0> y-y^^ Z— Zq,

 , ,

THE STRAIGHT LINE 479

are direction components of L. They are, then, proportional

to /, ??i, n :

(2) x — Xq = pi, 7j — yQ = pm, z — Zq = pn,

where the factor of proportionality, p, varies in value as P
changes position.*
Conversely, if P : (x, y, z) is any point other than Pq, for

which equations (2) hold for some value of p, i.e. for which
the quantities (1) are proportional to I, m, n, it follows that
the direction of PqP is that of L and hence that lies on L. P
The coordinates (xq, y^, Zq) of the excepted point Po obviously
satisfy (2), when p = 0. Consequently, equations (2) are satis-
fiedby the coordinates of those points and only those points
which lie on L and so they represent L.
Instead of equations (2), we can write

/3x a; - 0^0 ^ - .V ?/o ^ g - gp

I ra n

We call this continued equality the equations of the line,

remembering from § 1 that, of the three equations which in

general result from it, one is superfluous.

If one of the denominators in (3) is zero, so is the cor-

responding numerator thus, if n = 0, then z — Zq = 0.;

For,
equations (3) are but an abbreviated form of the equations of
proportionality t (2), and if n = the last equation in (2)
reduces immediately to z — Zq = 0.
Suppose, for example, that the line is to go through the
point (3, 4, — 6) and have 0, 5, 3 as its direction components.
Here I = 0, and hence a; — ccq = 0, i.e. ic — 3 = 0. Thus,

* From (2) PqP^ = p2 (^2 4. ^2 + ^2)

or p =± ^'^
\/l^ -f- m2 + n2
That is, the numerical value of p is proportional always to the distance
of P from Po- The sign of p depends on the side of Pq on which the
point P is situated.
t Cf. Ch. XVI, § 9, eq. (5).
 :

480 AISIALYTIC GEOMETRY

cc 3— = is one of the equations of the line. The other,
obtained from the equality of the last two members of (3), is

y — 4_2-h6 or 3a; -52 -42=0.

5 3 '

Hence the desired equations are

ic-3 = 0, 3 -5 2-42 = 0.
a; "

Again, let the point be (1, 2, — 3) and let the direction com-

ponents be 0, 0, 1. In this case, = 0, and m =

I therefore ;

a; — xq = and ?/ — ?/o
= 0, i.e.

a;-l = 0, y-2 = 0,
and these are the equations of the line. This result might
have been obtained directly, by inspection, since it is clear
that the line is parallel to the axis of z.

Reduction of a Continued Equality to the Form (3). Ex-

ample 1. What are the direction components of the line
x_—^_ y + 4 _2 + 3^
6 ~ 3 2 *

This continued equality will be of the form (3), if the minus

sign before the second fraction is associated with the denomi-

nator. Accordingly, the direction components are 6, — 3, 2.

Example 2. Consider a more complicated case :

2x-1 ^ 2-^y ^Zz *

3 4 2

To put this continued equality in the form (3), divide the

numerator and denominator of each fraction by the coefficient
of the variable in the numerator

3 4 2.
2 5 3

Here, each variable x, y, z, appears in the numerator, has

as it

the coefficient unity, and there is complete conformity to (3).

 THE STRAIGHT LINE 481

The direction components of the line are, then, ,

f |, f or
—
45, — 24, 20. Furthermore, it is clear that the line goes
through the point (|, |, 0).

EXERCISES

In each of the following exercises find the equations of the

line through the given point with the given direction com-
ponents.

Point Components
1. (2, - 3, 1), 5, 2, - 4.
2. (0, 0, 0), 3, - 1, 2.
3. (4, - 1, - 2), - 6, 5, 8.

4. (2, 0, - 3), 1, 1, 1.

5. (3, 2, - 8), 1, 3, 0.

6. (2, 0, 1), 4, 0, 1.

7. (-3,4,6), 0,1,0.

In each of the exercises which follow, find the direction

components of the line represented by the given equations
and the coordinates of a point on the line. Construct the line.

^~^= y+2 ^g-l

8
2 3
5*
3a; +2^ y ^ 2z-l
9
9 -6 4*
10. l-x = y — 2 = z — 6.
11. 3x-\- = 2 — 5y = 4:Z — T.
4:

12. 2x = l — y=z3z.
13. Show that the lines of Exs. 8 and 10 are perpendicular.
14. Show that the lines of Exs. 9 and 12 are parallel.

4. Line through Two Points. The line through the two

points {xi, 2/i, Zi)f (x2, y^, z^) has the direction components

^2 — ^l, Vl — 2/1, 2^2 — ^i.

482 ANALYTIC GEOMETRY

It can be considered, then, as the line which has these direction

components and goes through the point (xi, 2/1, ^i)- Conse-
quently, by (2), § 3, it is represented by the equations

(1) X— Xi = p(X2-Xi), 2/ - 2/1 = p(2/2 — 2/l)> Z — Zi = p{Z2-Zi).

Instead of (1) we can write, as the equations of the line, the
continued equality
^ — ^1 ^ y — Vi ^ z — zi
(2) ^

^2 — ^1 2/2 — ^2 — zi
2/1

Since (2) is an abbreviated form of (1), it follows that, if a

denominator in (2) is zero, the corresponding numerator is also
zero. Thus, if the two points are (3, 5, — 4), (8, 5, — 4), so
that 2/2 — 2/1 = ^ '^^^ Z2 — Zi = 0, we have y — y^ = and
z — Zi = 0, that is,

2/-5 = 0, 12 + 4 = 0.
These are, then, the equations of the line. The result might
have been obtained directly by noting in the beginning that
the 2/-coordinates, and also the ^-coordinates, of the two points
are equal and by concluding, then, that the line is parallel to
the axis of x.

EXERCISES
Find the equations of each of the following lines.

1. Through (2, 5, 8), (- 1, 6, 3).

2. Through (- 1, 0, 2), (3, 4, 6).

3. Through the origin and (5, — 2, 3).

4. Through (2, 0, 3), (0, 3, 2).

5. Through (2, -5,8), (2,3,7).

6. Through (3, - 2, - 5), (3, - 2, 6).

7. The edges of the tetrahedron of Ch. XIX, § 6, Ex. 9.

8. The edge of the quadrangular pyramid of Ch. XIX, § 5,

Ex. 10.
 THE STRAIGHT LINE 483

5. Line or Plane in Given Relationship to Given Lines or

Planes. Problem 1. To iind the equations of the line through
a given point perpendicular to a given plane.
The direction components of the line are those of a normal to
the given plane and hence can easily be found. The problem
then becomes that of finding the equations of a line through a
given point with given direction components.
For example, if the given point is the origin and the given
plane is

the required line goes through (0, 0, 0) and has the direction
components 3, — 4, 5. Hence its equations are

a;-0 ^ y-0 ^ g-0

3 -4 " 5 '

or 20a; = -15y = 12z.

Problem 2. To find the equation of the plane through a given

point perpendicular to a given line.
The direction of a normal to the plane is that of the given line
and therefore the direction components of the normal are easily
written down. We then have the problem of finding the equa-
tion of a plane through a given point with given direction com-
ponents of its normals.
Thus, if the point is (3, — 2, 1) and the line is given, as the
intersection of two planes, by the equations,
3a:-52/-224-6 = 0, 4ic 4-?/ + 3^-7 = 0,
the direction components of the line are, by (2), § 2,

-5 -2 -2 3 3 -5
1 3 3 4 4 1

i.e. — 13, — 17, 23, or 13, 17, — 23.

The required plane has
these direction components for normals and passes through
its

the point (3, — 2, 1). Consequently, its equation is

13(a; - 3) + 11 {y + 2) - 23(^ - 1)= 0,

or 13a; 17 - 23 2 + 18 = 0.
-|- 2/
484 ANALYTIC GEOMETRY

Problem 3. Let two non-parallel lines, L^ and Xo, and a point

P be given. Through P parallel to each of the lines there is a
unique plane. To determine this plane.
A normal to the plane is a common perpendicular to the lines
Li and L2 and so its direction components can be found by the
method of Ch. XVIII, § 5. The problem is then that of finding
the equation of a plane through a given point with given direc-
tion components of its normals.
If jLi and L2 have the equations

x—l_y — 2__z cc+ _y _

3 z —1
5 ~ 3 ~ 2' "X ~2~ 3
'

their direction components are, respectively, 5, 3, —2 and 4, 2,

3. The direction components of a common perpendicular to

them are, by Ch. XVIII, § 5, (6), 13, - 23, - 2. Thus the
plane parallel to Li and L2 and passing through a given point,
say (3, 2, — 4), has the equation

13 (x - 3) - 23 - 2 (^ 4- 4) = 0,
(2/
- 2)

or 13a^-23?/-22-l = 0.
Problem 4. Given two intersecting planes, Mi and M2, and
a point P. Through P parallel to each of the planes there is a
unique line. To find this line.
Since the line is parallel to each of the planes, it is parallel
to their line of intersection. It is, therefore, itself the line of
intersection of the two planes which pass through P and are
parallel respectively to Mi and M2.
For example, if P is (2, 0, — 1) and 3fi and M2 are
2x-3y-\-z-6 = 0, 'ix -2y -\-3z -\-9 = 0,
the planes through P parallel respectively to 3fi and M2 have
the equations

2(x-2)-3y-{-{z + l)=0, 4:(x - 2)-2y-\- 3(z -{-!)= 0,

or 2x-3y-\-z-3==0, 4.x -2y 3z - 5 = 0.
-{-

These equations, considered simultaneously, represent the re-

quired line.
 THE STRAIGHT LINE 485

The problem might have been solved by determining the di-

rection components of the line of intersection of M^ and M2J as
given by and by finding the equations of the line which
(2), § 2,
has these direction components and passes through P.

Problem 5. To find the equation of a line which passes

through a given point and is parallel to a given line.
If the line is given as the intersection of two planes, this is
the previous problem. If its equations are given in the form of
a continued equality, the solution is simple. We leave it to the
student.

Problem 6. To find the equations of a line which passes

through a given point and is perpendicular to each of two given
non-parallel lines.
The solution of this problem we also leave to the student.

EXERCISES
In each one of the following exercises in which it is possible,
solve the given problem directly, by inspection of a figure.

Find the equations of the line passing through the given

point and perpendicular to the given plane.

Point Plane
1. (2,-8,3), a;+ 2?/- 3^-2 = 0.
2. (0,0,0), •
6a?- 2^-1- 52 + 3 = 0.
3. (3, -4,0), 5a;- 3:^ + 4 = 0.
4. (-1,2,5), 2^ + 3 = 0.

Find the equation of the plane passing through the given

point and perpendicular to the given line.

Point Line
, .
^-^'-'^^'
^ .. |2a;-32/ + 62;-4 = 0,
^'
\4.x- + 5. + 2 = 0.
2/

a (C)a c^\ f3a; -h22/+ 2-3 = 0,

^^'^'^)'
^-
i + 22/ + 3. + 2 = 0.
a;
486 ANALYTIC GEOMETRY
 THE STRAIGHT LINE 487

Point Lines
24. (2, 4, 0) The lines given in Exs. 5, 6.

25. (0, 0, 3)1 The lines given in Ex. 13.

6. Angle between a Line and a Plane. Given a plane M

and a line L which is not perpendicular to the plane. Pro-
ject the line on the plane. The acute
angle, <^, which the line makes with this
projection is the angle between the line
and the plane. It may be determined
by finding the acute angle, 0, which the FiGTe"
'

line makes with a normal N

to the plane.
For, it is clear that 6 and <^ are complementary angles.

Example. Find the angle between the line

2a;H-2 = ?/-fl=-4^ + 28
and the plane
2x-\-3y-2z^ 1 = 0.
The direction components of the line are 2, 4, —1 ;
those of
a normal to the plane are 2, 3, — 2. Hence

cos 6
.
=±—
2.2 + 4-3+( -l)(-2)
,
^\ '
\ /
,

V4 + 16 + 1V4 + 9 + 4
where we are to take that sign which makes the right-hand
side positive.

Thus, cos ^ = -i?:^ = 0.9527, and (9 = 17° 42'.

. V357
Finally, <^ = 90° - ^ = 72° 18'.

EXERCISES
Find the angle between the given line and the given plane.

Line Plane
1. Sx+ 3 = 2y + 2=-6z-12, Sx + y -{-2z-^l =0.
|3a^-42/ + 2. = 0, 3a;-2^-12 = 0. ,

2.
[4:X— 3y + z = 5,
488 ANALYTIC GEOMETRY

3. Through (0, 0, 0), (1, 2, - 2), 2x - 6y + 3z - = 0. 4.

4. Through (2, 5, 3), (4, - 1, 6), y-]-2z + A = 0.

5. Find the angles which the plane given in Ex. 1 makes

with the coordinate axes.

6. Find the angles which the line given in Ex. 2 makes

with the coordinate planes.

7. Point of Intersection of a Line and a Plane. Given a plane,

(1) Ax-{-By-\-Cz + D = 0,
and a line, •

(2) A^x + B,y + Ciz + i)i = 0, A2X + B^y + C2Z + A = 0,

which intersects the plane. The coordinates of the point of
intersection satisfy the equation of the plane, since the point
lies in the plane. They also satisfy each of the equations of
the line, for the point lies on the line. They are, then, the
simultaneous solution of the three equations

Ax -{-By -\-Cz +D =0,

(3) A^x + B,y + C,z + D^ = 0,
A2X + B^y + C2Z + A = 0.

Example. Find the coordinates of the point of intersection

of the line and the plane of § 6. Take, as the equations of
the line, those obtained by equating the first and second mem-
bers, and then the first and third members, of the continued
equality which represents the line. Then the three equations,
which are to be solved simultaneously, are

2x-y = -l, x-\-2z = lS, 2x-\-Sy-2z = -l.

The solution is found to be x = l, y = 3, z = 6. Thus, the
line meets the plane in the point (1, 3, 6).

Intersection of a Curve and a Surface. The above method

applies also to the problem of finding the point (or points)
of intersection of a curve and a surface which are given by
their equations. The two equations of the curve and the
 THE STRAIGHT LINE 489

equation of the surface are to be considered as simultaneous

equations in the unknowns, x, y, z, and solved as such.
Plane and Line Arbitrary. The line (2) has one and just
one point in common with the plane (1), if and only if the
three planes (3) intersect in a single point, i.e. by the theorem
of Ch. XXIX, § 10, if and only if the determinant of the coeffi-

cients of X, y, z in equations (3) does not vanish.

If this determinant vanishes, it follows, either directly or
from the discussion in Ch. XXIX, § 10, that the line either is

parallel to the plane or lies in it.

EXERCISES
Show that the given line has just one point in common with
the given plane and find the coordinates of the point.
1. The line and plane of Ex. 1, § 6.

2. The line and plane of Ex. 2, § 6.

3. The line and plane of Ex. 3, § 6.

Find the points of intersection of the given curve with the

given surface. Draw a figure for each exercise.

Curve Surface
4. x=y= z, flj- -|- 2/-
-f 2;2 = 1.
5. + + 22 = 29, 2 = 2,
a;2 ?/2 4:X-Sy = 0.
6. 12-6x = 2y-{-2 = 3z-9, x^ -\-
y''
= 5.
7. + + 22 = 12, x = y,
a;^ 2/- x^-\-y^ = 8.
Eind out all you can about the relative positions of the given
lineand the given plane.
Li7ie Plane
\2x-3y-\-z=:0,
4.x-6y + 2z+l = 0.
9. x =y= z. 5x-{-Sy — Sz — S = 0.
,^
10.
\2x-{-Sy-S
^ ^ = 0, ' Q a;—
8 — 6^ = A0.
V
^
[3x-2y-\-l = 0,
490 ANALYTIC GEOMETRY

8. Parametric Representation of a Curve. The Straight Line.

Given a directed passing through
straight line
the point Pq {xq, 2/0, Zq) and having the direction
:

cosines cos a, cos y8, cos y. Let (x^ y, z) be an P :

arbitrary point of the line other than Pq and let

r be the algebraic distance from Pq to P, positive
if the direction from Pq to P is that of the line

and negative if this direction is opposite to that

of the line.
The projections of PqP, each divided by r, are
equal respectively to the direction cosines of the
line, by Ch. XVIII, § 1, Th. 2. Thus

X — Xc^
= cos a, y
-^
——
Va
= cos (i,
o z — Zq — cos
y.
r r r

These equations can be put into the form

(1) x = Xq-\- r cos a, ?/ = 2/0 + r cos fi, z = Zq-\- r cos y.

Equations (1) give the coordinates {x, y, z) of the point P on
the given line at the arbitrary distance r from Pq. If r is

allowed to vary through all values, positive, zero, and nega-

tive, P takes on all positions on the line, and always its
coordinates are given by equations (1). These equations, then,
represent the line. Since they express the coordinates of the
point P : (x, y, z) tracing the line in terms of the auxiliary
variable, or parameter, r, we call them a parametric representor
tion of the line.
If the line is determined by the point Pq : (xq, yQ, Zq) and its

direction components I, m, n, we have, according to (2), § 3,

the following parametric representation

(2) x= XQ-}-pl, y = yo + p'm, z = Zo-\-pn.

The parameter p is not, in general, equal to the distance from
Pq to P : (x, y, z), but is merely proportional to this distance.*

* Cf. footnote, p. 479.

 :

THE STRAIGHT LINE 491

The Helix. Given the cylinder

^2 _|_ 2^2 _ q2.

with the axis of z as axis. The circle in the (a;, 2/)-plane on

which the cylinder is erected has the parametric representa-
tion (Ch. yil, § 10)
X = a cos 6, y = CL sin 0, 0,

where the angle which the radius, OP, to the point

is

P : makes with the positive axis of x. On the ruling of

(x, y, 0)

the cylinder through P mark the point P' at a height above P *

equal to a constant multiple, kO, of the
angle 0. The coordinates of P' are

(3) = a cos ^, y=-asmO, z = kO.

a;

When = 0, P and P' coincide in the

point Pq on the axis of x. As $ increases
from to 2 TT, the point P traces the circle,
and the point P', always directly above P,
traces a locus on the cylinder, encircling it

just once. When increases from 2 7r to

4:7r, P retraces the circle, whereas P' con-
tinues on its rising path, encircling the
cylinder a second time. Consequently,
when $ increases through all positive values,
the locus traced by P' encircles the cylinder
infinitely many times.
As decreases from zero through all Fig. 8
negative values, P', starting from Pq, en-
circles thelower half of the cylinder infinitely many times.
The complete locus of P' is, then, an unbroken curve continu-
ously winding about the cylinder in both directions. It is this
curve which is represented parametrically by the equations (3).
Since the height of P' above (or below) P is always propor-
tional to the angle through which the radius OP of the circle
has turned, the curve (3) which P' traces is mounting on the
* Above P, if d is positive ; below P, if ^ is negative.
 :

492 ANALYTIC GEOMETRY

cylinder with a uniforiyi steepness. It is the curve of the

thread of a machine screw and is called a circular screw or a
circular helix.

The Twisted Cubic. Consider the curve represented para-

metrically by the equations

(4) x = at, y = hP, z = cfy

where t is the parameter, and a, b, c are constants, not zero.
Like the helix, this curve is a twisted curve. It is known as
a tivisted cubic.

Points of Intersection of a Curve and a Surface. Example 1.

The straight line of § 6 can be represented parametrically by
setting each of the members of the continued equality

2x + 2=y-\-l = -4.z-\-2S
equal to a parameter t and by solving the three resulting
equations for x, y, z:

(5) x = it-l, y = t-l, z = l-\t.

A point of this line lies in the plane of § 6,

2x + ?,y-2z + l = 0,
if and only if its coordinates, as given by (5), satisfy the equa-
tion of the plane ; i.e. if and only if ^ is a solution of the
equation
2(ii_l)+ 3(^-1)- 2(7 -10+ 1 = 0.
or f - 18 = 0.
^

Hence t = 4. But the i point = 4 of the

line (5), i.e. the point
corresponding to the value 4 for the parameter t, has the
coordinates

a; = 2-l = l, 2/
= 4-l=3, z = l -1 = Q.
Hence the line intersects the plane in the point (1, 3, 6).

Example 2. Find the points of intersection of the twisted

cubic
 THE STRAIGHT LINE 493

(6) x = t, = y z = t^, t^

with the plane 2x -\-y — z = 0.

A point of the cubic lies in the given plane when and only
when its coordinates, as given by (6), satisfy the equation of
the plane. Hence the solutions of the equation

for t determine all the points of intersection.

One solution is ^ = ; the others are t = 2, = —
t 1. The
points of the cubic (6) corresponding to these values of t are
respectively (0, 0, 0), (2, 4, 8), (- 1, 1, 1). Thus the cubic _
meets the plane in these three points and in no further point.

The above method may be used to find the points of inter-

section of any given curve with a given surface, provided the
curve is defined by a parametric representation. The sim-
plicity and effectiveness of the method is one of the advan-
tages of representing a curve parametrically.

EXERCISES

Find a parametric representation for each of the following

straight lines.

1. Through (2, — 3, 5) with the direction cosines |-, — S-,

f.
2. Through (2, —3, 5) with the direction components
2, - 6, 3.

3. Through {x^, y^, z{), {X2, 2/2, 22) ; cf. (1), § 4.

4. Sx-4:y-6z-\-T = 0, 2x-y-2z + l = 0.
Suggestion. First find the direction components of the line
and the coordinates of a point on it.

The same for the lines of the following exercises of previous

paragraphs.

5. Ex. 1, § 3. 7. Ex. 1, § 4. 9. Ex. 1, § 2.

6. Ex. 2, § 3. 8. Ex. 2, § 4. 10. Ex. 3, § 2.

494 ANALYTIC GEOMETRY

Find equations for the lines with the following parametric

representations.

11. a; = 3 — 2^, y = - i, 2 =
t 3^+2.
12. x = 3t, y = 5t, z=—St.
13. If from the point (3, 2, — 6) one proceeds 12 units in
the direction whose cosines are |,
— '^, |-, what are the coordi-
nates of the point reached ? Ans. (11, — 2, 2).

14. Draw to scale the circular helix for which a = 4, A: = 2.

15. Show that the twisted cubic (6) is the total intersection
of the parabolic cylinder y =
x^ with the cylinder z x^. =
Hence construct the cubic.

16. Find a parametric representation of the curve

y^ = 2x, z=Sy^.

17. The same for the curve of § 1, Example 4.

Suggestion. Let x = sin 2 0.

By the method of this paragraph find, in each of the follow-
ing exercises, the point (or points) of intersection of the given
curve and the given surface.
18. The line and plane of Ex. 1, § 6.

19. The line and plane of Ex. 3, § 6.

20. The line and cylinder of Ex. 6, § 7.

EXERCISES ON CHAPTER XX

1. Find the equations of the line which passes through

the point (1, —2, 3) and intersects the axis of z at right
angles.

Find the coordinates of the points in which the given

•

linemeets the coordinate planes and hence construct the

line.

^ \ x-{-oy- z-7=:0. . y-\-2 o

2x— oy-{-Sz-\-l = 0. 3
 THE STRAIGHT LINE 495

Show that the first of the two following lines intersects the
axis of z and that the second intersects the axis of x.

[2iB + 3?/- 2 + 2 = 0, a; + 4 ^ 2?/ + 6 ^ 32 +4

/*•
I
a;-22/ + 2z-4 = 0. ^'
4 3 2

6. What is the condition that the line

A^x + B^y + Ci2 + i)i = 0, A^x + B^y + O22; + Dg = 0,
where CiC^ ^ 0, meet the axis of 2; ? Ans. C1D2 — C2D1 = 0.
Show that the following lines are identical.

\2x + y-bz + 2 = 0, |4a; + 2/- 8)2 + 3 = 0,

[6a;-2?/-52 + l = 0; [
2x - 2?/ + 2 - 1 = 0.
fa; + ?/-2; + l = 0, ^_+l=:LzJ( = 2_l
|2a; + 32/-l = 0; 3 2

9. Find the equations of the altitudes of the tetrahedron of

Ch. XIX, § 6, Ex. 9.

10. Find the equation of the plane which contains the point
(2, — 1, 5), is perpendicular to the plane 2x~y + 3z = 4:, and
is parallel to the line

5x-h2y-{-3z = 0, ^x +y -{- 2z -8= 0.

Ans. 3a; - 9y - 50 + 10 = 0.
11. Find the equation of the plane which passes through the
points (2, — 1, 3), (5, 0, 2) and is parallel to the line

2x-5 = l-y = 2-3z.

12. Find the equations of the line which contains the point
(4, 2, — 3), is parallel to the plane x -\- y -\- z = 0, and is per-

pendicular to the line whose equations are x 2 y — z = 5, -{-

2 = 4. Ans. 6x — 24 = 3?/ — 6 = — 22 — 6.
13. A line is parallel to the plane 2x — 3y + = 0. If the 4:

perpendicular from the origin on the line meets it in the point

(2, 5, — 3), what are the equations of the line ?

14. Show that the equation of the plane which passes

through the point (xq, y^, Zq) and is parallel to two (non-parallel)
496 ANALYTIC GEOMETRY

lines having li, mi, ??i and I2, m^^ n^ as direction components
can be written in the form
X Xq y 2/0 ^ '^0

?i mi Til 0.

15. Two with the direction components l^, mi, Ui and

lines
Z2, m2, 712 intersect in the point
(xq, 2/0? ^^o). Find the equation
of the plane containing them.

16. Find the equations of the line determined by the point

(2, — 1, 0) and the point of intersection of the three planes

X — y -{-z = 0, Sx — 2y-\-4:Z = 0, 2x-{-y-\-z — 4 = 0.

17. Find the equations of the line through the origin and
the point of intersection of the plane and the line whose equa-
tions are x -r2y — 3z -\- A = and 3x-{-l = 2 — 2y = z-\-S.
Determine the equations of the line which lies in the
18.
plane 2x — y-\-z — 3 = and is perpendicular to the line

3x-^2y-z-8 = 0, x-3y + l =
in the point in which this line meets the plane.

Ans.
. x — 2 = ^y — 1 = z
12 19 5

19. A line through the origin with the direction cosines

fj
~f
f intersects the plane 3x-{-6y-\-2z
5
— =
Q> in the
point P. Find the length of OP. Ans. 14.
Suggestion. Represent the line parametrically.
20. A line through the point A: (3, — 2, 5) with the direc-
tion cosines -|,
i, -| meets the plane 2x + 3y — z-\-1 = 0\ti the

point P. What is the length of ^P?

Loci

21. Find the locus of a point which is always equidistant

from the three points (2, 0, 3), (0, - 2, 1), (4, 2, 0).
 THE STRAIGHT LINE 497

22. Determine the point in the plane x — y — 2z = which

is equidistant from the three points (2, 1, 5), (4, — 3, 1),
(-2,-1,3). Ans. (I, l,i).
23. Show that the locus of a point moving so that it is
always equidistant from three given non-collinear points is a
line perpendicular to the plane of the three points. In what
point does it intersect this plane ?
Suggestion. Choose the coordinate axes skillfully.

Find the locus of a point which is equidistant from the

24.
points (2, 3, 0), (4, — 1, 2) and also equidistant from the
points (5, 2, - 3), (3, 0, 1).

25.The previous problem for any four non-coplanar points

Pi, P2 and P3, P4. Show that the locus is a line which is
perpendicular to each of the lines P1P2, P3P4 and goes
through the center of the sphere determined by the four points.
26. Find the locus of a point which moves so that the dif-
ference of the squares of its distances from two given points is

constant.
 CHAPTER XXI
THE PLANE AND THE STRAIGHT LINE.
ADVANCED METHODS

1. Linear Combination of Two Planes. A linear combination

of two planes,
(1) A,x 4- ^1^ -f C,z + D, = 0,
(2) A2X + B^y -f C2Z -h A =0,
shall be defined as any plane

(3) Ai (A,x -\-B^+ Ciz + D,) + X^iAo^ + ^22/ + C2Z + A) = 0,

whose equation is obtained by multiplying the equations of
the two planes by constants, Ai, Xo, and adding the results.
The constants Ai, A2 can be chosen at pleasure, provided merely
that the coefficients of x, y, z in (3), namely Ai^i + A2^25
Ai^i + A2-B2, AiCi + A2O2, do not all vanish. In particular, the
case Ai = A2 = is thus excluded.
For example, if the given planes are

(4) x-2y-2z + 9 = 0,
(5) 2x-Sy-2z + S = 0,
and we multiply the equation of the first by — 3 and the equa-
tion of the second by 2 and add, the plane defined by the
resulting equation,

(6) x-^2z-ll = 0,

is a linear combination of the given planes.

If the planes (1) and (2) meet, the plane (3) passes through
their line of intersection. For, if (xq, y^ Zq,) is an arbitrary
point of this line, then
498
 :

ADVANCED METHODS 499

.^. ^la^o 4- ^i2/o + C,Zq + Z)i = 0,

since the point lies in both the planes (1) and (2). If it is also
to lie in the plane (3), the equation

A-i {AiXq + B^ijo -t- G^Zq + Dy) + As (^2^0 + ^2^/0 + C's^o -f A) =

must be a true equation ; and this it is, since the two paren-
theses on the left-hand side both vanish by virtue of equa-
tions (7). Thus the plane (3) contains the arbitrary point
(xq, 2/0, ^0) oil tlie line
of intersection of the planes (1) and (2)
and hence contains the whole line, q. e. d.
We have thus proved the theorem
Theorem 1. A linear combination of two intersecting jjlanes
is a plane through their line of intersection.
For example, the plane (6) passes through the line of inter-
section of the planes (4) and (5).
If the planes (1) and (2) are parallel, it follows, by Ch. XIX,
§ 7, Th. 2, that

The direction components of the normals to the plane (3),

namely,
Xi-4i 4- X2-42, ^iBi -r A2S2, Ai6\ + A2C2,
become, then,
(Ai -f PA2) A„ (Ai + M2) B„ (Ai + p Ao) Ci.
Now Ai -f pk2 =^ 0, since otherwise the coefficients of x, y, z in
(3) would all be zero ; that is, in this case we must exclude,
according to the definition, not only the values Ai = Ag = but
also the values of Ai and A2 for which A1/A2 = — p- It follows,
then, that the plane (3) is parallel to or identical with the
plane (1). Thus we have the theorem :

Theorem 2. A linear combination of tico parallel planes is

a plane parallel to them or coincident vnth one of them.
Plane through a Line and a Point. It is now a simple matter
to find the equation of a plane d-etermined by a line and a point.
500 ANALYTIC GEOMETRY

For example, let the line be the line of intersection of the

planes (4) and (5) and let the point be (—5, — 1, 2). By
Th. 1, the plane

(8) Ai(a^-22/-2:3 + 9)+A2(2a;-3?/-22 + 8)=0

passes through the given line. If it is also to contain the
given point, (—5, — 1, 2), we must have
Xi(- 5 + 2 - 4 9)+ A2(- 10 + 3 - 4 + 8)= 0,
-f-

2Ai-3A2 = 0.
This equation determines the ratio A1/A2. It will be satisfied
if, in particular, we take Ai = 3 and A2 = 2. Then (8) becomes

or 7x-12y- lOz + 43 = 0.
This is the equation of the required plane.
In the general case, when the given line is the line common
to two intersecting planes, (1) and (2), and {xi, 2/1, ^i) is the
given point, not on the line, the procedure is quite the same.
The plane (3) passes through the given line. Demanding,
further, that it contain the point (xi, t/i, ^i) leads to an equa-
tion for the determination of the ratio A1/A2, and any values
for Ai and A2 which have this ratio yield, when substituted in
(3), the equation of the required plane.

Converses of TJieorems 1, 2. Since every plane through the

line of intersection of the given planes (1) and (2) can be
thought of as determined by this line and a point (xi, y^, z-^
external to it, we have proved that every plane through the
line of intersection of the given planes is a linear combination
of them. This is the converse of Theorem 1.
The converse of Theorem 2 can be proved in a similar
manner. The details are left to the student. Both converses
can be stated in a single theorem.
Theorem 3. Any plane through the line of intersection of
two intersecting planes, or parallel to two parallel planes, is a
linear combination of the two planes.
 : :

ADVANCED METHODS 501

Projecting Planes of a Line. Consider the line L

^^ 2 a; - 3 - 2 2 + 8 = 0,
7/
I

in which the planes (4) and (5) intersect, and also the plane (6)

a; + 2 2 - 11 = 0.
This plane passes through L^ since it is a linear combination of
the planes (4) and (5). In particular, it is the plane through
L which is perpendicular to the {z, x)-
plane, for equation (6) contains no term
in y. It is, then, the plane which projects
L on the {z, x)-plane. Accordingly, it is

known as di projecting plane of L.

Equation (6) represents, in space, this
projecting plane. Considered merely in
the {z, a;)-plane, it defines the line which
is the actual projection of L on the Fig. i
{z, a;)-plane.
By combining equations (4) and (5) linearly so that the re-
sulting equation contains no term in x, e.g. by multiplying
the first of the equations by —2 and adding it to the second,
we obtain the equation of the plane which projects L on the
(?/, 2;)-plane, namely,
(10) 2/ + 2 2 - 10 = 0.
In a similar manner the equation of the plane which projects
L on the (x, 2/)-plane is found to be
(11) a; - - 1 = 0.
2/

The planes (6), (10), (11) are the three projecting planes
and
of L. Any two of the three projecting planes of a line will,
in general,* determine the line. Eor example, the pair of
equations

(12) a; + 22-ll=0, ?/ + 2z-10 =

* Exceptions occur when the line is parallel to, or lies in, a coordinate
plane ; cf . Ex. 13.
502 ANALYTIC GEOMETRY

isas proper an analytic representation of L as the pair (9),

and much more simple.
Equations (6), (10), and (11) are equivalent to the continued
equality

(13) x-ll = y-10 = -2z,

a.nd conversely. In other words, the representation of a line
by means same as
of its projecting planes is essentially the
the representation of by means of a continued equality of
it

the usual form. The three equations which result from

equating the members of the equality are the equations of
the projecting planes.
Furthermore, we now have a method of finding, from the
representation of a line as the intersection of two planes, a
representation of it by a continued equality. Thus, in the
case of L, we passed from the equations (9) to the continued
equality (13).

EXERCISES
Find the equation of the plane determined by the given
line and the given point.

Line Point

\5x + 2y-z + 5 = 0,
W">">

[3x— oy — 2z = S, ^ ^

3. Sx-5y = 6,2x-{-Sz = 9, (4,3,-5).

6. Find the equation of the plane containing the line of

Ex. 1 and having the intercept 2 on the axis of y.

6. Find the equation of the plane passing through the line

of Ex. 2 and having equal intercepts, not zero, on the axes of
a; and 2/. Arts. 13 a; + 13 2/
4- 22 2 + 8 = 0.
 ADVANCED METHODS 503

7. What is the equation of the plane which contains the line

of Ex. 4 and is perpendicular to the plane Sx — y -{• 4:Z = 0?
Ans. 4:X-{-Sy — z — 1 = 0.
8. Find the equations of the line which is the projection of
the line of Ex. 3 on the plane 2x-hy — 3z-\-5 = 0.

9. Find the equations of the projecting planes of the line

of Ex. 1 and from them determine a continued equality which
represents the line.
10. The preceding exercise for the line of Ex. 2.

11. The line of Ex. 3 is defined by two of its projecting

planes. What is the equation of the third ?
12. What are the equations of the projecting planes of the
line of Ex. 4 ?

13. A line which is not parallel to or in a coordinate plane

has three projecting planes, which are distinct; a line parallel
to or in one coordinate plane has three projecting planes, just
two of which are identical ; a line parallel to or in tivo coordi-
nate planes — i.e. with an axis
parallel to or coincident has —
but two projecting planes, which are distinct. Consequently
there are always at least two distinct projecting planes of a line
and the line is determined by them. Prove these statements.
14. A line not parallel to or in the (x, 2/)-plane can be
represented by equations of the form

x = az ->r^, y = cz -\- d\

a line parallel to or in the (x, ?/)-plane can have its equations

put into the form
y = ax -\- b, z = c,

unless it is parallel to or coincident with the ^/-axis ; in this

case, its equations can be written as

x = a, z = h.
Prove these statements.
15. Prove that the plane determined by the point (x2, 2/2, ^2)
and the line through the point (x^, 2/1, ^\) with the direction
504 ANALYTIC GEOMETRY

components I, m, n can have its equation written in the form

, X —xi y -yi z -z^

X2 —Xi 2/2
- 2/1 Z2-z^ = 0.
I m n

Suggestion. Determine the direction components of the

normals to the plane.

2. Three Planes through a Line. Three Points on a Line.*

By means of the results of the preceding paragraph we can
prove the following theorem.
Theorem 1. Three planes pass through a line or are paral-
lel, ivhen and only when any one of them is a linear combina-
tion of the other two.
If the three planes pass through a line, or are parallel, any
one of them passes through the line of intersection of the
other two, or is parallel to the other two. Consequently, by
Th. 3, § 1, this plane is a linear combination of the other two.
Conversely, if a x)a,rticular one of the planes is a linear com-
bination of the other two, it goes through the line of intersec-
tion of these two, if they intersect, or is parallel to them, if
they are parallel (Ths. 1, 2, § 1). It follows then, further-
more, by the first part of the proof, that any one of the three
planes is a linear combination of the other two, q. e. d.

For example, the three planes,

3aj-22/+ 2+ 6 = 0,
(1) 2a; + 52/ -32- 2 = 0,
4a; -9?/ + 52 -f 14 = 0,

pass through a line, inasmuch as the equation of the third can

be obtained by multiplying that of the first by 2, that of the
second by — 1, and by adding the results.

Three Points on a Line. The three points P^ : (ajj, 2/1, ^i)y

A : (aJ2, 2/2, 2:2), Pz '

{^i, 2/3, 2^3) lie on a line if and only if the

* It is assumed, liere and in § 4, that the given planes, or the given

points, are distinct.
 ADVANCED METHODS 505

direction components of P1P3 are proportional to those of

x^ — Xi = p{x2 — a^i), y^ — y^=: p(y2 — yi), z^ — z^^ p{z2 — 2i).

These equations can be rewritten as
X^={1- p)Xi-[-pX2,
(2) 2/3=(l-p)2/i+P2/2>
Z^=(l-p)Zi-\-p Zo.

They then say that the coordinates of P3 are a linear combina-

tion of the coordinates of Pi and P2 with constants of combina-
tion, 1 —p and p, ivhose sum is unity. Since any one of the
points might have been called P3, this result can be stated
more generally.

Theorem 2. Three points lie on a line lohen and only when

the coordinates of any one of them can he expressed as a linear

combination of those of the other two, with constants of combina-
tion ivhose sum is unity.

This theorem is of importance because of the analogy be-

tween it and Theorem 1, and because of its theoretical value
in later work. It has not the practical value of Theorem 1,
since testing three points for collinearity can be more easily
done directly.
Thus, if the three points are (2, -1, 5), (4, 2, 6), (-2, -7, 3),
the direction components of P1P2 and P1P3 are, respectively,
2, 3, 1 and — 4, — 6, — 2. Since these triples are propor-
tional, the three points lie on a line.

EXERCISES
What can you say of the three planes in each of the follow-
ing exercises ?
1. = 2, Sx + y -{-2z = l, 5x — 5y — 6z = 7.
2x — y — z
2. x-\-Sy-z= l, 3x-5y-^Tz = 3, Sx -^ 2y 2z = S. -{-

3. 6x-Sy + 9z = 2, 2x-y-\-3z = 0, -4:X-\-2y - 6z=S.

Are the three given points collinear ?
 ;

506 ANALYTIC GEOMETRY

4. (5, 3, 4), (1, 5, 10), (11, 0, - 5).

5. (- 13, 12, - 15), (- 5, 6, - 11), (7, - 3, - 5).
6. (2, - 3, 8), (5, 4, 7), (8, 10, 6).

7. Determine k so that the three planes

A;£c_32/^_;2 = 2, 3a; 4- 2?/ 4- 4^ = 1, x-^y-2z = S
will pass through a line.

8. Determine k so that the three points (2, 3, fc), (5, 5, 1),

( — 1, 1, 9) will be collinear.

Line in a Plane. From Theorem 1 of the preceding para-

3.

graph follows immediately the theorem

Theorem. A line lies in a plane, if and only if the plane is a

linear combination of any two planes which determine the line.

For example, the line of intersection of the first two of the
planes (1), § 2 lies in the third plane, since the third plane
was shown to be a linear combination of the first two.

A second method of testing whether or not a given line lies

in a given plane presents itself if the line is represented para-

metrically. The line through the point (2, — 1, 3) with the
direction components 3, 2, — 4 has the parametric representa-

tion (Ch. XX, § 8, (2)) :

a; = 3^-f-2, y = 2t-l, z = -4.t-\-S.

It will lie in the plane
2^ + 52/ + 42 -11 = 0,

if and only if the coordinates of every one of its points satisfy

the equation of the plane, i.e. if and only if

2(3 ^ + 2) + 5(2 - 1) + 4(- 4

^ f -h 3) - 11 =
is a true equation for all values of t. But the equation re-

duces to
• i + = 0,
and so is satisfied by all values of t. Consequently, the line
lies in the plane.
 ADVANCED METHODS 507

It is clear that this method can also be applied to test

whether or not a given curve, represented parametrically, lies
on a given surface.

EXERCISES
In each of the following exercises determine whether or not
the given line lies in the given plane. Apply both methods.
Line Plane
x— l _ y -^2 _ z -\-3
— z — 2 = 0'
1.
2
~ -3 ~ -1 '
x-^y

2. 2x-3 = l-oy = 3z, 4,x-\-oy -3z-7 = 0.

3. 'ix-l=-3y = l-2z, 2x — y-{-3z-{-5 = 0.
4. Find the conditions under which the line through the
point (a^o, 2/o, Zq) with the direction components Z, m, n will lie

in the plane Ax -\- By -{- Cz -{- D = 0.

A71S. Al + Bm H- Cn = 0, Axq + ByQ -i- Czq -\- D = 0.
5. Does the twisted cubic x = t, y = t^, z = 2f lie on the
surface 2q^ —z= 0?
6. Show that the curve
x = a sin^ t, y = ci sin t cos t, z = a cos t

lies on the sphere x^ -\- y^ -\- z"^ = a^.

4. Four Points in a Plane. Four Planes through a Point.
Given the four points Pi, Po, P3, P4, with the coordinates
(xi, yi, 21), (X2, 2/2, ^2), (xs, 2/3, 2:3), (0:4, 2/4, 24)- To determine
when they lie in a plane.
Unless all four points lie on a line, there will be three of
them which determine a plane. Let these three be P2, P3, P4.
The equation of the plane through them is, by Ch. XIX, § 6,
X 1

1
0.
1

1
508 ANALYTIC GEOMETRY

The four points will be coplanar if and only if Pi : (xi, i/i, ^i)

lies in this plane, i.e. if and only if

Xi

(1)
 ADVANCED METHODS 509

For, if the normals to the four planes are parallel to a plane,

the normals of any three are parallel to this plane, and hence
the determinant of their direction components vanishes, by Ch.
XVIII, § 6. But this determinant is one of the four determi-
nants A in question. Hence all four determinants A vanish.
Conversely, if all four determinants A vanish, the normals
of each set of three planes are parallel to a plane, and this
plane can be taken as the same plane in all four cases.
=*

Hence the normals to all four planes are parallel to it.

Suppose, now, that the planes have one and only one point
in common. Then the four determinants A are not all zero.
For, if they were all zero, the normals to the four planes would

all be parallel to a plane. Consequently, either the four

planes would be all parallel or "all the lines of intersection
obtained by taking them in pairs would be parallel or identical
(cf. Ch. XIX, § 10), and hence the planes would have either

no point in common or a whole line of points in common.

But this contradicts the hypothesis that they meet in just one
point. At least one of the determinants A, then, does not
vanish. Let us assume, say, that A4 = AiBoC-^ is not zero. | \

Then the lirst three planes meet in a single point (Ch. XIX,
§ 10), whose coordinates, found by Cramer's rule, are

X —— , y —— .
,
» —
or

(d) a._-_____, y-^__, .--___

Since this point lies in the fourth plane, we must have
- ^4 AO2 A + A
1 I I
^iCo A I
- A ^lAA
I
I
+ A ^lA A =
I I

* This is obvious if the four planes are parallel. In the contrary case,
when at least two of the planes, say the first two, are not parallel, the state-
ment is substantiated as follows. The normals to the first two planes and
the third are parallel to a plane M\^ and the normals to the first two planes

and the fourth are But J/i and 'M21 since they
parallel to a plane Mi-
are both parallel to the normals of the two planes are, in any case,
first

parallel to each other and hence can always be taken as the same plane.
 :: :

510 '
ANALYTIC GEOMETRY

or, since the expression on the left is the development of the

determinant |
ABGD \
by the minors of the last row,

(4) \ABCD\=:0.
Conversely, if |
ABCD = \
and not all four of the determi-
nants A are zero, the planes (2) meet in a single point. For,
we can assume that A4 ^t 0. Then the first three planes meet
in a single point (3) and this point lies in the fourth plane,
since by hypothesis (4) holds.
Thus we have proved the theorem
Theorem 3. The four planes (2) meet in a single point if and
only if the determinant of their coefficients vanishes and not all

four minors Ai, A2, A3, A4 are zero.*

If the normals to the planes (2) are all parallel to a
plane, the determinant \ABCD\ obviously vanishes, for then
Ai = A2 = A3 = A4 = and the expansion of |
ABCD \
by the
minors of the fourth column, namely
- AAi + AAo - AA3 -f A^4
has the value zero.
Conversely, if |
ABCD \
vanishes by virtue of the vanishing
of Ai, A2, A3, A4, the normals of the planes (2) are, by Theorem
2, all parallel to a plane.
Consec[uently, we can combine Theorems 2 and 3 in the more
general, though less useful, theorem
Theorem 4. The four planes (2) meet in a single point or
their normals are all parallel to a plane, if and only if the deter-

minant of their coefficients vanishes.

Finally, we enumerate the cases which can occur when the

normals to the four planes are parallel to a plane. First, the

* Stated algebraically this theorem reads The four equations (2)

:

are compatible and have, moreover, a unique solution, if and only if

I
ABCD\ = and Ai, A2, A3, A4 are not all zero. This theorem includes
the theorem of Ch. XVI, § 9, Ex. 8, and also its converse. It is to be
noted that it was geometric considerations which led us here to a proof
which covered the converse as well as the theorem.
 ADVANCED METHODS 511

planes can be all parallel ; this case can easily be detected by-
inspection. Secondly, the planes can all go through a line;
the test of § 3 reveals this case. Finally, whatever lines of in-
tersection the planes have, when taken in pairs, are all paral-
lel ; this case will make itself known by exclusion of the others.

Example. Consider the four planes

= 0,
x-\- + 3z—2/ 6
4.x = 0,
+ 2y-bz-\- 8
8i/-7z-f 22 = 0,
2X-32/+ z- 7 = 0.
Here |
ABCD = and A4 0, as can
4=.\
easily be verified.
Hence the four planes meet in a single point. Let the student
show further that the last three planes pass through a line,
which is intersected by the first plane in the point in ques-
tion.
EXERCISES
Do the four given points lie in a plane ? If so, do three,
or do all four, lie on a line ?
1. (2, 3, 1), (1, 5, 2), (- 3, 4, - 1),(- 2, 2, - 2).
2. (2, 5, 3), (0, 2, - 3), (1, 3, 7), (- 1, - 1, 15).

3. (1, 2, - 1), (3, 1, 2), (- 1, 3, - 4), (7, - 1, 8).

4. (0,2,1), (1,0,2), (-1,-1,1), (4,2,3).

5. For what values of k will the four points {k, — 5, 6),

(4, _ 4, A:), (5, 1, 2), (2, 0, 7) lie in a plane ?

What can you say of the relative positions of the four given
planes ?
6. The planes of Ch. XVI, § 9, Ex. 9.

7. The planes of Ch. XVI, § 9, Ex. 10.

3a;-5yH-22 + 3=0, 2x + Zy- 2;-|-4 = 0,
2iB + 4iy-32-l = 0, + 32-2 = 0,
a;-2?/
5a; _ 2/- ^-f2 = 0, '
3.r+ + 22; + 3 = 0,
2/

8a;-62/4- + 5 = 0. 2; 5a; + 47/+ 2-5 = 0.

 :

512 ANALYTIC GEOMETRY

10. For what value of k will the four planes,

kx-\- y-z -6 = 0, x-y-\- z = 0,

xJ^Jcy-\-z-3 = 0, 2a;H-?/ + 42-l = 0,
go through a point?

5. Two Intersecting Lines. Given the two distinct lines

(
A,x + B^y + Ciz + A = 0, A,'x + B,'y + d'z + i)/ = 0,
+ A = 0;
I
1 A^x + ^22/ + C22 I A^'x + B^'y + Cs'z + D^' = 0.
The two lines intersect in a point, when and only when the

four planes which in pairs determine them meet in a single

point. The condition for this is given in Theorem 3 of the

preceding paragraph.
The two lines are parallel if and only if the normals to the
four planes are all parallel to a plane. Theorem 2 of § 4 tells

when this occurs.

These results can be combined in the general theorem
Theorem 1. Two lines intersect or are parallel ivhen and
only ichen the determinant of the coefficients in the equations of

the four planes lohich in pairs determine the two lines vanishes.

The simplest way to decide in any case whether the two

lines intersect or are parallel is to compute the direction com-
ponents of the lines and compare them.
The above proof assumes tacitly that the four planes in
question are distinct otherwise the theorems of § 4 could
;

not be applied. Theorem 1 still holds, however, in the ex-

ceptional case when one of the planes determining one line
is identical with one of the planes determining the other line.
For, the two lines lie, then, in a plane and hence intersect or

are parallel on the other hand, the determinant in question

;

contains two rows which are proportional and hence it vanishes.

Lines Given by Continued Equalities. Let the first line be

determined by the point Pi (x^, yi, Zi) and the direction com-
:

ponents li, mi, rii and the second by the point P2 (^2> 2/2? ^2) :

and the direction components I2, m2, n^.

 ADVANCED METHODS 513

The lines are parallel if and only if l^, mi, rii are proportional
to I2, W2, ^2.
If the lines are not parallel, there is a unique plane which
contains the first line and is parallel to or contains the second.
A normal to this plane is perpendicular to both lines and
hence has the direction components mi7i2 ?ii^2 1 h'^2 cf. | \, | \ \
j

Ch. XVIII, § 5. The equation of the plane is, then,

(1) I
mi?i2 |(a; - «i) + 1 711^2 1
(2/
- 2/i) + I
^i^^2 1
(^ - ^i) = 0,
or
x-xi y-yi z-Zi
(2) Zi mj rii = 0.
?2 'WI2 ^2

If this plane contains the point P2 it will contain the entire

second and conversely. Consequently, the two non-parallel
line,

lines intersect if and only if

^2 "~~ ^1 2/2 "~ 2/1 ^2 ~ ^i

(3) ^1 mi ni =0.
U m2 ^2

Equation of Plane Determined by Two Intersecting or

the

Parallel Lines. two given lines intersect or are parallel, the

If
plane in which they lie can be determined by one of the lines,
say the first, and a point of the second which does not lie on
the first. Its equation, then, can be found by the method of
§ 1. This method, though always applicable, is designed
primarily for the case when at least one of the lines is given
as the intersection of two planes.
If both lines are represented by continued equalities, the
plane which they deterriine has (2) as its equation, in case the
lines intersect. If the lines are parallel, a similar equation
for their plane can be found ; cf. § 1, Ex. 15.

EXERCISES
Show that the given lines intersect or are parallel. In each
case find the equation of the plane which they determine.
514 ANALYTIC GEOMETRY

1.
= 0,
2x-Sy-\-l {2x-y-z = 0,
3x-y-2z = \x-2y-i-l = 0.
0',

Ans. The plane is 5x — 4:y — 2z -^-l = 0.

2.
j^x + 2y + 5 = 0, |2x + 72/ + 2 + l = 0,
[ x — y— 2 = 0; 3a; + 32/-;2 + 4 = 0.
1

3.
y-3^y-2^z-l
2 -5
4 — Sx _y _ _
3 ' -4^2
x—1 = w 4- 2o = 2—6 ,

4.
2 ~^' ~2~
-|
Sx + l = 4:-y = 2z-S.

5. I
4a; -2/ + 3^ + 1 = 0,
l2a;+32/ + 52-3 =
6. Distance of a Point from a Line. Distance between Two
Lines. Let it be required to find the distance D of the point
P2 {x2, 2/25 ^2) froiii the line
: L which passes
through the point Pi (xi, yi, : Zi) and has the
'^L direction cosines cos a, cos /S, cos y.
It is clear from the figure that

D= PiP2sme,
Fig. 2 where 6 is the angle which the line P1P2 makes
with L.
By Ex. 16 at the end of Ch. XVIII,

(1) Sin2 = I
/i,iV2 P + I
V1A2 1^ + I
A1/X2 |2,

where Xi, fxi, vi and A2, /xo, V2 are the direction cosines of P1P2
and L.
Now,
Xi-
X<) ^1
, u. - y-' - y^ V — ^2 — gj .

A2 = cos a, fM2 = COS 13, V2 = COS y.

Hence
?/2 - .Vi ^2

/XlVo = P1P2 P1P2 ^2 - 2/1 ^2- ^\

I

P1P2 COS yS COS y

COS fi COS y

Similar values are found for |

viA.2 1 and |
A1/X2 1.
 :

ADVANCED METHODS 515

Substitriting these values in (1), multiplying the resulting

equation through by PyP^ and extracting the square root of
both sides, we obtain, as the final result

Z2 — ^i ^2 — ^^1 ^2 "~ ~
^=v + + -^'i 2/2 2/1

cos j8 cos y COS y COS a COS a COS /3

Distance between Two Skew Lines. Let the line through

the point Pi : (xi, 2/1, ^1) with the direction components ?i, mi, nj
and the line through the point
P2 (^2> 2/2> ^2) with the direction
•

components Z2, m2, W2 be two

skew lines, i.e. two lines which
neither intersect nor are paral- Fig. 3

lel. To find the distance D be-

tween them measured along their common perpendicular.
The plane through the first line parallel to the second has
the equation (1), § 5, namely,

I
^11^2 I
(x — i»i) + | ^1^2 1
(2/
— 2/1)+ I
hm2 1
(z - %)= 0.
The required distance D is the uniform distance of the sec-
ond line from this plane, or it is the distance of the point
2/2) ^2) from
P2 : (a^2) this plane. Thus,

D = ± m ifh I
(X2 — a^i) + I
n^h I (2/2 - 2/1) + I h'^2 1
(^2 - ^i)
V I ?7^ln2
1

2+ I
^1^2 1
^ + I
h^h \^

or

D=±
V I min2 [2 H- 1
ni?2 [^ + | ^1^2
12

where A is the determinant in formula (3), § 5, and where

that sign is to be chosen which will make the right-hand side
positive.

Distance between Tivo Parallel Lines. The distance between

two parallel lines can be found as the distance of a point
on one of the lines from the other line.
516 ANALYTIC GEOMETRY

EXERCISES
Find the distance of the given point from the given line.

Point Line
4 — iC_?/_l — 2!
Ans. 4^ VlOi = 4.31.
2, 3, 4), •

6 3
X
2. 0, 0, 0),
2-5
-5 4
3. - 1, 2, - 3), Sx-\-l = 4.-y = 2z -3.

\3x-\-2y-{-2z-{-2
4. 2, - 1, 5),
0,
{
\6x-h5y + 6z-h2 = 0.
l2x-3y-\-l = 0,
6. 3, 1, - 1),
\3x-y-2z = 0.
Find the distance between the two given lines.

6. The lines of Exs. 1, 2. 7. The lines of Exs. 2, 3.

Ans. i|V89 = 1.59. '

8. The lines of Exs. 1, 3. 9. The lines of Exs. 2, 4.

10. The lines of Exs. 3, 4. 11. The lines of Exs, 4, 5.

12. A cube has edges of length a. Find the distance be-

tween a diagonal and an edge skew to it. Ans. ^ V2 a.

7. Area of a Triangle. Volume of a Tetrahedron. We

j&rst prove the following theorem.
Theorem 1. If a region of area in a plane A
is projected M
on a second plane M', the area of the projected region equals
A cos 0, where is the acute angle between and M'.* M
In the case of a rectangle whose sides are respectively
parallel and perpendicular to the line of intersection of M
and M', the proof is immediate. For, when the rectangle is
projected on M', one dimension remains the same and the
other is multiplied by cos 0.

* The theorem is trivial if M is parallel or perpendicular to M' ;

we
exclude these cases.
 : :

ADVANCED METHODS 517

The area A of an arbitrary region in is the limit ap- M

proached by the sum B of the areas of rectangles, of the type
just described, which are inscribed
in the region

A = lim B,
If A' is the area of the projected
region and B' is the sum of the

areas of the projected rectangles,

evidently
A' = lim B\
J Fig. 4
Since the area of each projected
rectangle is cos 6 times the area of the original rectangle,
B! = B cos 0.
Hence A' = lim B cos = cos 6 lim B
or A' = A cos 0, q. e. d.

Let the areas of the projections of the given region on the

coordinate planes be denoted by A^,, A,^, A^^, and let the
normals to M
have the direction angles a, /3, y. By Th. 1,=*
^^^=|^cosa|, ^„=|^cos^|, ^,^=l^cosy|.
Hence
(1) A-^ ^ AJ -{ AJ + A^^^
Thus we have proved the theorem
Theorem 2. The sum of the squares of the areas of the pro-
jectionsof a region on the three coordinate planes equals the
square of the area of the region.

Area of a Triangle. It is now easy to write down a

formula for the area A of the triangle whose vertices are at
the points P, {x^, yi, 2i), P2 {^2, V-i, ^2),
: (%, Vs, ^s)-
-
^or, A
the areas of the projections of the triangle on the coordinate
planes are, by Ex. 18 at the end of Ch. XYI,

A. = ±il2/i^2l|, A,, = ±i\x,z,l\, A,^=^±i\x,y,l\,

* The absolute value signs are necessary since a, /3, 7 are not neces-
sarily acute angles,
 1

518 ANALYTIC GEOMETRY

where yiZ^ 1
|
for example, is the determinant whose three
1
,

columns are y^, y^, 2/3 ^i, ^2, % 1, 1, 1- Hence, by (1),

; ;

(2) ^ = iV I
?/i22 1 1' + I
^1^2 1 |2 + I
X{y., 1 |2.

Volume of a Tetrahedron. Let the above triangle be the

base of the tetrahedron and let the fourth vertex be at the
point Pq : (xq, y^, z^. The volume V
of the tetrahedron is known from
Solid Geometry to be equal to one
third the area A of the base times the
length D of the altitude :

(3) F= i AD.
The equation of the plane of the
YiQ, 5 base, the ^lane of Pj, P2, P3, is given
form in Ch. XIX, § 6.
in determinant
This equation, when the determinant is developed by the
minors of the first row, becomes

\y^Z2l\x- \x^Z2l\y-\- \xiy2l\z - \x^y2Z^\ = 0.

The distance of the point Pq : (xq, 2/0, ^0) from this plane, i.e.

the length D of the altitude of the tetrahedron, is, by Ch.

XIX, § 9,

/^\ 2) __ . 1
2/1^2 1 1 ^o~ I
^1^2 1 1 .Vq-H I
^iJh 1 1 ^0 ~ I
^1^/2^3 .

V I
yiZ2 1 1^
+ I
^1^2 1 1^ + I
a^i2/2 1
1^

Substituting in (3) the values of and as given by (2) A D

and (4), and at the same time writing the numerator in (4) in
determinant form, we obtain, as the value of V,

(5)

where that sign is to be taken which yields a positive result.

 ADVANCED METHODS 519

EXERCISES
Find the areas of the following triangles.
1. With vertices at (2, -1, 3), (4, 3, -2), (3, 0, -1).
Ans. Wl34 = 5.79.
2. With vertices at (0, 0, 0), (x^, y^, z^), {x^, y^, z^-

3. Cut from the plane 2a; — 32/-f4s — 12 = 0by the coor-

dinate planes.
4. With vertices at (a, 0, 0), (0, h, 0), (0, 0, c).

Find the volumes of the following tetrahedra.

5. That of Ch. XIX, § 6, Ex. 9.

6. That of Ch. XIX, § 10, Ex. 5.

7. Included between the plane 2x — 3?/ + 42= 12 and the

coordinate planes.

8. Included between the coordinate planes and the plane

with intercepts a, 5, c on the axes.

9. With vertices at (0, 0, 0), (a^i, 2/1, 2^1), (^^2, 2/2, ^2), {^z^ 2/3, ^s)-

EXERCISES ON CHAPTER XXI

1. Find a parametric representation of the line

X— y — z-\-l = 0, 2 a.' — 2/4-2 — 8 = 0.

2. What are the equations of the projecting planes of an
arbitrary line passing through the origin ?
3. Find the equation of the plane which contains the line of
Ex. 1 and is parallel to the line 2x — 3 = y— 3 = 2z — 1.

4. Find the equation of the plane which is perpendicular to

the plane 2x-\- 5y —3z —2 = and meets it in the line in
which it intersects the (a;, 2/)-plane.

Ans. 6a.' -+- 15?/ + 29z- 6 = 0.

5. Do Ex. 16 at the end of Ch. XX without finding the
coordinates of the point of intersection of the three given planes.
Suggestion. Find two planes through the given point, each
containing the line of intersection of a pair of given planes.
520 ANALYTIC GEOMETRY

6. Do Ex. 17 at the end of Ch. XX without finding the

coordinates of the point of intersection of the given line and the
given plane.
7. Find the equations of the line which contains the point
(2, 0, — 1)and intersects each of the lines
\2x-y-\-3z=0, f x-\-y + 2z-5 = 0,
\Sx + y-2z=2', \3x-\-Ay-z + l = 0.
8. Eind the equations of the line which intersects each of the
lines given in Ex. 7 and is parallel to the line 4: — 6x = y-\-9 = 2z.
9. A plane intersects the (x, ?/)-plane in the line whose equa-
tion in the (x, i/)-plane is 2 a; + 3?/ = 12. If the plane cuts from
the first octant a tetrahedron whose volume is 12, find its
equation. Ans. 2x -\-Sy -\- 4:Z — 12 = 0.
10. There are two planes which contain the line

x-\-2y-{-z-\-l=0, 2x-\-y-z-7 =
and make angles of 30° with the plane x — z + 2 = 0. Find
their equations. Ans. x — y — 2z—8 0, 2x -\-y — z — 1 = 0.
=
11. Find the equation? of the planes which contain the line
given in Ex. 10 and are V2 units distant from the point
(2, 2, - 3).
12. The planes through the edges of a trihedral angle perpen-
dicular to the opposite faces pass through a line. Prove this
theorem in the case that the faces lie in the planes

2x — y + z = 0, 4.x—y-\-3z = 0, 3a; — 2?/ — 2 = 0.

13. Prove the theorem of Ex. 12 in the general case.

The Equations \u -\- ixv = 0, uv = 0*

14. Theorem. If u = 0, v = are the equations of two sur-
faces, the equation X.u-\-fxv=0, Aft^O, represents in general \ a

*Cf. Ch. IX, §^3, 4.

t In particular, it may represent a curve or a point ; cf . footnotes, pp.
445, 167.
 ADVANCED METHODS 521

surface ichich contains the total intersection of the tiuo surfaces,

if they intersect, and has no other point in common ivith either of
them. If the given surfaces do not intersect, the equation repre-
sents in general * a surface not meeting either of them or it has no
locus. Prove this theorem.

15. Find the equation of the sphere which contains the

circle
a^ _l_ 2/^
_|- 2;2 — 4 = 0, z —5=
and passes through the point (3, 0, 2).

Prove that the curve of intersection of the cylinders

16.
= 4, 1/2 -f 2;2 = 5 lies on the surface x^ — z^ -\-l = and is
a;2 -|- 2/2

the total intersection of this surface with each cylinder.

17. Theorem. If u =
0, v = 0, zo = are the equations of
three planes ichich meet in a single 2wint, the equation
\u -^ fxv -\- vw = rejjresents a plane through this point. Con-
versely, every plane ichich contains this point is a linear combinor
tion of the three given planes. Prove this theorem.

18. Find the equation of the plane which is determined by

the points (0, 0, 0), (1, 2, 0) and the point of intersection of the
three planes

2a; + + 32 + l = 0,
2/
a^ + -4^ + 2 = 0, 2x-2y + 6z -^ = 0.
?/

Ans. 2x — y — 6z = 0.

19. Theorem. TJie equation iiv = represents those points

and only those points which lie on the surfaces u = and v = 0.
Prove this theorem.
20. What do the following equations represent?

(a) a;2— 2/2 = ; (c) ^2— xy — xz -^ yz = 0;

(b) a^ - 2/' = ; {d) xy - xz - 2 y -\-2z = 0.

* In particular, it may represent a curve or a point ; cf . footnotes, pp.

445, 168.
522 ANALYTIC GEOMETRY

Bisectors of the Angles between Two Planes *

21. What is the locus of the inequality

2x-y+2z-4:>0?
22. There are four regions lying between the planes
2x-y-\-2z-4. = 0, Sx + 4.y + z - S ^0.
Find the pairs of simultaneous inequalities representing these
regions, specifying the region which each pair represents.

23. Find the equations of the planes bisecting the angles be-
tween the two planes of Ex. 22.
Ans. 2x-{-7y — 5z-{-4: = 0, Ux -\- y + 7 z - 20 = 0.
24. The same for the following pairs of planes

x-y-\-z-2 = 0, 3x-6y + 2z-^ = 0,

W a, + 2/-2 + 3 = 0; ^^ 6x-\-2y-9z-5 = 0.
25. Find the equation of that bisector of the angle between
the two planes of Ex. 24 (b) which passes through the region be-
tween the two planes which contains the origin.
26. The planes which bisect the dihedral angles of any
trihedral angle meet in a line. Prove this theorem when the
faces lie in the planes

x-\-y-^z — l = 0, x — y + z—l=0, 2x + y — z + l = 0.
27. Prove the theorem of Ex. 26 in the general case.
Suggestion. Cf. Exs. 28, 29 at the end of Ch. XIII and
Ex. 28 at the end of Ch. XIX.

28. Show that by a proper choice of axes two arbitrarily

chosen skew lines, Li and L2, can have their equations written as
x = c, z = my ; x = — c, z = — my ; cm =5^ 0.

29. Prove that, if the line L^ of Ex. 28 is taken as the 2;-axis,

the X- and ?/-axes can be so chosen that L2 has the equations,
x= c, z = my, where c^^^O.
* Cf Ch.
. Xm, §§ 6, 7, 8.
 CHAPTER XXII
SPHERES, CYLINDERS, CONES. SURFACES OF
REVOLUTION

1. Equation of the Sphere. The equation of the sphere

whose center is at the origin and whose radius is p is, accord-
ing to Ch. XIX, § 1,

(1) x^-{-y^-^z'' = p^.

It can be shown in a similarmanner that, if the center is
at the point (a, ft, y) and the radius is p, the equation of the
sphere is

(2) (x - ay + (2/
- /^y -^{z- yY = p".
Thus the sphere whose center is at the point (2, — 3, 4)
and whose radius is 6 has the equation
{x - 2)2 +(2/ + 3)2 +(^ - 4)2 = 36,
or ic2 _^ + 2^ - 4 + 6?/ - 82 - 7 = 0.
2/2 a?

EXERCISES
Find the equations of the following spheres and reduce the
results to their simplest form.

1. Center at (3, 1, 2) ; radius, 5.

Ans. x'^-\-y'^-\-z^ — ^x-2y — ^z—ll = 0.
2. Center at (— 2, 3, — 6) ; radius, 7.

3. Center at (4, 0, 0) ; radius, 4.

4. Center at (0, — 5, 0) ; radius, 2.

5. Center at (0, — 4, 3) ; radius, 5.

6. Center at (|,
— ^, 0) ; radius, 1.
523
524 ANALYTIC GEOMETRY

7. Center at (i —f , |-) ; radius, |.

8. Center at (0, 0, a) ; radius, a.

9. Center at (a, 0, a) ; radius, aV2.

10. Center at (a, a, a) ; radius, aV3.

2. General Form of the Equation. The equation of a sphere

can always be written in the form

(1) ^
x^-\-y'^-}-z'^-{-Ax + By-\-Cz + D = 0,
as is seen by expanding equation (2), § 1.

Let us investigate whether, conversely, every equation of

the form (1) represents a sphere.
Consider, first, the particular equation

(2) x'^ + y^-\-z^-2x + 6y + 4:Z-35 = 0.

If we complete the square of the terms in x, and do the same
for the terms in y and in z, the equation becomes
«

(3) - iy+(y + 3)2 +(2 + 2)2 = 35 + 1 + 9 + 4 = 49.

{x

This equation is of the form (2), § 1, where a = l, /? = — 3,

y = — 2, p = T, and hence represents a sphere whose center
is at the point (1, — 3, — 2) and whose radius is 7.

If the constant term, — 35, in (2) is replaced by 14, the

right-hand member of (3) becomes — 14 + 1 + 9 + 4 = 0. In
this case, then, we have
(cc- 1)2 +(2/ + 3)2 +(2 + 2)2 = 0.
T^e point (1, —3,-2) has coordinates satisfying this equa-
tion. For the coordinates (x, y, z) of any other point at least

one of the parentheses is not zero and the left-hand side of the
equation is positive. Consequently, the equation represents
the single point (1, — 3, — 2) or, if we define a null sphere (a

sphere of zero radius) as a point, it represents a null sphere.

If the constant term, —35, in (2) is replaced by 15, the
equation becomes

{x-iy+iij + 3)2 +(2 + 2)2 = - 1.

 SPHERES AND OTHER SURFACES 525

Since the left-hand member of this equation can never be

negative, no matter what values are assigned to x, y, z, the
equation represents no point whatever in space.
These three examples indicate what to expect of the general
equation (1). On completing the squares for the pairs of terms
in X, y, and z, respectively, in (1), the equation takes on the
form (2), § 1, where
A r, B C
(4)

Hence, we have the following

Theorem.
^4
Equation
^ A

a
I

(1) represents sjjhere, a single point,

or no point ivhatever, according as the quantity

is positive, zero, or negative. In case it represents a sphere, the

coordinates of the center and the square of the radius are given
by formulas (4).

Consider, more generally, the equation

(5) a (a;2 -\-

y^ -\- z^) + hx -[- cy -{- dz -{- e = 0.
If a = 0, but h, c, and d are not all zero, the equation repre-
sents a plane.
» li a^O, the equation can be divided through by a, and it

then becomes
x'^ + y'^ + z'^-\--x-\--y+-z-^- = 0.
a a^ a a
This equation is of the fotm (1) and hence the foregoing con-
siderations apply to it.

EXERCISES
Determine what the following equations represent. Apply
in each case the method of completing the square. Do not
merely substitute numerical values in formulas (4).
1. a;2-f2/2-f 22_^4a;-6y-2;? + 5 = 0.
Ans. A sphere, radius 3, with center at (—2, 3, 1).
526 ANALYTIC GEOMETRY

2. x'^-\-y'^-\-z'^—6x-\-Sy-\-4.z-\-29 = 0.
Alls. (3, - 4, - 2).
The point
3. a;2 + + — 2ic + 42/ + 224-9 = 0.
2/^ 2;2 Ans. No point.
4. x'^-\-y'^ + Z" — 4.x-\-2y-\-4.z = 0.

6. a;2+ + 2^ + 6ic-82/ + 16 = 0.
2/'

6. a;2 4-2/2 + 22 + 3;2_4 = 0.

7. x'^+ y^ + z^ — 2ay — 2bz = 0.

8. a;24- ?/2
-f =2
2;2 aaj.

9. x^-\-y^-\-z^-4:X-6y + 13 = 0.
10. a;2-(-2/^ + ;22^9 = 0.

11. 3^2 -h 32/2 + 3^2 _2ic- 12?/ + 6:^ + 14 = 0.

12. 2a;2 + 2?/2 + 22!2-|-2a;-6?/-22-7 = 0.

13. 3a;2 + 3^/2 + 32:2 -4ic + 22/ + 4;2 + 6 = 0.

14. 5x'- + 5y'^ + 5z'^-5x + 6y-Sz-5 = 0.

3. Sphere through Four Points. Through four points which

do not lie in a plane there passes a single sphere. If the
points are {x^, y^, z^), {x2, y^, z^), {x^, y^, z^), (x^, y^, z^), the equa-
tion of the sphere, in determinant form, is

a;2 ^2/2 +22

 SPHERES AND OTHER SURFACES 527

ofany one of these points makes the first row of the deter-
minant identical with a later row, so that the determinant
vanishes. Therefore the equation actually represents a sphere
and this sphere is the one through the four given points.

EXERCISES
Find the equations of the spheres through the following
sets of four points.

1. (1, 0, 0), (0, 1, 0), (0, 0, 1), the origin.

Ans. a;2 + y2 _|_ 2;2 _ — — 2 = 0.
aj 2/

2. (1, 1, 1), (- 1, 1, 1), (1, - 1, 1), (1, 1, - 1).

3. (2, 3, 1), (5, - 1, 2), (4, 3, - 1), (2, 5, 3).

4. The vertices of the tetrahedron of Ch. XIX, § 6, Ex. 9.

5. The vertices of the tetrahedron formed by the coordinate

planes and the plane 2x —3y -\-4:Z — 12 = 0.
6. The vertices of the tetrahedron of Ch. XIX, § 10, Ex. 5.

7. When will five points, no four of which are coplanar, lie

on a sphere ?
Do the five given points lie on a sphere ?

8. (0, 0, 0), (- 1, 0, - 1), (3, 1, 0), (2, 4, - 4), (3, 3, - 4).

9. (0, 2, 3), (4, 1, 0), (- 4, 5, 0), (1, 5, - 1), (4, 2, - 5).

4. Tangent Plane to a Sphere. Let P be a point of a sphere

and let the radius to P be drawn. The plane through P per-
pendicular to the radius is the tangent plane to the sphere at P.
If the sphere is

(1) x'^ + y^ + z^ = p2
and P has the coordinates {x^, i/i, z^), the radius to P has the
direction components x^, ?/i, Zi.
The tangent plane at P is the
plane through (x^, i/i, Zi), whose normals have these direction
components. Consequently, its equation is

Xi(x - x{) + yi{y - ?/i) -\-Zi{z- z^)= 0,

or Xix + yiy + z^z = x^^ + 2/1^ + ^i^-

528 ANALYTIC GEOMETRY

Since tlie point (xi, y^, Zi) lies on the sphere (1),

0^1^ + 2/i' + ^i' = P',

and hence the equation of the tangent plane, in final form, is

(2) x^x + y^y + 2:12; = p^.

In a similar manner the equation of the tangent plane to
the sphere *

(3) (x-ay-\-{y-f3y-\-(z-yy = p2

at the point (xi, 1/1, Zi) of the sphere can be shown to be

(4) {x, - a){x - «) + (2/i

- /3){y - y8)4-(^i - y){z-y) = p'.
The use
of (4) to find the equation of the tangent plane to
a sphere whose equation is in the form

(5) x^-\-y^-{-z'^ + Ax + By-\-Cz + D =

involves the reduction of the equation of the sphere to the
form (3). Thus, if the sphere is

(6) x^-{-y^-\-z'^-2x-j-6y-\-4:Z-S5 = 0,
the equation must first be rewritten as
(X- 1)2 + (2/ + 3)2 + (2 H- 2)2 = 49.
The equation of the tangent plane at the point (3, — 6, 4), for
example, is then, according to (4),

(3 _ l)(aj _!) + (_ 6 + 3)(2/ + 3) + (4 + 2){z + 2)=49,

or

(7) 2x-3y-\-6z-iH = 0.
The coordinates of the center (a, (3, y) and the square of the
radius, p'^, of a sphere whose equation is in the form (5) are
given by formulas (4), § 2. If these values for a, (3, y, p^ are
substituted in (4) and the equation obtained is simplified, the
result is

(8) x,x-{-y,y-^z,z + ^{x-\-x,) + ^(y + y,)-\-^{z-\-z{) + D=0.

This is the equation of the tangent plane at the point
(xi, 2/1, Zi) to a sphere whose equation is in the form (5). By
 SPHERES AND OTHER SURFACES 529

means of it the equation of the tangent plane to (6) at the

point (3, — 6, 4) can be written down directly. have, We
namely,
I
Sx-6y-\-4:Z-{x + 3)4-3(2/ - 6)4-2(2 4- 4)- 35 = 0,
and this reduces to the equation (7) obtained by the indirect
method.

EXERCISES
Find the equation of the tangent plane to each of the follow-
ing spheres at the given point.
1. 3.2
^ 2/2= 9 at (2, - 2, - 1).
4- 2^

2. -f
ic2 2/2= 49 at (3, - 6, 2).
4- 22

3. (aj_ 1)2 -^(y- 2)2 + (^ 4- 3)2 = 81 at (2, 6, 5).

4. 4.(2/ + 5)2 + (2 - 4)2 = 9 at (1,
a^2 - 3, 2).
6. 4-
ic2 4- 2;2_2ic — 42/4-42 =
2/2 at the origin.

6. 3^4- + 22 -6a; 4- 42/ +102 -11 = Oat (1, 1,

2/2 1).

7. Find the volume of the tetrahedron cut from the first

octant by the tangent plane at (1, 2, 3) to the sphere

2a;2 4- 22/2 4- 222 4- 2a; - 32/ - 42 - 12 = 0.

8. The coordinates of one of the points of intersection of the
plane 2x — y — 2 = with the sphere of Ex. 1 are (2,2, 1).
Find the angle between the plane and the sphere.
Ans. 72° 39'.

9. Find the angle which the line x=y =z makes with the
sphere of Ex. 5. ^
Ans. 11° 6'.

5. The Circle. A plane intersects a sphere in a circle, is

tangent to it, or fails to meet it, according as the distance D

of the center of the sphere from the plane is less than, equal
to, or greater than, the radius p of the sphere.
In other words, the equations of a sphere and a plane, con-
sidered simultaneously, represent a circle, a point, or no point
whatever, according sls D< p, D = p, ot D > p.
530 ANALYTIC GEOMETRY

Consider, for example, the sphere and the plane

^^ x-2y-2z-\-7 = 0.

The center of the sphere is at the point (1, — 3, — 2) and its

radius is p=7 -, cf. § 2. The distance of the center from the

plane is

j^ l+(-2)(-3) + (-2)(-2)+T ^18^^^

Vl + 4 + 4
3

Consequently, the plane meets the sphere in a circle, and equa-

tions (1), considered simultaneously, are the equations of the
circle.

It is readily seen that p^ — D^, i.e. 7^ — 6^ = 13, is the square

of the radius of the circle. Hence the radius of the circle is

Vl3. The center of the circle is the point of intersection of

the plane and the line through the center of the sphere perpen-
dicular to the plane. Its coordinates are thus found to be
(-1,1,2).
Radical Plane of Two Spheres. Given the two spheres

.ox x'-\-y^ + z^ + 2x-2z-l = 0,

^ ^
a;2 -f 2/2 + 22 + 4^^ _ 102 _ 9 = 0.
_|_ a^

Subtract the eq'uation of the second from that of the first.

The resulting equation,

(3) X- 4?/ -f- 82 + 2 = 0,

represents a plane. This plane is known as the radical plane
of thetwo spheres. Since its equation is a linear combination
of the two spheres, we conclude, by Ex. 14 at the end of
Ch. XXI, the following :

If the spheres intersect, the radical plane is the plane of

their common circle ; if the spheres are tangent, it is their
tangent plane at the point of tangency and if the spheres ;

fail to meet,^ the radical plane intersects neither of them.

* Two concentric spheres have no radical plane ; this is the only ex-
ceptional case.
 SPUERES AND OTHER SURFACES 531
/

Conversely, the spheres intersect in a circle, are tangent, or

fail to meet, according as their radical plane intersects one of
them in a circle, is tangent to it, or fails to meet it. Thus
the question of the relationship between two spheres is re-

duced to that of the relationship between a plane and a sphere,

and this we have already discussed.
The center of the first of the spheres (2) is at the point
(— 1, 0, 1) and its radius is 3. The distance of the center
from the plane (3) is found to be 1. The radical planeand
the first sphere intersect, then, in a circle, and consequently
this is true of the two spheres.
Equations (2), considered simultaneously, are a pair of equa-
tions of the circle. A simpler pair consists in one of the
equations (2) and the equation (3).

EXERCISES
In each of the following exercises, determine what the given
equations represent. If they represent a circle, find its center
and radius if they represent a point, find its coordinates.
;

1. a;2 + _ 25 = 0, 2 = 4.
2/^ -I- ^2

2. aj2-f + 22_ 6aj — 42/==0, 2aj + + 22 — 1 =0.

?/2 2/

3. a;24-2/2+;22_4a._2^ + 62-2 = 0, 6x-\-2y -3z+5 = 0.

4. a;2-f-2/2 + 22_2a; = 0, 8a;-2/-3=0.

Find the radical plane of the spheres given in the following

exercises. If the spheres intersect in a circle, find its center
and radius. If the spheres are tangent, find the coordinates
of the point of contact.

5. a.2 _|. ^2 _|. ^2 ^ 13^ a;2 + 2/2 + 22 _|_ 3a; _ 4 = 0.

^'
I
a;2 + 2^ + 22/ - 52 - 5 = 0.
+ 2/2

7. a;2-j-2/2-f22_2 = 0, 2 2 2/^4- 2 224- 3 -4 + 2 -f 2 = 0.

a;2-|- a;
2/

Find the equation of the sphere determined by the given

circleand the given point cf. Ex. 15 at the end of Ch. XXI. ;
532 ANALYTIC GEOMETRY

8. The circle of Ex. 1 and the origin.

9. The circle of Ex. 2 and the point (1, 1, 1).

0. The circle of Ex. 5 and the point (1, — 2, 3).

ICylinders.Given a plane curve, not a straight line, and

igh each poijit of the curvedraw an indefinite straight
line perpendicular to the plane. The surface generated by
these lines is called a cylhider. The lines are its rulings, or
generators, and the given curve its directrix.

We shall consider here only cylinders whose rulings are

parallel to a coordinate axis.
If the rulings of a cylinder are parallel, for example, to the
axis of z, the equation of the cylinder does not contain z.

For, the directrix can be thought of as lying in the {x, y)-

plane, and its equation in this plane will represent in space

the cylinder, inasmuch as whose coordinates
the points
satisfy the equation are those points and only those points
which lie on the directrix, or directly above or below it, i.e.
which lie on the cylinder. But this equation does not con-
tain z, q. e. d.
Conversely, a curved surface represented by an equation in
which z does not appear is a cylinder whose rulings are
parallel to the axis of z. Eor, the equation defines in the
(x, y)-plane sl curve, and in space it represents those points

and only those points which lie on, or directly above or

below, this curve, i.e. which lie on the cylinder erected on
the curve.
We have, then, the following theorem : A curved surface is

a cylinder with rulings parallel to a coordinate axis when and

only when its equation does not contain the variable correspond-
ing to that axis.

Quadric Cylinders. A cylinder whose directrix is a conic is

known as a quadric cylinder. In particular, it is called elliptic

(or circular), hyperbolic, ox parabolic, according as the directrix
is an ellipse (or circle), a hyperbola, or a parabola.
 SPHERES AND OTHER SURFACES 533

Figure 1 shows the quadric cylinders of the three types,

whose equations are

(1)

The two hyperbolic cylinders represented by the second

known as
equation are conjugate hyperbolic cylinders, and the
planes

(2)
X^
-^62 = 0,
a?

shown also in Fig. 1, as their common dsymptotic planes.

The elliptic cylinder, or either hyperbolic cylinder, of Fig. 1
is symmetric in each point of the axis of z. That is, every

Fig. 1

quadric cylinder whose directrix is a central conic is symmetric

in each point of the line drawn through the center of the conic
parallel to the rulings. This line is called the axis of the
cylinder.

Sections of Quadric Cylinders. The curve in which a quadric

cylinder is met by a plane, M, which is not parallel to the rul-
ings, we shall call a (plane) section of the cylinder.

Theorem 1. A section of a quadric cylinder is a conic of the

same type as the directrix.

We give the proof in the case in which the directrix, Z), is a

 :

534 ANALYTIC GEOMETRY

central conic. Let Jif intersect the plane Koi Din the line L*
As coordinate axes in K take Ox parallel to L and Oy perpen-
dicular to L, as shown. The
equation of D, referred to these
axes, is of the form (Ch. XII,
§3):
(3) Ax'- + Bxy + (7/ + F'= 0.

Draw in M the rectangular

axes, O'x', O'y', whose projections
on Kare respectively Ox, Oy.
Let P' (x', y') be an arbitrary
:

Fig. 2
point of M
and let P {x, y) be :

its projection on K. Then

(4) X = x', y = y' cos 6,

where is the acute angle between M and K.

Since D is the curve in K into which the section, S, of the
cylinder by ilif projects, the equation of S is obtained from
equation (3) of D by substituting for x and y in (3) their values
as given by (4). Thus the equation of S is

(5) Ax'^ + Bx'y' cos (9 + Cy'- cos2.(9 + i^ = 0.

This equation represents a conic and, furthermore, a conic of
the same type as D, since the discriminant of the quadratic
terms
^2 cos2 e-4.A0 cos2 = {B'-4.AC) cos2
is of the same sign as the discriminant, B^ — 4: AC, of the
quadratic terms in (3), q. e. d.
It is clear that (5) is independent of the height 00' at which
M cuts the axis of the cylinder. In other words, the sections by
two parallel planes are congruent conies.

Suppose, now, that B^ — AAC^O and that to F' is given in

turn the values 1, — 1, 0. Then (3) represents in turn a hyper-

* If M is parallel to K, the section by M

is congruent to the directrix ;

in this case, then, no further proof is required.

 :

SPHERES AND OTHER SURFACES 535

bola, the conjugate hyperbola, and the common asymptotic

lines ; but this is true, also, of (5). We
have, then, the follow-
ing result.

Theorem 2. The sections of two conjugate hyperbolic cylinders

by a plane M are tivo conjugate hyperbolas ichose common asymp-

totes are the lines in ichich M cuts the common asymptotic planes

of the cylinders.

Returning to the general case, we assume that there is given

a second cylinder with vertical rulings, whose directrix, Z>, is
similar and similarly placed to Z), or, if Z) is a hyperbola, is
similar and similarly placed to D
or to the conjugate of D.
The equation of T), as a curve in K, can be written, according
to Ex. 40, p. 260, in the form

(6) Ax' + Bxy -\- Cy'--{- Dx + Ey + F=0.

The equation of the section S of the second cylinder by the
plane M is, then,

(7) xix'^ -}- Bx'y' cos + C?/'^ cos2 -\- Dx' + Ey' cos 6 -{ F = 0,
Since equations (5) and (7) fulfill the conditions of Ex. 40,
p. 260,it follows that S and /S are in the same relation as D
and 2>. We have thus proved, in the case in which D and D
are central conies, the following theorem.

Theorem 3. If the directrices of two cylinders (ivith parallel

rulings) are similar and similarly placed conies, or if, in the case

of hyperbolas, each is si7nilar and similarly placed either to the

other or to the conjugate of the other, then the sections of the

cylinders by the same plane or by two parallel planes stand in like
relationship.

The converses of Theorems 1, 2, 3 are true, as is readily seen.

The three theorems and their converses can be stated equally
well in terms of projections. Thus Theorem 1 and its con-
verse are equivalent to the theorem : A plane curve is a conic
of a certain type if and only if its projection on a plane not per-
pendicular to its plane is a conic of this type.
536 ANALYTIC GEOMETRY

EXERCISES
1. Do Exs. 9, 10, 11 of Ch. XIX, § 1.

2. Do Exs. 20, 21, 22 of Ch. XIX, § 1.

What does each of the following equations represent ?

3. 4a;2 + 2/2— + 42/ — 4 = 0.
8i);

4. 3a;2 + 6a; -2?/ + 1 = 0.

5. xy-\-2x~y — 6 = 0.
6. Prove Theorem 1 when D is sl parabola.

7. Prove Theorem 3 when i) is a parabola.

8. State Theorems 2 and 3 and their converses in terms of

projections.

9. Show that if a central conic iS in a plane jlf projects into

the central conic D in the plane K, then the center 0' of S
ppeJ]ects into the center of D.

'\Cones.
7. Let a plane curve, not a straight line, and a
point 0, not in the plane of the curve, be given. Draw an
indefinite straight line through and each point of the curve.
The surface formed by these lines is known as a cone. The
lines are its rulings, or generators, and the point is its vertex.

If the given curve is a circle and lies on the line L

through its center perpendicular to its plane, the cone can be

generated by the rotation about L of any ruling. Accordingly,
it is known as a cone of revolution or a circular cone. The line

L is its axis and the constant angle between L and a ruling

is the generating angle.

Problem. To find the equation of a cone of revolution whose

vertex is at the origin, whose axis is the axis of z, and whose
generating angle is <^.

P
Let : (aj, be any point of the cone other than 0. The
y, z)
ruling R on which P lies determines with the axis of 2 a
plane M which cuts the (x, 2/)-plane in a line L. Direct the
line L as shown in the figure and denote the projection of OP
 SPHERES AND OTHER SURFACES 537

on X, thus directed, by r. Then the directed line L and the

axis of z form in the plane M
a system of coordinate axes,
with respect to which the point P has
the coordinates (r, z).

The equation of the ruling i2, as a line

of Mj is
r = z tan <^.

Since P lies on R, its coordinates (?*, z)

satisfy this equation. But, clearly.

where the plus sign or the minus sign is

to be taken, according as P lies on the
Fig. 3
upper nappe of the cone (as shown) or
on the lower nappe. Consequently, the coordinates (a;, y, z)
of P satisfy the equation *

± Va;2 + 2/2 = 2: tan ^,

or
(1) a;2 + 2/2_;22tan2<^ = 0.
Conversely, every point whose coordinates satisfy this equa-
tion lies on the cone, for the steps can be retraced. Hence
this is the equation of the cone.

Equation (1) is homogeneous in x, y, z. This is characteristic

of the equation of a cone with its vertex at the origin. In fact,
we can state the theorem : A curved surface a cone ivith its
is

vertex at the origin, when and only when its equation is homogeneous
in X, y, z.
'

Before giving the proof we consider a particular homo-

geneous equation

(2) 4:x''-3y^ + 12z^ = 0.

If Xj, 2/i, z^ is a solution of this equation, i.e. if

(3) 4.x,'-3 7j,^-^12z,-^ = 0,

* The coordinates of the origin, originally ruled out, clearly satisfy the
equation.
538 ANALYTIC GEOMETRY

kxi, kyi, kzi, where k is any constant, is also a solution. For,

the equation

4.(kx,y - S(ky,y + 12(kz,y = or k\4.x,^ -3 t/i^ + 12 z,') =

is true, inasmuch as the parenthesis has, by (3), the value zero.
P (x^, 2/1, ^i) is an arbitrary point, not
It follows that, if :

the origin, on the surface represented by (2),* any point with

coordinates of the form {kx^, kyi, kzi) is also on the surface.
But the points (kx^, ky^, kz^), where k is an arbitrary constant,
are all the points of the line OP passing through the origin
and P (Ex. 6 at the end of Ch. XYIII). That is, the line OP
through the origin lies wholly on the surface. But P was an
arbitrary point on the surface, other than 0, and therefore the
surface is formed by lines through the origin, i.e. it is a cone
with its vertex at the origin.
The cone can be constructed by drawing the curve in which
it intersects a plane not passing through and by joining
with the points of this curve by
straight lines. If y = 1 is the
plane taken, the curve of intersec-
-y tion is the ellipse

Ax'' + 12z^ = 3,
Fig. 4 and thus the cone is as shown in
Fig. 4.
This example suggests the following proof for the theorem.
An equation in x, y, z is homogeneous if and only if, when
Xi, 2/1, Zi (not all zero) is a solution,
kx^, kyi, kz^, where k is an

arbitrary number, is also a solution. On the other hand, a

curved surface is a cone with its vertex at the origin if and
only if, when P
(x^, yi, z^) is a point of the surface other than
:

0, an arbitrary point, (kx^, ky^, kz^), of OP is also a point of

the
surface. Thus the algebraic condition that an equation be
homogeneous is equivalent to the geometrical condition that a

*The locus of (2) is actually a surface (and not a curve or a point),

since all pairs of values (x, z) lead to points on it.
 SPHERES AND OTHER SURFACES 539

curved surface be a cone with its vertex at the origin, and the
theorem is proved.
"^^^^f^iadvic Cones. A cone represented by a homogeneous equa-
tion of the second degree in x, y, z, i.e. by an equation of the

form,
(4) ax^ + by^ + cz^ + dxy + eyz + fzx = 0,
is called a quadric cone.
We state, without proof, that equation (4), if it represents
a cone,=* can always be transformed by a rotation of axes
(cf. Ch. XXIY, § 6) into the equation,

(5) Ax^ + %2 _ Cz"- = 0,

where A, B, and C are positive constants.
The quadric cone (5) is, in particular, a cone of revolution,
when and only when it can be written in the form (1), i.e.
when and only when A = B.
All quadric cones are of one general type. They cannot be
classified into three types, corresponding to those for quadric
cylinders. For ellipses, hyperbolas, and parabolas can all be
obtained as plane sections of any one of them.f

EXERCISES

Construct the cones represented by the following equations.

If the cone is circular, determine its axis and the generating
angle.

1. a:2+?/2-22^0. 4. = 0.
4a;2-|_2/2_422

2. ic2- 32/2 + 22^0. 5. 60^-32/2-2 22 = 0.

3. 4aj2-2/2-;z2^0. e. y'^-2xz = 0.
* Equation (4) represents in general a curved surface, and hence, by
the theorem, a cone. Under special conditions it may, however, repre-
sent two planes, a single plane, a line, or merely the origin. For example,
x2 _ 2/2 — represents two planes ;
(x — ?/)2= 0, a single plane ;
x2 + ?/2 = 0,
a line and x"^ -\- y^ -\- z^ = 0, only the origin. These cases are here excluded.
;

t This was proved geometrically for the case of a cone of revolution in

Ch. Vni, § 10. An analytical proof covering all cases will be given later,
Ch. xxm, §5.
540 ANALYTIC GEOMETRY

Find the equations of the following cones.

7. The cone of revolution whose vertex is at the origin,
whose axis is the axis of y, and whose generating angle
is 30°.

8. The cone of revolution whose vertex is at (0, 0, a),

whose axis is the axis of z, and whose generating angle is 45°.

Ans. x^ -\-]p- — {z — a)2 = 0.

9. The quadric cone which has its vertex at the origin and
intersects the plane= 1 in the ellipse whose center is on the
2;

whose transverse axis is parallel to the axis of y, and

2:-axis,

whose major and minor axes are 6 and 4.

10. The quadric cone which has its vertex at (0, 1, 0) and
intersects the (2, a;)-plane in the hyperbola 2 aj^ — 2;2= 4.
11. The cone of revolution which has the line bisecting the
angle between the positive y- and z-axes as axis and which
contains the y- and 2:-axes. Ans. x^ = 2 yz.
12. The cone of revolution which has the line x— y = z as
axis and passes through the coordinate axes.
Ans. xy + yz -\- zx = 0.

Surfaces of Revolution. Let a plane curve and a line L in

)lane of the curve be given. The surface generated by the
curve when the plane is rotated about L through 360° is known
as a surface of revolution. The line L is its axis.

It is clear that spheres are surfaces of revolution. So also are

circular cylinders and circular cones ; in the one case the gener-
ating curve is a line parallel to i, and in the other, it is a line
which intersects L.

Quadric Surfaces of Revolution. The surfaces obtained by

rotating the conies about their axes, together with the spheres
and the circular cylinders and cones, are known as quadric
surfaces of revolution.

Problem. To find the equation of the ellipsoid of revolution

generated when the ellipse in the (?/, 2;)-plane, whose equation
 SPHERES AND OTHER SURFACES 541

in that plane is
2 2
^+?.=i, a>5,
a"- IP-

is rotated about its conjugate axis, the axis of z.

M be the rotating plane in an arbitrary position, and

Let let

P : be an arbitrary point on
{x, y, z)

the in M.
ellipse Establish in M the
same system of axes as was up in set
the plane M of Eig. 3 in finding the
equation of a cone of revolution. The
equation of the ellipse in 3/, referred
to these axes, is

Fig. 5

Since P lies on the ellipse, its coordinates {r\ z) satisfy this

equation. Consequently, inasmuch as

= x^ +
r^ 2/ J

the coordinates (ic, ?/, z) of P satisfy the equation

+ y'+'-=1.
x'' z
(1) :^!
. .

a>b,
a^ a^ 6^

and this is the equation of the ellipsoid of revolution.

In a similar manner, the equation of the ellipsoid of revolu-

tion obtained by rotating the ellipse about its transverse axis,
the axis of y, is found to be

^ ^
JP a^ b^

The first of the two ellipsoids of revolution (Fig. 6) is often

called an oblate spheroid ; and the second (Fig. 7), a prolate
spheroid. Both approach as their limits the sphere, whose
center is at the origin and whose radius is a, when b is made
to approach a as its limit.

The hyperboloids of revolution generated when the hyperbolas

r2 /y2 „2 2*^

(3) O TO -^J
?/
-=-1,
a' 62 a2 62
542 ANALYTIC GEOMETRY

situated in the (?/, 2)-plane, are rotated about the axis of z, — the
conjugate axis of the Urst hyperbola, and the transverse axis
of the second, —
have respectively the equations :

(4)
a^ or h

(5)
o;- v^ _^ -1, or
a^ a- ¥

Fig. 6 Fig. 7

The two hyperboloids of revolution are shown in Figs. 8 and

9. The first is known as a hyperholoid of one sheet or an
imparted hyperholoid ; the second, as a hyperholoid of two sheets,
or a hipaHed hyperholoid.
Taken together, the hyperboloids (4) and (5) are known as

conjugate hyperboloids of revolution. They are generated by

Fig. 8 Fig. 9

the conjugate hyperbolas (3) revolving about the same axis, the
axis of z. The cone which results from the rotation about this
 SPHERES AND OTHER SURFACES 543

axis of the common asymptotes of the hyperbolas, namely, the

cone

(6)
a? a' b'
544 ANALYTIC GEOMETRY

4. a;2_^2/2-|-422 = 4. 8. Sx^- - z''= 12.

4.y'^ 4:

5. 4.x^+ 9y^-\-9z^=S6. 9. Sx^ - 5y^ 3z'':=lo.

-\-

6. 2a;2-|_32/2-f 2z2= 6. 10. + 1/2=82;.

a;2

7. x^ + y^ — 2z'^ = 2. 11. + = — 4aj.

2/^ 2!2

12. The parabola of the text is rotated about the axis

of y. Find the equation of the surface generated and con-
struct it.

13. The surface generated by the rotation of a circle of

radius a about a line L in the plane of the circle at the dis-
tance h>a from its center is called an anchor ring, or torus.
Find its equation, if L is the axis of z and the circle is in the

(y, 2;)-plane with its center on the axis of y.

Ans. (a/^TF - by 4- ^2 = «^.

EXERCISES ON CHAPTER XXII

1. Find the equation of the sphere having the line-segment
joining the two points (3, 2, — 1), (5, 4, 3) as a diameter.

2. Find the equation of the sphere which has its center at

the point (5, —2, 3) and is tangent to the plane 3 a; +2 y-j- 2=0.

3. A sphere has its center in the plane x -\- y -\-3z —2=

and passes through the three points (2, 3, 1), (2, — 1, 5),
(—2, — 3, 3). Find its equation.
Find the equation of the sphere which has its center on
4.

the line 4a;-|-8 = 32/ + 7 = 42 and passes through the points

(4, 3,- 1), (3, 2, 3).
5. A sphere is tangent to the plane x ~2y — 2z = 7 in the

point (3, — 1, — 1) and goes through the point (1, 1, — 3).

Find its equation. Ans. y^ — lOy — lOz — 31 = 0.
x"^ -\- -\- z"^

6. There are two spheres passing through the points

and having the radius 3. Find their
(4, 0, 3), (5, 4, 0), (5, 1, 3)
equations.

j
a;2+ y^+ 2^ - 6x-4:y-2z-\- 5 = 0.
•

I
Sx^ + 3/ -h 3^2 _ 22a; - 82/ - 22 + 19 = 0.
 .

SPHERES AND OTHER SURFACES 545

7. rind the equations of the spheres which are tangent to

the plane x -^2y — 2z — 12 = and pass through the three .

points (3, - 2, 0), (2, - 3, 0), (3, 1,-3).

8. Find the equations of the spheres which are tangent
to the planes 2x — y-\-2z + l = 0, 6x-\-oi/ — 2z + o = and
have their centers on the line 1 —x = ij -\-l = 2z.
9. Find the equation of the sphere inscribed in the tetra-
hedron formed by the plane 2x-{-2y-[-z — 4^ = and the
coordinate planes.
10. A sphere goes through the point (4, 6, 3) and meets the
(x, 2/)-plane in a circle whose center is at the point (1, 2, O)
and whose radius is 5. Find its equation.
Ans. 0^24. 2/2 + ^2- 2a; -4?/ -32; -20 = 0.
11. Find the equations of the tangent line to the circle

x2-\.y2-^z^-x+ 4:Z = 0, 3a; -2?/ + 42; + 1 =

at the point (1, - 2, - 2).
12. Does the line 2x — 1 = 2/4-3=4 — intersect the sphere 2;

x2^y'2^z^--Qx-{-Sy -4.z-\--i = 0?

13. Show that the radical planes of three spheres, taken in

pairs,* pass through a line or are parallel.
I

Orthogonality f

14. Prove that the plane x + y -\-z — l = intersects the

sphere of Ex. 12 orthogonally.
15. When does the plane ax + by + cz-{-d = intersect the
sphere
x'^ -\-
y^ -\-
z"^
-\- Ax -{- By -j- Cz -}- D=
orthogonally ?

Ans. When and only when a A -^ bB -\- cC = 2 d.

* It is assumed that no two of the three spheres are concentric ;

cf

footnote, p. 530.
t Cf. Exs. 18-26 at the end of Ch. IV.
546 ANALYTIC GEOMETRY

16. Show that the line 3x + S = — 6y — 7 = 2z -\-13 inter-

sects the sphere of Ex. 12 orthogonally.

17. Find the condition that the line

a^x -\- hiu -\- CiZ -\- di = 0, a^x -f- 622/ -f C22; -|- ^2 =
intersect the sphere of Ex. 15 orthogonally.

18. Eind the angle of intersection of the spheres (2) of § 5.

19. Prove that the two spheres

a;2 + 2/2 _^ ;22 _ 9 = 0^ a;2 + 2/2 + ^2 _ 6a; -f 83/ + 9 =
intersect orthogonally.

20. When does the sphere of Ex. 15 intersect the sphere

a;2 + 2/- + ^^ + ^'^ + B'y + C'2 + i)' =

orthogonally ?
Arts. When and only when AA' + BB' -\- CC = 2 D-\-2D'.
21. Find the equation of the sphere containing the circle of
Ex. 11 and intersecting orthogonally the first of the spheres
of Ex. 19.

Loci
22. Do Ex. 1 of Ch. V, § 1, when P is not restricted to lie
in a plane.
23. The same for Ex. 4 of Ch. V, § 1.

24. A point P moves so that the ratio of its distances from

two fixed points is constant. Find its locus.

25. A point P moves so that its distance from a given plane

M is proportional to the square of distance to a given point
its

Pq, not in M. If P remains always on the same side of as M

pQ, find its locus.
26. If, in the preceding exercise, Pq lies in J[f and P may be
on either side of M, what is the locus of P?
27. What is the locus of a point which moves so that its

distance from a given line is proportional to its distance from

a given plane perpendicular to the line ?
 SPHERES AND OTHER SURFACES 547

28. What is the locus of a point which moves so that its

distance from a given point is proportional to its distance to a

given plane through the point ?

29. What is the locus of a point P which moves so that the
difference of the squares of its distances from a given point
and a given sphere is constant, if the distance from P to the
sphere is measured along a tangent line to the sphere
through P?
30. Agiven point Pq is distant 2 a units from a given plane
and P' isan arbitrary point in the plane. What is the locus
of the point P so chosen on the line PqP' that PqP PqP' = 4a2? •

Suggestion. Take the coordinates of P' as auxiliary

variables.

31. J. is a fixed point and B

an arbitrary point of a given
sphere whose center is 0. The radius OA is produced four
times its length to the point A^ and the radius OR, twice its
length to the point R'. Wliat is the locus of the point of
intersection of AR and A! R ?
 CHAPTER XXIII

QUADRIC SURFACES

The Ellipsoid. A quadric surface is any surface defined

equation of the second degree in x, y, z. The sphere and
the quadric cylinders, cones and surfaces of revolution studied
in the previous chapter are special types of quadric surfaces.
We proceed to consider more general types.
The surface defined by the equation

(1) +^=1
is known as an ellipsoid. If two of the three numbers a, b, c

are equal, it is in particular an ellip-

soid of revolution (Ch. XXII, § 8).

To construct the surface in the general

case when no two of the three numbers
a, b, c are equal, plot first the sections
by the coordinate planes, that is, the
curves of intersection with the planes
Fia. 1 a; = 0, 2/ = 0, 2 = 0. These are, respec-
tively, the ellipses

"^ + '-=1, ^+y-=i.

62 c2 ' a2 a^

The parts of these ellipses which lie in the quarter-planes

bounding the first octant connect the points (a, 0, 0), (0, b, 0),
(0, 0, c), as shown.
The section of (1) by a plane z = k parallel to the {x, y)-
plane has the equations :

548
 ;

QUADRIC SURFACES 549

If Ti^ < c2, these equations represent, in the plane z = k, an

ellipse whose center is on the axis of z and whose axes lie in
the {z, x)- and (y, 2) -planes and have the lengths

2a \/i-!' Wi-J
As k increases from toward c as its limit, this ellipse, rising
from the section by the (x, 2/)-plane, grows continuously
smaller and shrinks finally to a point, the point (0, 0, c). —
Similarly, if k decreases from toward — c as its limit.
The surface generated by the
changing ellipse is the ellipsoid.
Fig. 2 (or Fig. 6 of Ch. XXII *)
shows it in its entirety. The sur-
face is evidently symmetric in the
origin, 0, and in the coordinate
axes and coordinate planes. O is

called the center of the ellipsoid Fig. 2

the coordinate axes, the axes of the

ellipsoid ; and the coordinate planes, the principal planes of the
ellipsoid. The sections by the principal planes are known as
the principal sections.
The dimensions
of the ellipsoid, measured along the axes,
are 2a, 2b, These numbers, in the order of their magni-
'2c.

tude, are known as the major axis, mean axis, and minor axis
of the ellipsoid.

Here, and throughout the chapter, we speak of a (plane)

section of a surface only when the plane in question meets the
surface in a curved line. Sections by parallel planes we shall
call parallel sections.

* Figs. 6-10 of Ch. XXn, drawn originally to represent quadric sur-

faces of revolution, picture equally well the corresponding general quad-
ric surfaces studied in this chapter. One has merely to imagine that a
different ratio of foreshortening along the axis of x has been chosen.
550 ANALYTIC GEOMETRY

EXERCISES
Construct the following ellipsoids, drawing accurately the
principal sections and the sections parallel to one principal
plane. What are the lengths of the axes ?

1. -+r
a;^
_^?!^i. 2. 9 a;2 + 36 2/2 + 4 22 = 36.
16 25 9

3. Discuss the generation of the ellipsoid (1) by sections

parallel to the (2, ic)-plane.

4. Prove that the sections of the ellipsoid (1), which are

parallel to a principal plane, e.g. the [x, 2/)-plane, are similar
and similarly placed ellipses ;
cf. p. 260.

2^ The Hyperboloids. Tlie Hyperholoid of One Sheet. The

surface represented by the equation

(1) 52 ^2
a2

is called a hyperholoid of one sheet or an unparted hyperholoid.

If a = b, it is in particular a hyper-
holoid of revolution of one sheet
(Ch. XXII, § 8).

In the general case, a =^h, the

surface can be constructed by the
method of § 1. The sections by
the vertical coordinate planes,
X = and y = 0, are the hyper-
bolas
 ;

QUADRIC SURFACES 551

The Hyperholoid of Two Sheets. This surface, also known

as the biparted hyperholoid, is defined by the equation

(3)
t-t=-l or
x^ ip-

62
j^ = 1.
a2 c2

A particular case, when a = 6, is the hyperholoid of revolution

of two sheets.
In the generalcase, a ^^ 5, the sections by the vertical coor-
dinate planes are the hyperbolas conjugate to the hyperbolas
(2). The (ic, 2/)-plane, 2 0, does not =
intersect the surface. This is true
of all the planes 2; = for which
A:,

fc2 < c^ The planes 2;= ± c meet the

surface in the points (0, 0, ± c), and

the planes z = where ^'2 > c^ meet

^',

it in ellipses, which increase in ^size

as A: increases in numerical value
cf. Fig. 9, Ch. XXII.
Center, Axes, Principal Planes.
Each hyperboloid is symmetric in Fig. 4
the origin and in the coordinate
axes and coordinate planes ; is the center, the coordinate
axes, the axes, and the coordinate planes, the principal planes
for each surface. The sections by the principal planes are the
'principal sections.

The Asymptotic Cone. The hyperboloids

(1) and (3) are
called conjugate hyperboloids. We
have seen that each vertical
coordinate plane intersects them in conjugate hyperbolas whose
common asymptotes pass through the origin. This is true also
of any vertical plane,

(4) y = mx,
which passes through the axis of z. For, the sections of (1)
and (3) by the plane (4) are also the sections by this plane of
the cylinders.
1 ?7A 2^ = -r,
ic- + '-^\--
.

.a^
552 ANALYTIC GEOMETRY

whose equations are obtained by eliminating y from (4) and

(1), and from (4) and (3). But these cylinders are conjugate
hyperbolic cylinders whose common asymptotic planes are

m
(5) X' 4 + 1^-^=0.
,a' 62 C2

Consequently, by Ch. XXII, § 6, Th. 2, their sections by the

plane (4) are conjugate hyperbolas whose common asymptotes,
defined by equations (4) and (5), pass through
the origin, q. e. d.

The equation of the locus of these asym-

ptotes, as the plane (4) rotates about the axis
of z, is obtained by eliminating m (now an
auxiliary variable expressing the motion of
the plane) from equations (4) and (5). The
result is

(6)
t + t-t = 0.
a^

The locus of the asymptotes is, therefore, a

Fig. 5
cone whose vertex is at the origin. This
cone is called the asymptotic cone of each of the hyperboloids
(1) and (3). Evidently (1) lies wholly without the cone, that
is, on the convex side of it, while (3) lies wholly within it.

EXERCISES
Construct the following hyperboloids, drawing accurately
the principal sections which exist and the sections parallel to
one principal plane.
x^ y^ ^'--1

3. 9a;2_16?/2+ 36^2 = 144. 4. -4.x'' -\-36y'' - 9z^ = 36.

5. What is the equation of the hyperboloid conjugate to the
hyperboloid of Ex. 3 ? of Ex. 4 ? Give also in each case the
equation of the common asymptotic cone.
 QUADRIC SURFACES 553

6. Show that the sections of the hyperboloid of two sheets

(3), which are parallel to a vertical principal plane, are similar
and similarly placed hyperbolas.
7. The sections of the hyperboloid of one sheet (1) by two
planes parallel to a vertical principal plane are similar and
similarly placed hyperbolas, or are hyperbolas each of which
is similar and similarly placed to the conjugate of the other.
Prove this theorem and determine when each of the two cases
occurs.

/A M 3. /'The Paraboloids. TJie Elliptic Paraboloid. The surface

Vaefined by the equation

(1)
^'
+ ^'=2.
a" Z>2

is called an elliptic paraboloid, li a — b, it is in particular a

paraboloid of revolution (Ch. XXII, § 8).

In the general case, a^b, the

sections of the surface by the verti-

cal coordinate planes, y =^ and

X = 0, are the parabolas

aj2 = 2 a% 2/' = 2 b%
both of which open upwards. The
(x, ?/)-plane intersects the surface
only in the origin. A plane parallel
to the {x, 2/)-plane and below it does
not meet the surface, while a plane
parallel to the (x, 2/)-plane and Fig. 6
above it intersects the surface in
an ellipse, which increases in size as the height of the plane
increases ; cf Fig. 10, Ch.
. XXII.

The Hyperbolic Paraboloid. This is the surface defined by

the equation

(2)
^-y^ = 2z.
554 ANALYTIC GEOMETRY

It is never a surface of revolution, no matter what values are

assigned to a and b.

The sections of the surface by the vertical coordinate planes

are the parabolas
x''' = 2u% »
y^ = -2b%
of which the first opens upwards and the second, downwards.
The section by the (x, 2/)-plane
consists of the two lines,
X
OA:
a
+ 1 = 0,
X
OB:
a b

A section parallel to and above

the (x, 2/)-plane is a hyperbola
whose vertices are on the parabola
Fig. 7
opening upwards, whereas a sec-
tion parallel and below the (x, 2/)-plane is a hyperbola
to
whose vertices are on the parabola opening downwards. It is
seen, then, that the surface is saddle-shaped; it rises along the
parabola which opens upwards, and falls along the parabola
which opens downwards. The (z, ic)-plane contains the pommel
and the (y, 2;)-plane, the
stirrups.
The surface can best
be plotted by drawing
the sections parallel to a
vertical coordinate plane,
for example, the (y, z)-
plane. These sections
are all parabolas opening
Fig. 8
downwards and having
their vertices on the parabola in the {z, a;) -plane. Figure 8
shows part of the surface constructed by means of them.
Vertex, Axis, Principal Planes. Each paraboloid is sym-
metric in only one line, the axis of and in only two planes,
z,
 QUADRIC SURFACES 555

the vertical coordinate planes. The line is known as the axis,

and the planes as the principal planes. The sections by the
principal planes are called the principal sections, and the point
0, the vertex.
EXERCISES
Construct accurately the following paraboloids.

9 4 9 4
3. 2ic2 + 322 = 12?/. 4. x^-4.i/ = -Sz.
5. Prove that the sections of a pa.raboloid of either type,

which are parallel to a principal plane, are equal and similarly

placed parabolas.
6. Prove that the elliptic paraboloid (1) can be generated
by the parabola x'^=:2 a^z moving so that its vertex traces the
parabola y^ = 2 fz in ic = 0, while its axis remains vertical and
its plane parallel to the (z, a;)-plane.

Describe and prove a method of generating the hyper-

7.

bolic paraboloid, which is similar to that given in Ex. 6 for

the elliptic paraboloid.
8. Show that the equation xy = a'^z represents a hyperbolic
paraboloid.

4. Rulings. The Hyperboloid of One Sheet. The equation

^^ a- h' e-

can be written in the form

Consider the equations

h c \ a) \b cj a

obtained from (2) by setting the first factor on the left equal
to the parameter u times the first factor on the right, and
556 ANALYTIC GEOMETRY

then tlie second factor on the right equal to u times the sec-

ond factor on the left.

These equations represent a one-parameter family of lines,
each line being given by a particular value of the parameter u.
All the lines lie on the surface (1). For, if P is an arbitrary
point of the line u = Uq, the coordinates (x, y, z) of P satisfy
equations Ui for u Uq =
hence they also satisfy equation (2),
;

since, ii u =
Uq is eliminated from equations Ui by multiplying
them together, side for side, the result is precisely equation (2).
There will be just one line Ui through an arbitrary point
(xq, yQ, Zq) of the surface (1) if the equations

(3)^ ys+?9=Ji+^\ u(y^-'-A=i-^

b c \ aj \h cj a

have a unique simultaneous solution for u. Let us see when

this is the case.
If 1 + x^/a ^ 0, the first equation determines u uniquely,
and the value obtained is seen to satisfy the second, since,
when it is substituted in the second, this equation takes on
the form (2) for (ic, y, z) (xq, yo, Zq). =
In this case, then,
there is just one line Ui through (xq, yQ, Zq).
If 1 -f Xq/u 0, but yo/b — Zq/c =^ 0, it follows from (2) that
=
y^/l) _j_ Zq/c = 0. Then the first equation of (3) is satisfied, no
matter what value u has. The second equation determines
u uniquely, and so in this case, too, there is just one line Ui
through {xq, 2/o,
Zq).

Finally, if 1 + Xq/u — and ?/o/6

— Zq/c = 0, at least one of
the equations (3) is contradictory and there no line JJi is

through (xq, 2/o, Zq). It is, however, natural to supplement the

lines Ui by the line

be a

for, ifwe divide each of the equations Ui by u and then allow

u to become infinite, the line Ui approaches Uq as its limit.
We have proved, then, that the lines U consisting of the
family of lines Ui and the line Uq fill out the surface just
 QUADRIC SURFACES 557

once. They form what is called a set of rectilinear generators

or rulings of the surface.
There is a second set of rulings, V, consisting of the family
of lines
2-i=y
y 2;_
1 + X- , <^'>'-''
h c \ a a
and the line

Fo: 1 + - = 0.
b c a
It is readily seen that this set has the same properties as the
set U. Hence we have the theorem :

Theorem 1. A liyperholoid of one sheet contains two sets of

rulings. Through each point of the surface j^asses one ruling of
each set.

It is conceivable that the ruling U and the ruling V which

go through the same point coincide. This is not the case,
however, as will appear later * cf. Theorem 5. ;

The lines through the origin parallel to the lines U have the
equations
V z X
4--=^-,
,

(4) I
be
f a
b c
x = 0.
Jb cj a

Their. locus, obtained by eliminating u, is the asymptotic cone

^ I y z'
= 0.
Moreover, the lines fill out the cone just once, as can be shown
by the method used in proving Theorem 1.
Similarly, the lines through the origin parallel to the lines
F, i.e. the lines
y
be =
^ — z
V ~,
X
a
C
(5)
fy ,
z\ X
\b c a
* It is not difficult to give a direct proof of the fact at this point.
 : j

558 ANALYTIC GEOMETRY

alsofill out the asymptotic cone just once. Consequently, we

have proved the theorem
Theorem 2. Tlie lines luhich pass through the center of a /
hyperholoid of one sheet and are parallel to the rulings U (or V)
are precisely the elements of the asymptotic cone. In other words
there is one and only one ruling of each set which is parallel to a
given element of the cone, and conversely.
From this theorem we can draw the following conclusions.
Theorem 3. No three rulings of one set are jxirallel to a plane.
For otherwise there would be three elements of the cone
lying in a plane, and this is impossible.

Theorem 4. Two rulings of one set neither intersect nor are

parallel; that is, they are never coplanar.

For, they are not parallel, since no two elements of the cone
are parallel and they do not intersect, since otherwise there
;

would be a point on the surface, through which pass two rul-

ings of the same set.
Theorem 5. Two rulings of different sets either intersect or
are parallel; that is, they are alioays cop)lanar.

For, first, the rulings of the two sets are parallel in pairs,
since there is just one ruling of each set which is parallel to
a given element of the cone. From equations (4) and (5) it
appears that ii = Wo(^ 0) ^"^^ ^ = '*''o(=?^ ^) determine a pair of
parallel rulings if and only if 1 + UqVq = 0, and that the ruling
w = is parallel to Vq, and the ruling ?; = to Uq.
Secondly, two non-parallel rulings of different sets intersect
in just one point. For, it is easily shown that the four equa-
tions Ui and Vi* which define in pairs two rulings which are
not parallel, i.e. for which 1 + wv =^ 0, have one and just one
simuitaneous solution for x, y, z, namely,

/c\
x = ^^ 1 — UV = b ^^
n U-{-V
= c-U — V -I ,
ci
1-1-1^^=5^0. ^
(6) , y , z ,

1 + UV 1 -\- UV 1 -\- UV
* The proof in the special cases, in which Uq or Vq or both are in-
volved, is left to the student.
 : :

QUADRIC SURFACES 559

If u and V take on all possible pairs of values for which

1 -f uv ^ 0,
equations (6) give the coordinates of the points of
intersection of all the lines Ui with the lines F], that is, the
coordinates of all the points of the surface (1) except those
on the lines Uq and Vq. They constitute, then, a parametric
representation of the surface (1) in terms of the two parameters
u and V.

The HyiJerholic Paraboloid. The equation

(7)^
^
^-t=2z
a^ b^

can be written as

<«)
(M)(M)=^^-
Accordingly, there are two sets of rulings on the hyperbolic
paraboloid, namely

a b \a b)

a b \a bj

The first equation of U represents a plane parallel to or

coincident with the plane AOz (Fig. 8)

a b

Consequently, the rulings U lie one each in the^ planes parallel

to (and including) the plane AOz. Moreover, they are the
total intersection of these planes with the surface ; for, if the
first equation of U is solved with equation (8) of the surface,
the result is precisely the second equation of U. Similarly,
the rulings Flie one each in the planes parallel to (and in-
cluding) the plane BOz:

a
and are the total intersection of these planes with the surface.
The planes AOz and BOz are known as the directrix planes.
560 ANALYTIC GEOMETRY

There is Just one ruling of each set through each point of the
surface, for the planes parallel to (and including) a directrix
plane exhaust all points of space just once, and hence their
lines of intersection with the surface exhaust all points of the
surface just once.
It is easily shown that the
direction components of a
ruling U are a, —
and that those of a ruling V are a,
6, u,

6, V. Since the two triples are never proportional, two rulings

of different sets are never parallel or coincident. In particu-
lar, the two rulings which pass through one and the same point

of the surface are distinct.

The following theorems are now easily proved.

'
Theorem 6. Three rulings of one set are ahvays parallel to

a plane.
For, all the rulings of a set are parallel to a directrix plane.

Theorem 7. Two rulings of one set are never coplanar.

For, they do not meet since they lie in parallel planes, and
they are not parallel, as inspection of the direction components
just found shows.

Theorem 8. Tu)o rulings of different sets always intersect.

For, their projections on the {x, ?/)-plane, being lines in the

direction of OA and OB respectively, intersect in a point M.
Now there is but one point, P, on the surface which projects
into M, since a line perpendicular to the {x, ?/)-plane meets the
surface just once. Consequently, the two rulings in question
intersect at this point P.
The coordinates of P, found by solving the four equations
U and V simultaneously for x, y, z, are

(9) x=^a-^, y = h-^-, z= — '

These equations constitute a parametric rej^resentation of the

surface (7) in terms of the parameters u and v there are no ;

exceptional points.
 ;

QUADRIC SURFACES 561

EXERCISES
1. Find the equations of the rulings which pass through the
point (3, 2, 5) of the hyperboloid of one sheet of Ex. 1, § 2.

2. The same for the hyperbolic paraboloid of Ex. 2, § 3, the

point on the surface being (9, 2, 4).

Exercises 3-5. Use considerations of symmetry in the proofs.

3. The two rulings through a point P on a principal section

of a hyperboloid of one sheet are equally inclined to the plane
of the section and lie in a plane M which is perpendicular to
the plane of the section.
4. The same for a hyperbolic paraboloid.

5. If P
and P' are points of a hyperboloid of one sheet
which are symmetric in the center, the rulings through P are
parallel to those through P'.

6. Assuming Th. 1, § 7, prove that the plane of Ex. 3 M

passes through the tangent line at P
to the principal section.

Hence show that the projections of the rulings of either set

on a principal plane are the tangents to the principal section
in that plane.

7. The same for a hyperbolic paraboloid, applying the re-

sults of Ex. 4.

8.Prove that the plane determined by two parallel rulings

of a hyperboloid of one sheet is tangent to the asymptotic cone
along the element which is parallel to the two rulings.
9.. Prove that there are no straight lines on (a) an ellipsoid
(b) a hyperboloid of two sheets (c) an elliptic paraboloid.
;

5. ParaUel Sections. Equations (1), (3), and (6), § (2), of a

hyperboloid of one sheet, Hi, of the conjugate hyperboloid of
two sheets, Hz, and of the common asymptotic cone, C, can be
written as the one equation

^ ^
a? (y- c^
562 ANALYTIC GEOMETRY

where X is given the values 1, — 1, and in turn. The three

\

surfaces can thus be considered simultaneously.

We propose to determine the sections of the surfaces (1) by
an arbitrary plane,

(2) Ax-\-By-\-Cz = k.
At least one of the coefficients A, B, C is not zero. Assume
that C=^0. Then the sections in question are also the curves
in which the plane (2) meets the cylinders whose equations
result from the elimination of z from equations (1) and (2),
namely, the cylinders

a;"- y- (k- Ax - Byf ^

(S) ^

Considering the sections from this point of view, we conclude

the following theorems.

Theorem 1. The section of a hyperboloid or a cone is a conic.

For, the cylinders (3) are quadric cylinders, and a section

of a quadric cylinder is a conic ; cf Ch. XXII, § 6, Th. 1.
.

Theorem 2. Tiuo parallel sections of a hyperboloid or a cone

are conies of the same type. TJiey are, moreover, similar and
similarly placed, or, in the case of tico hyperbolas, each is similar
and similarly placed, either to the other or to the conjugate of the
other.
To prove this theorem, we fix our attention on one of the
surfaces (1), say the hyperboloid Hi, and give to k two arbi-
trarily chosen values, ki and ^"2, thus obtaining two arbitrary
parallel sections of Hi. The coefficients of the quadratic
terms in the two equations (3) which result are respectively
equal, since these coefficients in the general equation (3) do
not contain k. Hence, by p. 260, Ex. 40, the directrices of the
cylinders defined by the two equations are similar and simi-
larly placed conies, or, in the case of two hyperbolas, each is
similar or similarly placed either to the other or to the conju-
gate of the other. Consequently, by Ch. XXII, § 6, Th. 2,
 QUADRIC SURFACES 563

this is true also of the sections of the cylinders by the two

planes, q. e. d.

Theorem 3. The sections of two conjugate hyperboloids and

the common asymptotic cone by the same plane or by parallel
planes are similar and similarly placed conies, or, in the case of
hyperbolas, one of any two is similar or similarly placed either
to the other or to the conjugate of the other.

It is sufficient to prove the theorem for the sections of the

three surfaces by a single plane, since its truth for sections by
parallel planes will then follow from Theorem 2. Here, then,
k is fixed, and \ takes on successively the values 1, — 1, 0.

But the coefficients of the quadratic terms in (3) do not

contain A, and hence we reach the desired conclusion imme-
diately, by reasoning identical with that used in the proof of
Theorem 2.
The following theorem is now obvious.
Theorem 4. A jolane ivhich intersects a hyperboloid or a cone,
but not in a non-degenerate conic, cuts it in a degenerate conic,
ivhich is of the same type as any section by a parallel plane.
Accordingly; to ascertain the type of conic (degenerate or
non-degenerate) in which a plane intersects a hyperboloid, it
isnecessary merely to determine the type of degenerate conic
in which the parallel plane through the center meets the
asymptotic cone. But the planes through the center intersect
the cone in degenerate conies of all three types. Consequently,
a hyperboloid has sections of all three types.

EXERCISES
1. Show that every plane section of an ellipsoid is an ellipse

and that parallel sections are similar and similarly placed

ellipses.

2. Prove that an elliptic paraboloid has no hyperbolic sec-

tions, that sections by parallel planes cutting the axis are
similar and similarly placed ellipses, and that sections by
564 ANALYTIC GEOMETRY

parallel planes parallel to the axis are equal and similarly-

placed parabolas. •

3. Prove that a hyperbolic paraboloid has no elliptic sec-

tions, that sections by two parallel planes cutting the axes
are hyperbolas, one of which is similar and similarly placed
either to the other or to the conjugate of the other, and that
sections by parallel planes parallel to the axis, but not to a
directrix plane, are equal and similarly placed parabolas.

6. Circular Sections. Consider a section of the ellipsoid

(1)
y
+ •^
z'
+ - = 1, a^h ^ c^
a2

by a plane M passing through the axis of y. The section is an

ellipse, one of whose axes is always the mean axis, 26, of the
ellipsoid, no matter how M is situ-

ated. When M^ starting from the

(x, 2/)-plane, rotates in either direc-
tion into coincidence with the (y, zy
plane, the second axis of the ellipse,
starting from the major axis, 2a, of
the ellipsoid, decreases continuously
to the minor axis, 2c. Consequently,
there must be a single position of My
Fig. 9
in each direction of "rotation, for
which the second axis of the ellipse takes on the value 26
equal to the first. But then the ellipse is a circle.
These two positions, KOB and LOB^ of the plane can M
be constructed by describing in the upper half of the (2, x)-
plane a semicircle whose center is at and whose radius is h.
The semicircle will meet the ellipsoid in the desired points K
and L.
Since the sections of (1) by the planes KOB and LOB are
circles, so also are the sections by planes parallel to KOB and
LOB, by § 5, Ex. 1. The ellipsoid has, then, these two sets of
circular sections and, as can be shown (Ex. 1), only these two.
 QUADRIC SURFACES 565

It can be proved in the same way that the hyperboloid of

one sheet

contains just two sets of circular sections. Hence it follows,

by § 5, Th. 3, that this is true also of the cone and the hyper-
boloid of two sheets. It is to be noted, however, that there
are no the cone by planes through the
circular sections of
vertex and none of the hyperboloid of two sheets by planes
through the center.
The results obtained we now consolidate into a theorem.

Theorem. An ellipsoid, a hyperboloid, or a cone, ivhich is

not a surface of revolution, contains just two sets of circular

sections.

If, in Eig. 9, b approaches a as its limit, the planes KOB

and LOB both approach as their limits the (x, ?/)-plane. Con-
sequently, an ellipsoid of revolution has but one set of circu-
lar sections. This is true also of the hyperboloids and cones
of revolution.
The circles inwhich the planes KOB, LOB intersect the
ellipsoid evidently lie on the sphere whose center is at and
whose radius is b :

W b"' b^ b""

They therefore lie on the surface whose equation results from

subtracting (1) from (3) :

X
V52 ay \b^ cy
or c2 (a2 - b-) x"^ - a^ (b^ - c2) z^ = 0.
But this surface consists of the two planes

(4) c-Va^ -b^x ± a^b^-c^z = 0.

Consequently, these are the equations of the planes KOB,
LOB.
 .

566 ANALYTIC GEOMETRY

EXERCISES
1. To show that the ellipsoid (1) has but two sets of circu-
lar sections, prove first, using the fact that the centers of the

circles of on a line (§ 8, Problem 1), that every

any set lie
must be symmetric in a principal plane then
circular section ;

show that a section of (1) by a plane passing through the x-

axis or the 2;-axis is never a circle.

Prove geometrically that the hyperboloid (2) has two

2.

Give a construction for the planes

sets of circular sections.
through the origin which yield circular sections.

3. Find the equations of the planes just mentioned.

4. Show that the elliptic paraboloid (1), § (3), where a > b,

has two sets of circular sections, by proving first that this is

true of the elliptic cylinder

^-^y-=l, '
a>b.
a^ b^

5. A hyperbolic paraboloid has no circular sections.

Why?

7. Tangent Lines and Planes. Let the line L through the

point Pq {xq, ?/o, Zq) ^i^h the
: direction cosines cos a = X,
cos fi
= cos y = v meet the ellipsoid
fJLj

(1)
^ ^
^ + ^V^'=1
a^ b^ c^

in two distinct points, P^ and Pg- To find the coordinates,

{^1, Vi, ^i) and {X2, 2/2, ^2), of Pi and P^.
The parametric representation of L is
(2) x = XQ-]-Xr, y = y^ + ixr, z = ZQ-\-vry
where r is the algebraic distance from Pq to P: (x, y, z) ;
cf

Ch. XX, § 8. The point P of L lies on the ellipsoid, if and

only if its coordinates (x, y, z), as given by (2), satisfy (1),
that is, if and only if r satisfies the equation
 ;

QUADRIC SURFACES 567

(^)
e^s^3^^+<f+f^?)
+(^4+^4+'4-A=o.
a'

Since L intersects the ellipsoid in two distinct points, (3)

has two distinct roots. If we denote them by r^ and ra, the

coordinates of Pi and Pj ^^^

Xi = Xo-\- Xvi, 2/1 = 2/0 + /^^'i> 2:1 = Zo + V9-1

iCs = + Ar2,
a^o 2/2 = 2/0 + /^^2, 2:2 = ^o + »'^2- ^

Tangent Line. Suppose, now, that P^ lies on the ellipsoid.

Then one of the points of intersection of L, say P., coincides
with Pq. Analytically, we have

('4^ V _L ^^ 4- ?L = 1,

so that the absolute term in (3) is zero ; also, 1-2 = and ri is

the distance from Pq to Pi.

Imagine a curve drawn on the surface through P^ and P^
for example, an arbitrary plane section, 0, through Pq and Pj.

The line L is the secant PqPi of C and its

limiting position, as Pi moving along C
approaches Pq as its limit, is the tangent

I
to C at Pq. We define this tangent as the
tangent line to the surface at Pq in the direc-
tion of the curve C. Fig. 10

When P approaches Pq, then no matter

what curve C of approach is chosen ri approaches zero. But,
when ri approaches zero, the coefficient of r in (3) approaches
zero, and conversely. Consequently, the line L is a tangent to
L the ellipsoid (1) at the point Pq on the ellipsoid if and only if

^^ a^ h^ &
Tangent Plane. There are evidently infinitely many lines
L tangent to the surface at Pq. For them \, /x, v have varying
568 ANALYTIC GEOMETRY

values, which, however, always satisfy (5). To obtain the

locus of all the tangent lines i, we have only to eliminate the

auxiliary variables A, /u,, v, r from equations (2) and (5). Sub-

stituting the values of X, /x, v as given by (2) into (5) and sup-

pressing the factor 1/r, we get the equation

Xo(x-Xq) yo{y-yo) , go(g-go)_Q

a'' b^ c^

which reduces, by virtue of (4), to

(6)
?c2
a^
+ M + ?|=i.
b^ c^

But this is the equation of a plane. Hence we have the

theorem :

Theorem 1. The tangent lines at a point Pq of an ellipsoid

all lie in a plane.
The plane of the tangent lines at Pq we define as the tan-
gent plane to the ellipsoid at Pq. Its equation is given by (6).

EXERCISES
Find for each of the following surfaces the condition that
a line L through a point Pq of the surface be tangent to the
surface. Prove the analogue of Theorem 1 and deduce the
equation of the tangent plane at Pq.
1. The unparted hyperboloid. 2. The biparted hyperboloid.
3. The elliptic paraboloid. 4. The hyperbolic paraboloid.
5. The cone, Pq not being at the vertex.

6. Prove that the tangent plane to a hyperboloid of one

sheet at a point Pq is the plane determined by the rulings

which pass through Pq. Hence show that the two sets of
rulings found in § 4 exhaust all the straight lines on the
surface.

7. The same for a hyperbolic paraboloid.

8. Let Q be a quadric surface, not a cone or a cylinder.

Prove that a plane is tangent to Q if and only if it intersects

 QUADRIC SURFACES 569

Q in a point (a degenerate ellipse) or in two intersecting lines

(a degenerate hyperbola).

8. Diameters. Diametral Planes. Problem 1. Find the

locus of the centers of parallel sections of the ellipsoid

Let the common normals to the planes of the sections have

the direction components A, B, C. Let P\ (X, Y, Z) be the
center of one of the sections and let L, with the direction
cosines A, /x, v, be an arbitrary line through P, which lies in
the plane of this section. The parametric equations of L are,

then, ^

(2) a;=X4-Ar, y=Y-\-ixr, z=Z + vr,

where r is the algebraic distance from P to {x, y, z).

Since the points of intersection of L with the ellipsoid are

equally distant from P, their algebraic distances, Vi and rj,

from P are negatives of each other : Ti -f- r» =0. But Vi, r^ are
the roots of the quadratic equation

and consequently, by Ch. XIII, § 5,

This equation says that the direction whose components are

X/a2, YjW-^ Z/c' is always perpendicular to L. But L is an
arbitrary line in one of the planes of the sections and the only
direction which is always perpendicular to it is that of the
normals to these planes, that is, the direction whose compo-
nents are A, B, C. Consequently, X/a-, 7/6-, Z/c- are pro-
portional to A, B, C, or

^ ^
a' A WB c'C
 :

570 ANALYTIC GEOMETRY

The centers of the sections lie, then, on a line through the

center of the ellipsoid. Such a line is known as a diameter.
Accordingly, we can state our result as follows.

Theorem 1. The locus of the centers of parallel sections of

the ellipsoid (1) is that portion of a diameter which lies ivithin

the ellipsoid. components of the normals to the

If the direction
planes of the sections are A, B, C, those of the diameter are a^A,
¥B, c'C.

Exercise. The tangent planes at the extremities of a

diameter are parallel to the sections whose centers the diame-
ter contains.

Problem 2. Find the locus of the mid-points of a set of

parallel chords of the ellipsoid (1).

Let A, /x, V be the given direction cosines of one of the chords
and let P {X, T, Z) be the mid-point of the chord. Then
:

equations (2) represent the chord parametrically.

Since P
is the mid-point of the chord, the algebraic dis-

tances, ri and ro, from it to the end points of the chord are
negatives of each other : ri + ro = 0. Hence, as in Problem 1,

we obtain the equation (4). But, whereas in that problem

X, /x, V were auxiliary variables, here they are given constants,
and hence (4) is the equation satisfied by the point of the P
locus. But (4) represents a plane through the center of the

ellipsoid. Such a plane is known as a diametral plane. Thus

our result is

Theorem 2. The locus of the set of mid-points of a

parallel chords of the ellipsoid (1) of a dia- is that portion

metral plane lying within the ellipsoid. If the direction com-

ponents of the chords are Z, m, n, the equation of the diametral
plane is

Ix my , ^_ Q
 QUADRIC SURFACES 571

Conjugate Dia^neters and Diametral Planes.

Theorem 3. If a diameter D
contains the centers of sections
parallel to a diametral plane M, then bisects the chords M
parallel toD, and conversely.
For, let D
have the direction com-
ponents I, m, n, and let be the plane M
Ax-\-By+Cz = 0.

The condition that D contain the cen-

ters of sections parallel to M is, by Th.
that ^
1, Fig. 11
I :m :n = a^A : IrB : c^C.

The condition that M bisect the chords parallel to D is, by

Th. 2, that

A'.B:C = -'.-.-.
a^ 62 ^2

The two conditions can both be written in the form

/^\ I m n
^
^
aM.~¥B~^'
and are, therefore, equivalent, q. e .d.
A diameter D and a diametral plane M in the relationship
described are said to be conjugate. We have, then, the follow-
ing theorem.

Theorem 4. The diameter D with the direction components

I, m, n and the diametral plane Ax -|- By -\- Cz = are conjugate
if and only if

(6)
I _ m _ n
d'A~b^B~"Jc'
Exercise. Show that an axis and the principal plane per-
pendicular to it are conjugate, and that in no other case is D
perpendicular to its conjugate, M.
Theorem 5. If tivo diameters, Di and D2, are conjugate in
the ellipse E in which their plane meets the ellipsoid, each lies in
the diametral 2ila.ne of the other.
 ; :

672 ANALYTIC GEOMETRY

For, since Z>i is conjugate to D2 in E, D^ bisects the chords

of E parallel to A- ^^t the diametral plane M2 conjugate to
D2 also bisects these chords. Hence
Di must be the line in which M2 meets
JA^ X \^-^'>^^^ the plane of E, and so lies in M2. A
Similarly, A lies in the diametral plane,
Ml, conjugate to A«
Theorem
If one diameter lies in
6.
Fig. 12
the diametral plane conjugate to a second,

then the second diameter lies in the diametral i^lane conjugate to

the first.
Suppose that Di lies in the diametral plane M2 conjugate to

A- It will follow, then, by Th. 5, that A lies in the di-

ametral plane J/i conjugate to D^, if we can show that D^ and

A are conjugate diameters in the ellipse (Fig. 12). This is E
the case, for, since M2 bisects all chords parallel to Aj then

Di bisects all chords of E parallel to A- .

"

Conjugate Diameters. Conjugate Diametral Planes. Given

three diameters Dj, A? A and three diametral planes ilfi, itfo,

Mz such that Di, A, A are the lines of

intersection of M^, M2, M^ or Mi, M^, M^
are the planes determined by Di, A? A
(Fig. 13). Consider the following re-
lationships :

Ri-. Di, A, A and 3/i, M2, M^ are re- Fig. 13

spectively conjugate
R Each diameter contains the centers of sections parallel to
the plane of the other two ;

R Each diametral plane bisects the chords parallel to the

line of intersection of the other two.

According to the definition of conjugacy of a diameter and a

diametral plane, these relationships are equivalent
Theorem Aoiy one of the relationships
7. is equivalent to R
each of the other two; that is, if any one holds, so does each of

the other two.

 QUADRIC SURFACES 573

Three diameters in the relationship R2 are called conjugate

diameters, and three diametral planes in the relationship E^
are called conjugate diametral planes.
There are iniinitely many sets of three diameters and three
diametral planes in the relationship Ri. For, let be an E
by a plane through 0, and
arbitrary section of the ellipsoid
let Di and D2 be any two diameters conjugate in E. Then
the diametral planes Mi and Mo conjugate to D^ and D2 will,
by Th. 5, pass through A
and Di respectively. Finally, the
diametral plane, M^, conjugate to the diameter, D^, in which
3/1 and M2 intersect will, by Th. 6, contain Di and and A
hence must be the plane of Di and A-
The following theorems follow directly by application of
Theorems 2 and 1, respectively.

Theorem 8. TJie diameters with the direction components

h, *^i, ni, I2, m2, no, ^3, ^%j '^^3 ci^'e conjugate if and only if

y^ mi»?i2 ,
^1^2 _Q
6-^
a'' c2

l^ m^m^ Ti^fh
^Q
a^- h'- e-

52 "^
a2 c2

Theorem 9. The diametral planes

AiX+Biy-\-CiZ = 0, A2X-\-B2y-hC2Z = 0, A^x -{- B^y -{- C^z =

are conjugate if and only if
a'-A.A^+ lfBiB2 + c^CiC^ = 0,
«M2^3 + 6252^3 4- c2Co(73 = 0,
a2^3^i + 62^3^1 4- G^C^G, = 0.

EXERCISES
State and prove for the following surfaces the theorems
analogous to Theorems 1, 2.

1. The unparted hyperboloid.

574 ANALYTIC GEOMETRY

2. The biparted hyperboloid.

3. The elliptic paraboloid. Show, in particular, that the
diameters and diametral planes are all parallel to the axis.
4. The hyperbolic paraboloid. Describe the positions of
the diameters and diametral planes.

Give reasons for the following exceptions to the theorems

just proved.

5. Hyperboloids. There is no analogue to Theorem 1 for

parallel sections parallel to an element of the asymptotic cone,

and no analogue to Theorem 2 for lines parallel to an element
of the asymptotic cone.

6. Paraboloids. Thereno analogue to Theorem 1 for

is

and no analogue to Theo-

parallel sections parallel to the axis,
rem 2 for lines parallel to the axis. In the case of a hyper-
bolic paraboloid there is also no analogue to Theorem 2 for a
set of parallel lines parallel to a directrix plane.

State and prove for the following surfaces the theorems

analogous to Theorems 3, 4.

7. The unparted hyperboloid.

8. The biparted hyperboloid.

9. There are no conjugate diameters and diametral planes

for a paraboloid. Why ?
10. Find for the ellipsoid (1) the equation of the diametral
plane conjugate to the diameter through the point (xq, i/q, Zq)

of the surface.

11. Prove that the pairs of conjugate diameters and dia-

metral planes are the same for two conjugate hyperboloids.
12. Discuss the conjugacy of three diameters or three dia-
metral planes for either hyperboloid.
13. Prove Theorems 8, 9.

14. State and prove the analogues of Theorems 8, 9 for

either hyperboloid.
 :

QUADRIC SURFACES 575

9. Poles and Polars. Through a pomt Pq •

(^o? 2/o) ^o) iiot on
the ellipsoid

(1)
.
-.+f2+^=i
a^ ¥ &
an arbitrary line L is drawn meeting the ellipsoid in Qi and
^2- What is the locus of the point P which with Pq divides
Q1Q2 harmonically?
By the method used in solving the corresponding problem
in the plane, Ch. XIV, § 9, the locus is found to be the plane

or a portion of this plaije.

The point Pq(not on the ellipsoid) and the plane (2) are
said to be poleand ^lolar in the ellipsoid Pq is the pole of (2),
:

and (2) the polar of Pq. A point on the ellipsoid and the
tangent plane at the point are defined to be pole and polar.
By the methods of Ch. XIV, §§ 9-11, the following theorems
can now be proved.

Theorem 1. Let Q he a central quadric (an ellipsoid or

hyperholoid). Every point in space, except the center of Q, has
a polar with respect to Q.

Theorem 2. Every point in space has a polar with respect

to a paraboloid.
Theorem 3. Let Q he a central quadric or a paraboloid.
Every plane in space, which is not a diametral plane of Q, has

a pole with respect to Q.

The poles and polars considered in the following theorems
are taken with respect to an arbitrarily chosen central quadric
or paraboloid.

Theorem 4a. If one point lies in the polar plane of a second,

the second point lies in the polar plane of the first.

Theorem 46. If one plane contains the pole of a second, the

second plane contains the pole of the first.
576 ANALYTIC GEOMETRY

Theorem 5a. If a number of points lie on a line L, their

polar planes pass through a line L' or are parallel.

L'l
Theorem 56. If a number of planes pass
^^7 through a line L' {or are parallel), their poles

I
^ / lie on a line L.

Theorems Qa, ob are peculiar to the geometry

I
/ of space. We give a proof of Theorem 5a.
""^^^^ Let Pi : (x„ ?/i, 2i), Fo : (.^2, 2/2, ^2) be two
Fig. 14 distinct points. Their polar planes are

.,_^' ,M 4-51? _ 1=0 '

^ = ^4-M + !2?_l = 0.
a' b' c' a'' b' c^

Let Ps:{xs, y^s, 23) he an arbitrary point of the line P1P2.

Then, brCh. XXI, § 2,

X^ = pXi +(1 - p)X2, Vs = pVl +(1 - p)y2, 23 = p2;i -h(l - p)22-

Consequently, the polar plane of P3, namely

«3« , M4.?3?_l =
a' b' c'

can have its equation written in the form

pu+{l-p)v=^0,
and hence passes through the line of intersection of the polar
planes of Pi and P2, if they intersect, or is parallel to them, if
they are parallel, q. e. d.

Two lines L and L', in the relationship described in The-

orem 5a or 5b, are each said to be polar or conjugate to the
other. The following theorems concerning polar lines are
readily proved cf. Ths. 8a, Sb of Ch. XIV,
; 11. §,

Theorem 6a. TJie polar of a line intersecting the quadric in

two distinct points is the liyie of intersection of the .tangent planes
at these points.

Theorem 66. The polar of a line not meeting the quadric is

the line joining the points of contact of the two planes through the
line tangent to the quadric.
 QUADRIC SURFACES 577

Theorem 6c. The polar of a tangent to a quadric, not a

ruling, is a second tangent line with the same point of contact.
A ruling of a quadric is self-polar.

EXERCISES
Establish formula (2) and the analogous formulas for the
other quadrics. Prove the theorems stated without proof in
the text.
Discuss poles and polars with respect to a sphere (cf. Ch.
XIV, § 9, Exs. 9-11), showing, in particular, that two polar
lines are always perpendicular.

10. One-Limensional Strains, with Applications.* The one-

dimensional strain which stretches all space directly away from
the (y, 2;)-plane (or compresses all space directly towards the
(y, 2)-plane), so that each point is carried, along a parallel to
the axis of x, to a times its original distance from the (y, z)-
plane, where a is a positive constant not unity, has the equa-
tions
i) x' = ax.
578 ANALYTIC GEOMETRY

One-dimensional strains do not in general preserve angles

or areas. They never preserve volumes for example, i) car- ;

ries a portion of space of volume into a portion of space ofV

volume aV.
The product of the three one-dimensional strains i), ii), iii)

is the transformation

T: x' = ax, y' = by, z' = cz.

It is clear from the foregoing that T carries a portion of space
of volume V into a portion of space of volume ahc V.

Applications. The sphere

(1) x^ + y^ + z'^ =l
is carried by the transformation T into the ellipsoid
/,./2 ,/'2 5;'2

(2)
\ / ^ + 7i + ^=l-
a^ 0^ c^

The volume of the sphere is |7r. That of the ellipsoid is,

then, ^Trabc.

Theorem 1. TJie volume of the ellipsoid (2) is

Y= ^irabc.
Let the triples

(3) . Ai, [Xi, Vi, A.2, 1^2, V2, X3, /A3, V3

be the direction cosines of three mutually perpendicular (and

hence conjugate) diameters. A? A? Aj of ^^^ sphere (1).
They are, then, also the coordinates of three points, Pi, P2, P3,
on the sphere, which are respectively extremities of Di, D2, A*
Now T carries Di, D2, A
iiito three diameters, Di, A'? A'j

of the ellipsoid (2), and carries Pi, P2, P3 into three points,
P/, P2', P3', on the ellipsoid, which are respectively extremi-
ties of A'j A'j A'- Evidently, the triples

(4) a\i, bjjLi, cvi, aX2, 5/x2, cv2, aX^, bfi^, cvz

are both the coordinates of P/, P2', P3' and the direction com-
ponents of Z>/, D2', A'-
 : ;

QUADRIC SURFACES 579

Considered as the direction components of Z)/, A'? A'j the

triples (4) satisfy the conditions of Th. 8, § 8. Hence we
have the theorem
Theorem 2. Tlie transformation T carries three mutually
perpendicular (and therefore conjugate) diameters of the sphere
(1) into three conjugate diameters of the ellipsoid (2).

It follows, by Th. 7, § 8, that T carries three mutually

perpendicular diametral planes of (1) into three conjugate
diametral planes of (2), and carries a diameter and the per-
pendicular diametral plane of (1) into a diameter and the con-
jugate diametral plane of (2).
Considered as the coordinates of P/, P2, P/, the triples
(4) give immediately, as the squares
of the half-lengths of the
diameters Di, D2', D^,

a%2 ^ 52^^2 _|. c2y2^ ^2X2^ + bW + CV, ^2X32 + 62^g2 4. 02^32.

The sum of these squares is

a' (Ai2 -f Ao^ + X3-) 4- ?>

-(/xi^ + /X2^ + /X32) + c^ (vi^ + V2^ + V32).
Since the triples (3) are the direction cosines of three mutu-
ally perpendicular lines, so also are the triples

^b ^25 ^3) f^i) f^2) f^Z) ^1) ^2) ^3

of. Ch. XXIV, § 6. Hence the above sum has the value
a2 + 62 + c"-

and is therefore independent of the three conjugate diameters

A', A', A' taken. ^ '

Theorem The sum of the squares of the lengths of

3. three

conjugate diameters of an ellipsoid is constant.

EXERCISES
1. Prove analytically the properties A, B, C of one-dimen-
sional strains.
2. What angles and what areas does the transformation i)

preserve ?
 :

580 ANALYTIC GEOMETRY

3. Prove that i) carries a region of volume V into a region

of volume a V.

4. Show that T carries a line with the direction components

Z, m, n into ^ line with the direction components al, bm, en.

5. The plane M goes into the plane M' under T. If the
direction components of the normals to M are A, B, C, what
are those of the normals to M' ?
6. Assuming the equation of the tangent plane to the sphere

(1) at the point deduce by means of the transforma-

(o^o^ 2/oj ^o))

tion T the equation of the tangent plane to the ellipsoid (2)

at the point (xq', ?/o', Zq').

7. Show that a hyperboloid of general type can always be

carried into a hyperboloid of revolution by means of a trans-
formation of the form T.

EXERCISES ON CHAPTER XXIII

Find the equation of the quadric surface generated by

1.

the lines x— Xz = 0, Xy — z = 0, where A is a parameter. De-

termine the equations of the second set of rulings and set up
a parametric representation of the surface.
2. The same for the lines y — X — 1 = 0, \x — z-\-2=:0.
3. Find the equations of the planes which pass through the
line y = 2, X -{-2z = and are tangent to the ellipsoid

a;2 4. 3^/2 + 2:^2 3^ 6.

4. Prove that the sections of the hyperbolic paraboloid and
hyperbolic cylinder

'
a2 62 a2 b^

by the same plane or by parallel planes, oblique to the 2;-axis,

are hyperbolas, each of which is similar and similarly placed

to the other or to the conjugate of the other.

5. The umbilics of a quadric surface which has circular sec-

tions are the extremities of the diameters which contain the
 QUADRIC SURFACES 581

centers of these sections. Find their coordinates in the case

of the ellipsoid (1), § 1.

Similar Quadrics

Definition. Two central quadrics are said to be similar if

the principal sections of one are similar, respectively, to the

principal sections of the other.

6. Prove that the ellipsoids defined by the equation

^^ a" b"' c'-

where A is a parameter, are similar.

7. ShoAV that, of the hyperboloids represented by the
equation

^^ .
a'' b' c-

those for which A is positive are all similar, and that this is
true also of those for which X is negative. Prove that all the
hyperboloids have the same asymptotic cone.
8. Prove that all the ellipsoids (1) have the same pairs

of conjugate diameters and diametral planes.

9. The same for the hyperboloids (2).

Definition. Two paraboloids of the same type are similar,

if the principal sections of one are proportional in scale
(Ch. YI, § 1) to the principal sections of the other.

10. Prove in each case that the paraboloids defined by the

given equation are all similar :

(«) 5 + g = 2^^, ^^0; (b)

J-g = 2X^, X#0.

Ruled Surfaces
11. that the pencil of planes through the ruling Vq of
Show
the hyperboloid (1), § 4, cuts the surface in the set of rulings
U and that the pencil of planes through Uq cuts it in the
rulings V.
582 ANALYTIC GEOMETRY

12. Let P be a point on the minimum ellipse of the hy-

perboloid (1), § 4, and let <^ be the common angle which the
two rulings through P make with the 2:-axis. Prove that
tan <^ =6i/c, where hi is the half-length of the diameter of the
minimum ellipse which is conjugate to the diameter through P.
13. Using the result of Ex. 12, show that a hyperboloid of
revolution of one sheet can be generated by the rotation of a
ruling of either set about the axis which does not meet the
surface.

14. Prove that the rulings of one set on a hyperbolic para-

boloid intercept proportional segments on two rulings of the
other set.

Loci

15. Find the locus of a point which moves so that its dis-
tance from a fixed point bears to its distance from a fixed

plane, not through the point, a constant ratio, Tc.

Ans. A quadric of revolution which is an ellipsoid, an

elliptic paraboloid, or a hyperboloid of two sheets, according
as k is less than, equal to, or greater than unity.

16. A point moves so that its distance to a fixed point bears

to its distance to a fixed line, not through the point, a constant
ratio. Find its locus.

Exercises 17-19. In connection with these exercises, Exs.

28, 29, p. 522 will be found useful.
17. Eind the locus of a point which moves so that its dis-

tances to two skew lines are always in the same ratio, k.

18. Prove that a line which is rotated about an axis skew

to it generates a hyperboloid of revolution of one sheet ; cf.

Ex. 13.
19. Let L and L' be two fixed skew lines and let and M' M
be two planes, which pass through L and U
respectively and
so move that they are always mutually perpendicular. Find
the locus of their line of intersection.
 QUADRTC SURFACES 583

20. The locus of a line which so moves that it always inter-

sects three fixed skew lines, not parallel to a plane, is a hyper-

boloid of one sheet. Prove this theorem in the case that the
fixed lines are

I x = c, j X —— c, \ x = — z cot 6,
1 2/ = 2 cos ^; [y = — zcosej \y = csindy
where c ^ and 6 =^ 0, n-.

21. The moves that it always inter-

locus of a line which so
sects three fixedskew lines, parallel to a plane, is a hyperbolic
paraboloid. Prove this theorem in the case that one of the
fixed lines is the axis of z and the others have the equations
a; = c, = my x-=— c^z = — my, where cm 4^ 0.
2; ;

22. A line moving so that it is always parallel to a fixed

plane, 3/, and always intersects two fixed skew lines, neither

of which is parallel to 3f, generates a hyperbolic paraboloid.

Prove this theorem when M
is the (x, 2;)-plane and the two fixed

lines are the last two of the three in Ex. 21.

 CHAPTER XXIV
SPHERICAL AND CYLINDRICAL COORDINATES. TRANS-
FORMATION OF COORDINATES

1. Spherical Coordinates. Griven a point 0, a ray OA issu-

ing from 0, and a half-plane m bounded by the line of the ray

OA. Let P be any point of space. Join P

to and construct the half-plane, v, deter-
mined by OA and OP. Denote the distance
OP by r, the angle AOP by <^, and the angle
from the half-plane m to the half-plane p by 0.
Then (?•, <^, 0) are the spherical coordinates of
the point P.
For a given value, ?'o) of the radius vector r,
the point P lies on a sphere whose center is
at and whose radius is r = ro. The angle
is the longitude of P, measured from the
Fig. 1
prime meridian m, and the angle <^ is the
colatitude(complement of the latitude), at least for a point P
on the upper half of the sphere.
The radius vector r is, by definition, positive or zero. The
colatitude <j> shall be restricted to values between and tt in-

clusive : < <^ < TT. The longitude 9 shall be unrestricted ;

it

shall be taken as positive if measured in the direction shown,

and as negative, if measured in the opposite direction.*

* It is possible to define spherical coordinates so that r or or both

are also unrestricted. Systems of these extended types are not often
necessary, and when exceptional need for them occurs, they can easily

be introduced.
584
 SPHERICAL AND CYLINDRICAL COORDINATES 585

It is clear that the r- and (^-coordinates of a given point P

are unique, while the ^-coordinate has infinitely many values,
each two differing by an integral multiple of 2 tt* Conversely,
if ?', (^, 6 are given, such that r > and < <^ < tt, a unique
point P is determined.
Let i'q, <t)o, 00 be particular values of r, 6, such that ?'o > <fi,

and < c^o < T^- The equation r = ?'o represents a sphere,
whose center is at and whose radius is <^ = <^o defines ?-o ;

one nappe of a circular cone whose vertex is at and whose

axis lies along OA ; finally = Oq represents a meridian half-
plane issuing from the line of OA.

Transformation to and from Rectangular Coordinates. Let

P be any point of space whose coordinates, referred to a system
of rectangular axes, are (a;, y, z). Let P have the spherical
coordinates (r, (^, 6) with respect to 0,
OA, and m, as chosen in the figure. It is
clear that

X = ON cos 6, y = ON sin 0,
and ON = r sin <^, z = r cos <^.
—y
Hence the values of x, y, z in terms of ?-,

(fi, 6 are

x = r sin <^ cos 6, y = r sin ^ sin 0,

(1)
z = r cos <^,

Since r is the distance from to P : {x, y, z), we have also that

(2) r2 = cc^ + y'-

-h zK

EXERCISES
1. Plot the points (2, 90°, 180°), (4, 60°, - 30°), (8, Itt, Itt).

2. Find the rectangular coordinates of the points of Ex. 1.

3. Eind the spherical coordinates of the points (0, 2, 0),

(3, 4, 12), (2, — 2, — 1), checking each result by a figure.

* It is to be noted, however, that for every point P on the line of OA

6 is undetermined and that for O in particular is also undetermined.

Cf. Ch. X, § 1.
 : ; ; ; ; ; ; ;:

586 ANALYTIC GEOMETRY

4. What do the following equations represent ?

(a) r = 7; (b) = 30°

<^ ;
(c) ^ = |,r;
(d) tan^ = l; (e) tan<^ = l; (/) 4 cos"- <^ = 1.
5. Find the equations in spherical coordinates of the follow-
ing surfaces :

(a) The sphere of radius 5, center at ;

(&) The meridian half-plane of longitude 35° ;

(c) The complete plane determined by this half-plane

(d) The upper nappe of the circular cone whose vertex is

at 0, whose axis is along OA, and whose generating angle is

60°;
(e) The lower nappe of the cone of (d) ;

(/) The cone of (d).

6. What do the following pairs of equations represent ?

(a) r = S, = 120°
cf> (d) r = 3, tan2 = 1 c^

(6) r = 3, = |7r;
(9 (e) tan = 2, = f tt;
(9 <^

# (c) = 30°, = 45° <^ (/) tan ^ = - 1, cos^ = <^

i.

7. Find the equations in spherical coordinates of the fol-

lowing curves

(a) The small circle on the earth of colatitude 47°

(6) The semicircle on the earth of longitude 135°
(c) The complete circle determined by the semicircle of (6)
(d) The ray from of colatitude 60° and longitude 25°.

8. Determine the locus of each of the following equations

(a) r = 4 cos <^ ; (5) r = 6 sec <^ ;
(c) r = 3 esc <^.

Find the equations in spherical coordinates of the following

surfaces. Identify each surface.

9. a;2-f 2/2+^2= 9. 13. 3x-^2y = 0.

10. a;2 + ?/2_A:V = 0. 14. 3^-4 = 0.
11. 0^2
+ 2/2+22^4?/. 15. 2a; + 5 = 0.

12. 4(aj2 + 2/'^)+922 = 36. 16. x^ = 2yz.

 SPHERICAL AND CYLINDRICAL COORDINATES 587

2. Cylindrical Coordinates. Given a point 0, the axis of z

through 0, and the plane through K

perpendicular to the
axis of z. In K
introduce a system of polar coordinates, as
shown. Let P be any point of space and let be its projec- N
tion on K. Then the polar coordinates, r and $, of and the N
directed line-segment NP = z deter-
mine the position of P. The three
numbers, taken together, are known
as the cylindrical coordinates (r, 6, z)

of P.
As in the case of polar coordinates
in the plane (Ch. X, § 1), r is re-

stricted to be positive or zero, while

is unrestricted. The positive direc-
tion of rotation for the measurement Fig. 3

of 6 is as indicated in the figure.

If ro (> 0), 6q, Zq are particular values of r, 0, z, the equation
r = ?'o
represents a circular cylinder whose axis is the axis of
2 6 = Oq defines
; a half-plane issuing from the axis of z, and
z = Zq represents a plane perpendicular to the axis of z.

Transformation to and from Rectangular Coordinates.

Choose in the plane K
the Cartesian axes of x and y shown in
Fig. 3. Referred to these axes and the axis of z, P has the
rectangular coordinates (x, y, z).

It is clear that the z of the rectangular coordinates of P is

precisely the z of the cylindrical coordinates of P. The for-

mulas for X, y in terms of r, 6 and for r, in terms of x, y

are those of transformation in a plane from polar to rectangu-
lar coordinates, and vice versa (Ch. X, § 6). In particular,

(1) x = r cos 6, y = 1' sin 6,

(2) r2 = aj2_,_2/2.

EXERCISES
1. Plot the points (2, 40°, 5), (4, -f tt, - 3), (0, 122°, 1).

2. Find the rectangular coordinates of the points of Ex. 1.

 : : ; ; ;

588 ANALYTIC GEOMETRY

3. Find the cylindrical coordinates of the points (3, 4, 8),

(12, — 5, — 3), (0, 0, — 6), checking each result by a figure.

4. What do the following equations represent ?

(a) r = 5; (6) 6> = 225° ;

(c) tan (9 = 1.
5. Find the equations in cylindrical coordinates of the fol-

lowing surfaces
(a) The circular cylinder of radius 7 whose axis is the axis
of 2;;

(6) The half-plane bounded by the axis of z, the angle from

OA to it being 60° ;

(c) The complete plane determined by this half-plane.

6. What do the following pairs of equations represent ?

{a) r=3, ^ = —Itt; (c) 2^ = 5, 6* = 120° ;

(6) r = 5, ^ = - 6; (d) 3^ - 8 = 0, tan^ = 2.

7. Find the equations in cylindrical coordinates of the
following curves
(a) An arbitrary line parallel to the s-axis
(6) The circle of radius 3, whose center is on the 2;-axis and
whose plane is parallel to K and 5 units below it

(c) An arbitrary ray perpendicular to the axis of z and issuing

from a point on it

(d) The line of this ray.

8. Determine the locus of each of the following equations :

(a) r'2 + 2^ _ 9 .
(5) r = 4 sin ^ ;
(c) r sin (9 = 5.
Find the equations in cylindrical coordinates of the follow-
ing surfaces. Identify each surface.
9. Ex. 9, § 1. 10. Ex. 10, § 1. 11. Ex. 11, § 1.

12. Ex. 13, § 1. 13. Ex. 14, § 1. 14. Ex. 15, § 1.

15. 3 {x^ + 2/-) —2 ;2^ = 6. 16. xy -j-yz-\- zx = 0.

17. Prove that a plane through OA (Fig. 1) or Oz (Fig. 3)
is represented by the same equation in both spherical and
cylindrical coordinates.
 ; ;

SPHERICAL AND CYLINDRICAL COORDINATES 589

3. Triply Orthogonal Systems of Surfaces. Consider the

three sets, or families, of planes which ^re parallel to the co-
ordinate planes of a Cartesian system. These families of
planes evidently have the following properties (a) Through :

each point of space there passes just one plane of each family
(5) Two planes of different families intersect at right angles.
We say, then, that the three families of planes form a triply

orthogonal system of planes.

The equations of the families are, respectively,
(1) x = 'k, y = h ^ = 'i^h

where k, I, m are arbitrary constants, or parameters, each tak-

ing on any value, positive, zero, or negative.
Since through a point P\ (xq, yo, Zq) there pass just three
planes, one from each family, namely the planes x = x^, ?/= ?/o,
z — Zq, and since, further, P is the only point which the three
planes have in common, the position of P can be thought of as
determined by the three planes. From this point of view,
then, the basis of the rectangular coordinate system is seen to
be the triply orthogonal system of planes (1).

In the case of a system of cylindrical coordinates, consider

the three families of surfaces :

(2) r = k, .0 = 1, z = m,
where, of the parameters k, I, m, k cannot be negative (or zero),
I may be restricted to the range of values : < < 27r,
^ and m
is unrestricted.
The first family of surfaces consists of the
circular cylinders with the axis of z as axis
the second family is made up of the half-planes
issuing from the axis of z ; and the third, of
the planes perpendicular to the axis of z. It
is easily seen that through each point of space,
with the exception of those on the 2:-axis, there
passes just one surface of each family, and
that two surfaces of different families intersect Fig. -4
590 ANALYTIC GEOMETRY

orthogonally. We say, then, that the three families form a

triply orthogonal system of surfaces.
A point P: (ro, 60, Zq), not on the 2;-axis,* is the single point
of intersection of the three surfaces, r = rQ,d = 6o,z = Zq which
pass through it. Thus the basis of cylindrical coordinates is

the triply orthogonal system of surfaces (2).

, EXERCISE
Write the equations of the three families of surfaces peculiar
to a spherical coordinate system. Describe each family and
draw a figure showing three surfaces, one from each family,
their curves of intersection and their coi^mon point. Prove
that the three families constitute a triply orthogonal system
of surfaces and show that this system can be considered as the
basis of spherical coordinates.

4. Confocal Quadrics. Consider the quadrics

^^ a'^-k b'^-k c^-k

'

^^
(2)
-^+-t
a^-l^ b^-l
^=1,
l-c^
(^<l<b\
^

(S)^ —^— 7n m ^— ^— = 1, 6- < m < a\

^ a^ 6^ m — c^
where a, b, c are positive constants such that a>b> c and
A:, I, m
are parameters subject to the given restrictions.
The surfaces (1) are all ellipsoids the surfaces (2), all ;

hyperboloids of one sheet opening out along the axis of z and ;

the surfaces (3), all hyperboloids of two sheets cutting the

axis of X.

The coordinate planes are the principal planes of all the

* To remove add to the family of cylinders, r=k>0,

this exception,
the axis of Through a point P of this axis passes, then, one
z, r = 0.
surface each from the first and third families, and every surface of the
second. However, all these surfaces have but the one point P in com-
mon and hence can be considered as determining the position of P.
 ;

SPHERICAL AND CYLINDRICAL COORDINATES 591

surfaces. It is readily shown that the sections by the (x, y)-

plane of all the surfaces are confocal conies, the common foci
being at the points ( ± Va'^ — 6"^ 0, 0). Similarly, for the sec-
tions by the (z, aj)-plane of all the surfaces, and for the sections
by the {y, 2)-plane of the surfaces (1) and (2), — the (y, 2;)-plane
does not cut the surfaces (3). This property of the surfaces
(1), (2), (3) is expressed by calling them confocal quadrics.
It can be shown that through each point of space, with the
exception of those in the coordinate planes, there passes just
one surface of each type and that two surfaces of different
types intersect orthogonally all along a curve.* Consequently,
the confocal quadrics forma triply orthogonal system of surfaces.
This triply orthogonal system differs in one respect from
those studied in § 3, in that the three surfaces, one of each
type, which pass through a
point P situated in a given
octant intersect not only in P
but also in one point of each
of the other octants ; this is

clear since all three surfaces

are symmetric in each coordi-
nate plane. Consequently, in
the so-called ellipsoidal coordi- Fig. 5
nate system based on the con-
focal quadrics there are eight points with the same coordinates.
This ambiguity can be avoided, however, by considering only
a restricted region of space, for example, the first octant.
The equations (1), (2), (3) can be written as the single
equation
Ct"^
.r z'
(4)
a^-\
+
b'^-x
+ c^-x = 1,
•

where X is arbitrary except that it shall not take on the values

c2, 62^ a^. If A < c^,equation (4) defines the surfaces (1), etc.
finally, if A > a^, (4) has no locus.

* Cf Osgood, Differential and Integral Calculus,

. p. 326.
 i

592 ANALYTIC GEOMETRY

EXERCISE
Show that the equation

where A .is a parameter not taking on the values a^ and b^j

represents three families of paraboloids defined by the in-

equalities \ < b-, 6- < A < a^j a^ < A. Describe each family
and show that the sections of all the surfaces by the common
principal planes — the (y, z)- and (z, a5)-planes — are confocal
parabolas with the axis ot z as axis. The surfaces are known
as confocal paraboloids. They form a triply orthogonal system.

Transformation of Coordinates
5. Transformation to Parallel Axes. To transform from a
system of rectangular axes to a new system of axes having the
same directions as the old, but with a dif-
3-
ferent origin, consider a point P whose
, coordinates with respect to the two systems
are, respectively, {x, y, z) and (a;', y'^ z').

Then
Vo^ i /-j^x X =x' + Xq, y =y' + 2/0,

.-^~—- z =z'-\-Zo,
Fig. 6
(2)
X = X — Xq. y =y — y^^
Z = Z — Zq,
 .

. TRANSFORMATION OF COORDINATES 593

that is, on transforming to parallel axes with the new origin

at the point (1, — 1, ^), the equation becomes

2 a;'2 ^. 3 ^/'^ -4.z'^= 12.

Equation (3) is thus seen to represent a hyperboloid of one

sheet whose center is at (1, — 1, ^) and whose axes are parallel
to the coordinate axes. The hyperboloid opens out in the
direction of the axis of z and the semi-axes of the minimum
ellipse have the lengths V6 and 2.

EXERCISES
Determine and draw roughly the surface represented by
each of the following equations.
1. z^-4.x-6z-\-lS = 0.
2. 92/2-42;2^-36?/-f 82-4 = 0.
3. 2a;24.2?/2-f-32^-4x-f-8y-122-f 16 = 0.
4. a;2 -
- 3 4- s2 8 + 12 + 6 + 13 = 0.
?/2 a; ?/ 2!

5. 31/2 4- 422 + 4a; -62/+ 16^ + 27 = 0.

6 2 + 4 _ 3 22 _ 8 - 2 4 y - 3 - 1 9 == 0.
a;2 ?/2 a; 2;

7. 2a;2-3?/2 + 22 + 8a;H-187/-162-3 = 0.
8. a;2-|-2i/2 + 622_2a;-2?/ + 182 + 9 = 0.

9. 2a;2-52/2 + 322 + 20?/ + 62-47 = 0.

10. ?/2-2a;2-2a;-6?/H-22 + ll = 0.

6. Rotation of the Axes. Through the origin of the usual

(right-handed) system of (a;, y, 2;)-axes, choose arbitrarily three
mutually perpendicular directed lines to
serve as the axes of a new (right-handed)
system of coordinates (x', y', z'). Let the
direction angles of the axis of x', referred
to the old system, be cci, /3i, yi, let those
of the axis of y' be 02, ^2^ 72? ^''^^ those
of the axis of z', a-^, (3^, 73.

Let an arbitrary point P of space Fia. 7

 :

594 ANALYTIC GEOMETRY

have the coordinates {x, y, z) and (x', y\ %') with respect to the
two systems. Join to P by the two broken lines OMNP
and OM'N'P, where
0M= X, MN =y, NF=z', OM' = x', M'N' = y\ N'P = z'.
Then
(1) Proj. OM^ Proj. MN -\-Vto]. NP
= Proj. OM' + Proj. M N' -\- Proj. N'P,

no matter on what directed line the projections are taken.

Choosing the positive axes of x, y^ and z in turn as this directed
line, we have

x = x' cos «! + cos 4- cos a^,

2/' oto 2;'

(2) y = ^' cos pi + y' cos + z' cos ^3, /32

2 = cos yi +
0; cos y2 + COS 73.
2/' 2:'

Here cos^i, cos yi

cos «!, the coefficients of x' — are the —
direction cosines of the axis of x' cos 025 cos /Sz, cos y2, those of
;

the axis of y' and cos a^, cos ^3, cos yg, those of the axis
;

of z'. Let us denote these direction cosines, for the sake of

brevity, by Xj, fxi, vi, Ao, 1x2, V2, A3, /1A3, vg, respectively.
Since these triples of numbers are direction cosines and,
moreover, direction cosines of three mutually perpendicular
lines, we have
•^1^ + fJ-i^ H-vi^ = 1) ^A2 + /A1/A2 + V1V2 = 0,
(3) X2^ + fX2^ + V22 = 1, X2X3 + /X2/A3 + V2VS = 0,
V+ i^3^ + V32 = 1, A3A1 + /A3/X1 + V3V1 = 0.
Since the three directed lines form a right-handed system, it
follows by Ex. 19 at the end of Ch. XVIII that the determi-
nant of their direction cosines has the value plus one

(4) \\ fx, v\ = l.

Equations (3) and (4) express completely the fact that the
three given lines through which serve as the new axes are
directed, mutually perpendicular lines forming a right-handed
system.
 :

TRANSFORMATION OF COORDINATES 595

The direction cosines of any one of the three directed lines

can be expressed simply in terms of those of the other two

(Exs. 17, 18 at the end of Ch. XVIII)

(5) X2 = /^3n — /A1V3, /W,2 = V3A1 — V1X3, V2 = As/Xi — Xi^3,

A3 = /X1V2 — 1^-2^1, H-S = ^l'^2 -~ ^2'^lJ ^S — ^1/^2 — A2/X1.

Since «!, a2, a^ are the angles which the axes of cc', 2/', z'

make with the axis of x, they are the direction angles of the
axis of X with respect to the new axes. Similarly, ^1, ^2? ^^3

and yi, y2) 73 ^^^ respectively

the direction angles of the axes
of y and z, referred to the new system. Consequently, the
equations of transformation from the new axes to the old are

X' — X cos «! + cos ^x-\- Z cos yi,

2/

(6) y' = x cos a2 + y cos ^2 + ^ cos y2,

z' — X cos H- cos ^3 + cos y3.
ttg 2/ 2;

The direction cosines of the old axes with respect to the new
are, in our notation, Ai, A2, A3, fx^, 1x2, ^3, vi, V2, V3- ^^ is clear
that between these three triples there exist relations similar
to the relations (3), (4), (5) for the original triples.*
The accompanying diagram gives equations
(2) and (6) in skeleton. Reading across we
obtain (2) and reading down we get (6). Also,
the rows give the direction cosines of the old
axes with respect to the new, and the columns,
those of the new axes with respect to the old.

Example 1. Transform the equation of the surface

(7) 13 ar^ + 13 2/^ + 10 z- -j- S xy - 4.yz - 4:xz - 36 =

to new axes through 0, whose direction cosines are respectively
122
"3' 3"' 3>
2.
3'
12.
3' 3'
2.2.
3' 3'
i.
3*

* Of thenew equations only those of the form (3) are different from the
old. The new equation (4) is obtainable from the old by interchanging
rows and columns in the determinant. Similarly, if in the present equa-

tions (5) the columns are written as rows, the result is the new equations

(5).
 :

596 ANALYTIC GEOMETRY

Here x = i{- x' -\-2y' -}-2z'),

2/ = i( 2x'- y'-h2z%
z = i( 2x'-\-2y'- z').
Substituting these values for Xy ?/, z in (7) and simplifying the
result, we obtain

This equation represents an ellipsoid of revolution about

the 2:'-axis as axis. Hence (7) represents an ellipsoid of revo-
lutionwhose axis is the line through with the direction
components 2, 2, — 1.
Example 2. What surface is represented by the equation

(8) 13a;2-|-132/' + 10;2- + 8a.'2/-4?/2-4ic^

-10aj + 262/ -402 + 22 = 0?
We make the transformation to parallel axes

(9) x = x' + Xq, y=zy'-^yQ, z = z' -\- Zq,

aiming to choose the new origin (ccg, yo, Zq) so that in the equa-
tion resulting from (8) the linear terms in x', y', z' do not
appear. Substituting the values of x, y, z given by (9) into
(8), collecting terms, and then setting the coefficients of x', y',

and z' equal to zero, we obtain the equations

13aJo+ 42/0-220- 5 = 0,
4aJo + 13 2/0 -2^0 + 13 = 0,
2^0+ 2/0 — 5 20 + 10 = 0.
These equations have a unique solution, namely, Xq = 1,
2/0 = - 1, 2o = 2.
If (8) is transformed to parallel axes with the new origin at
the point (1, — 1, 2) thus determined, it becomes
13a;'2 + 132/'' + ^^^"' -^^xy - 4.y'z' -^x'z' - 36= 0.
But this is the same equation in x' y', 2' as (7) is in x, y, z. ^

Hence (8) represents an ellipsoid of revolution whose center is

at the point (1, — 1, 2) and whose axis of revolution has the
direction components 2, 2, — 1.
 TRANSFORMATION OF COORDINATES 597

Example 3. Consider an equation in which only one of the

terms in xy, yz, zx is present, for example, the equation

(10) 2 x"' - y"'

- z' - 2 yz - 4. X -j- 6 y-^ 2 z-\- 2 = 0.
The term in yz in this equation can be removed by rotating
the y- and 2;-axes about the axis of x through a suitable acute
angle 6 ; that is, by application of the transformation

(11) x = x% y = y' cos 9 — z' sin 6, z = y' sin 6 -\-z' cos 0.

According to Ch. XII, § 5, the desired angle 6 is 45°. Trans-
forming (10) by the rotation of axes (11), where 6 = 45°, we
obtain
x'2 - 2/'2 - 2 x' 4- 2 V2 2/' - V2 2' + 1 = 0.

This equation can be written in the form

(x' - 1)2 - {y' - V2)2 = V2(0' - V2),

and hence becomes
x"'^-y"^=z^2z",

when referred to axes through the point (1, V2, V2) parallel
to the axes of x\ y', z'.

It follows, then, that (10) represents a hyperbolic paraboloid

whose vertex, referred to the (x', y', 2')-axes, is at the point

(1, V2, \/2) and whose axis is parallel to the axis of z'.

Remark. The general method of procedure to determine

the surface represented by an equation of the second degree in
«, y, z is that of Examples 1 and 2 the equation is first trans- ;

formed by a change of origin to remove the linear terms in

«, ?/, z and is then subjected to a rotation of the axes to get
rid of the terms in xy, yz, zx. This method cannot be applied,
however, to equation (10) of Example 3, for it is impossible to
transform (10) so that the linear terms disappear, since, if this
were possible, the surface would be symmetric in the new
origin (Ex. 11 at the end of the chapter), whereas we know
that a paraboloid has no point of symmetry. Accordingly,
in Example 3 and in similar cases, the axes are first rotated to
598 ANALYTIC GEOMETRY

remove the terms in xy, yz, zx and then a proper change of

by the new equation, is made.
origin, as suggested

EXERCISES
1. Find the equations of the rotation of the axes which

introduces the two directed lines through with the direction

cosines |^, |, — f and f f 4 as the axes of x' and y' respectively.
,
,

2. Find the equations of a rotation of the axes which in-

troduces the planes,

x + 2y + 2z = 0, 2x-2y + z = 0, 2x-\-y-2z = 0,
as the (x', y')-, (y', z')-, and (z, x')- planes, respectively.

3. Find the equations of a rotation of the axes which in-

troduces the planes

Sx-\-4:y-j-12z = 0, 12x-{-Sy-4.z =
as the (x'y y')- and (?/', z')- planes.

4.^ Transform the equation of the hyperbolic paraboloid

07- — 2/2 = 2 m2
by a rotation of the x- and ?/-axes through an angle of — 45°
about the axis of z. Ans. x'y' = mz'.

Determine the surface represented by each of the following

equations.

5. 5a;2 4-52/2-f 4^2 -6x2/ -32 = 0.

6. - 2 y^ - 5z'^ -4:yz-\-4:y-\- 10 z -^1 = 0.
Sx"^

7. x'^-2y'^-\-z'^-2xz-x-\-3z-3 = 0.

8. Transform the equation of the surface

5x'^-2y''' + llz^-^12xy i-12yz-U =
by the rotation of the axes of Ex. 1. Thus identify the sur-
face.

9. The equation
10 x'- -\- 13y'' -\- ISz^ - 4.xy - 10 yz - 4:xz - 36y -^ 36z =
 :

TRANSFORMATION OF COORDINATES 599

represents a central quadric whose axes have the direction

components 2,-1, — 1, 0, 1, — 1, 1, 1, 1. Identify the
quadric.

10. The equation

represents a paraboloid whose principal planes are x-^ y = 0,

X —
y -\- z 0. =
Identify the paraboloid.
11. Show that the equation of* a sphere whose center is at
the origin is not changed by any rotation of the axes. Actu-
ally carry through the transformation.

7. The General Equation of the Second Degree. We have

defined a quadric surface as any surface represented by an
equation of the second degree in x, y, z, that is, by an equation
of the form
(1) Ax"^ + By'- -\-Cz'''-{-2A'yz+2B'xz-\-2 C'xy
- -\-2A"x-^2B"y + 2C"z + F=0.
We propose now to ascertain whether there are
types of
quadric surfaces other than those already discussed in Chs.
XXII, XXIII, and to sketch a method whereby the type of
surface defined by a given equation of the form (1) can be
determined.
As in the corresponding problem in the plane (Ch. XII),
transformations of coordinates play an important role. In
particular, the expressions formed from the coefficients of (1),
which are invariant (Ch. XII, § 6) under any change of axes,
are fundamental. Chief among these invariant expressions
are the determinants

A
D=
 :

600 ANALYTIC GEOMETRY

We state, without proof, the following theorems

Theorem 1. If equation (1) represents a surface, and if

D ^0, the surface is symmetric in just one point. If D = 0,
there is in general no point of symmetry, and when there is one,

there are infinitely many.

Theorem 2. If equation (1) represents a surface and if

D =^0, the surface is symmetric in three mutually perpendicular
planes, and these are, in gen'eral, all the planes of symmetry. If
D = 0, it is symmetric in tico perpendicular planes and these
are, in general, all the planes of symmetry.
It is clear from these theorems that, in discussing equation
(1), two essentially different cases arise, according
as D ^fc
or Z> = 0.
Case 1. D^O. A by an equation of the
surface defined
form (1) for which Z> ^^ symmetricis in a unique point 0',
by Th. 1. The coordinates of 0' can be found by the method
of Example 2, § 6. A transformation to parallel axes with

the new origin at 0' removes the linear terms in (1), leaves the
quadratic terms unchanged, and, as can be shown, makes the
constant term into A/Z). Thus (1) becomes

(2) Ax'' + By'^ + Cz'^ + 2 A'y'z' + 2 B'x'z' + 2 C'x'y' + A/i)=0.

Since D =^ 0, the surface is, by Th. 2, symmetric in three
mutually perpendicular planes whose common point, since it
is a point of symmetry, must be 0'. To determine from equa-
tions (1) or (2) the precise positions of these planes through
0' is a problem of intrinsic difficulty which we shall not
attempt to discuss. When once the positions are known, how-
ever, a rotation of the axes which brings the coordinate planes
into coincidence with them serves, either immediately or
eventually, to remove the terms in y'z', z'x', and x'y' in (2);

cf. Ex. 12 at the end of the chapter. We obtain, then, the

final equation

(I) aa;"2 + b^j"' + cz"^ + A/D = 0,

 TRANSFORMATION OF COORDINATES 601

For this equation, D = abc and hence, since Z) ^ 0, no one of

the coefficients a, b, c can be zero.

If^=^0, (I) and hence (1) represents a central quadric (an

ellipsoid or hyperholoid) or, in case a, b, c, A/Z) are all of the

same sign, it has no locus.*
If A = 0, (I)and hence (1) represents a cone, or, in case a, b, c

are all of the same sign, a point.

Case 2. B = 0. A surface defined by an equation of the

form (1) for which D = has in general no point of symmetry
(Th. 1), and hence it is in general impossible to transform to
parallel axes so that the linear terms in (1) disappear. There
are,however, at least two mutually perpendicular planes of
symmetry, by Th. 2. If the positions of two such planes are
known, a rotation of the axes whereby two of the coordinate
planes become respectively parallel to them serves, either im-
mediately or eventually, to remove the terms in yz, zx, and xy
in (1) cf. Ex. 13 at the end of the chapter.
;
Thus (1) becomes

(3) aa;'2 + + cz"--\-2 a"x' + 2 b"y' -f 2 c"z' -\-F=0.

by"^

For this equation, D = abc, and, since Z) = 0, abc = 0. Now

a, b, care not all zero, since otherwise (3), and hence (1),
would not be a quadratic equation. Two cases then arise,
according as one or two of the coefficients a, b, c vanish.
A. One of the coefficients a, b, c vanishes. Since (3) bears
equally on x', y', z' it is immaterial which one of the coefficients

a, b, c we assume to be zero. Suppose that c = :

ax'^ + by'^ + 2 a' V + 2 b"y' + 2 c"z' -\-F=0, ab^O.

By a change of origin to the point (— a" /a, — b" jb, 0), this

equation becomes

(Ila) ax'"' + by'"' + 2 c"z' + /= 0, a5 =^ 0.

'
* In reducing (1) to the form (I), it was assumed that (1) represents
a surface the method of reduction is quite the same, however, if (1) has
;

no locus. Similarly, the method of reduction in case 2 is always apphca-

lole, both when (1) represents a surface, as assumed, and when it does not.
 : :

602 ANALYTIC GEOMETRY

For (lla), A=— abc"^ and hence A :9^ or A= 0, according

as c" =/= or c" = 0. If c" ^t 0, a change of origin to the point

(0, 0, -//2 c") reduces (Lid) to

ax''^ + by''^ + 2 c'^z'' = 0, ab^ 0.

If c" = 0, (Ila) becomes
ax,ff2

Hence we conclude the following

IfA=^0, (Ila) and hence (1) represents a paraboloid {elliptic

or hyperbolic).
1/^ = 0, (Ila) and hence (1) represents, in the case f=^0,
an elliptic or hyperbolic cylinder, or it has no locus; if f= 0, it

represents two intersecting planes or a line.

B. Two of the coefficients a, b, c vanish. Here again it is

immaterial which two of the three coefficients we assume to

be zero. Suppose that =c= 6 :

aaj'2 + 2a'V + 26 y + 2c'V + i^= 0, a^O.

By a change of origin to the point (— a'' /a, 0, 0) and by a
proper rotation of the axes about the axis of x', this equation
becomes
(116) aa;"2 + 2 dz'' +f=0, a^O.
Here A is always zero. Equation (116), and hence (1), repre-

sents a parabolic cylinder, if d =?t 0; if d = 0, it represents two

parallel planes, a single plane, or has no locus.

Summary. The new types of loci of equations of the form

(1) which have resulted from this investigation, are
i) A point, —
which is a limiting form of an ellipsoid and is
frequently spoken of, in this connection, as a null ellipsoid. It
is to be noted, from the discussion of (I), that the correspond-

ing limiting form of a hyperboloid is a cone.

ii) Two intersecting planes, a line, two parallel planes or a

single plane, all of which are limiting forms of cylinders. We

shall call them degenerate cylinders; two intersecting planes,
a degenerate hyperbolic cylinder ; a line, a degenerate (or null)
 :

TRANSFORMATION OF COORDINATES 603

elliptic cylinder ; two parallel planes, or a single plane, a de-

generate parabolic cylinder.
We can now summarize our results

Theorem 3. An equation of the form (1), if it has a locus,

represents a central quadric, a paraboloid, a cone or a point, or
a cylinder (non-degenerate or degenerate).
The following table shows when each of the four cases
occurs and thus furnishes a means of determining the type of
surface defined by any given equation of the form (1).
604 ANALYTIC GEOMETRY

5. 10 a;2 +2 2/2 + 5 z^ -f- 6 2/2 +2 + 4 ajy

0:2;

- 26 X ~ Uy - ISz- IS = 0.
6. T'x"^ +7 y"^
+ 4:z''' — S yz + S xz — 2 xy
- 42 + 18 - 12 z = 0. 05 2/

7. The surface symmetric in three planes whose

(3) is

normals have the direction components 6, — 3, 2, 2, 6, 3,

— 3, — 2, 6. Determine the precise nature and position of
the surface.
8. The surface (4) is symmetric in three lines whose direc-
tion components are 2,-1,-1, 0, 1, — 1, 1, 1, 1. De-
termine its precise nature and position.

9. The surface (5) is symmetric in two planes whose

normals have the direction components 1, — 1, — 2, 3, 1, 1.

Determine its precise nature and position.
10. Two principal planes of the surface (6) are parallel
respectively to the planes

X — y -\-z = 0, X + y = 0.
Determine the precise nature and position of the surface.

EXERCISES ON CHAPTER XXIV

1. Prove that a curved surface whose equation in spherical
coordinates does not contain r is a cone with the pole as vertex.

2. Show that a curved surface whose equation in spherical

coordinates does not contain ^ is a surface of revolution.
What is its axis?
3. Prove that a curved surface whose equation in cylindri-
cal coordinates does not contain 2: is a cylinder.
4. Show that a curved surface whose equation in cylindri-
cal coordinates does not contain ^ is a surface of revolution.

Transformation of Axes
5. Prove that the transformation to new axes through
whose direction cosines are — ^, f , J, f , — -J, f, f f
, ,
—i is
 TRANSFORMATION OF COORDINATES 605

identical with a rotation of the original axes about the line

X = =
y z through 180°.

Suggestion. Show that the equations of transformation are

equivalent to those connecting the coordinates of two points
symmetric in the line.
6. Find the equations of the transformation which intro-

duces as axes the three mutually perpendicular lines through

the point {xq, ijq, Zq) with the direction angles «i, fii, yi,
«25 /^2, 72: «3j ^3j 73-

Set up the equations of a rotation of the axes which

7.

introduces the plane x -^y -{- z = sls the (x, 2/)-plane.

8. Determine the precise nature of the curve of intersec-

tion of the plane x -\- y -{- z = with the surface
x^ — xy -{- yz — zx ~ X — y — z = 0.
Suggestion. Use the result of Ex. 7.

9. A line of symmetry of the surface

xy -\- yz -\- xz =2
is the line x =y = z. Determine the precise nature of the
surface.

10. A plane of symmetry of the surface

x^ -\-y^ -\-z'^ -\- xy -\- yz — xz — X -]-y — z =

is the plane x — y — 2z = 0. What is the exact nature of the
surface ?
11. Show that, if a quadric surface is symmetric in the ori-
gin, its equation contains no linear term in x, y, z, and con-
versely.

12. A is symmetric in each of the coordinate

quadric surface
planes. Prove that either the equation of the surface is
of the form

or that the surface consists of two coordinate planes. Show

that in the latter case the equation can be reduced to the
606 ANALYTIC GEOMETRY

desired form by rotating two of the axes about the third

through an angle of 45°.

13. A quadric surface is symmetric in two planes which _

are parallel to or identical with two coordinate planes. Show I

"
that either the terms in yz, zx, xy do not appear in the equation
of the surface or the surface itself consists of two planes of
the type described. Prove that in the latter case the terms
in yz, zx, xy can be removed from the equation by rotating two
of the axes about the third through an angle of 45°.

14. Show that, if a quadric surface is symmetric in a coordi-

nate plane or in a plane parallel to a coordinate plane, its
equation contains, in general, at most one of the three terms
in yz, zx, xy. When does the exception occur ?
15. Prove that the conclusion of the previous exercise fol-
lows if the surface is symmetric in a coordinate axis or in a
line parallel to a coordinate axis. When does the exception
occur in this case ?
 ; ; ;; ; ;; ; ; ;; ; ; ; ,

INDEX
Abridged notation, 171 Central conies, 299, 313.
problems in —
189, 287. , Central line of a strain, 304.
Absolute value, 4 Central quadrics, 575, 601.
use of—, 11, 41, 42, 51, 466. Circle, see Contents, Ch. IV
Acoustical property of conies, of. null 67 —
—
,
;

Focal property. equation of in determinant

Affine transformations, 342 form, 394, 398
homogeneous —
348, 355 nine-point 78 —
—
, ,
;

—
;

singular 358. as limit of ellipses, 103, 140

—
,

Anchor ring, 544. parametric representation of

Angle, in the plane, positive sense 119;
for measurement of, 13 common chords of three 170 — ,

slope — 13 equations of
;

in polar coordi- —
—
,

between two lines, 38. nates, 194, 211;

Angle, in space, between two lines, inscribed in a triangle, 283 —•

407, 425, 435, 442 four points on a 395, 398. — ,

;

direction —
of a directed line, 420 Circle in space, 529
— —
; ;

between two planes, 455 tangent line to 545 ,

—
; ;

between a line and a plane, 487. as section of a quadric surface,

Asymptotes, 129, 259. 564.
Asymptotic cone, 543, 552, 557. Cissoid, 214.
Asymptotic planes of a hyperbolic Coefficient of a strain, 309.
cylinder, 533. Colatitude, 584.
Axes of a conic, 88, 102, 125, 127 Columns and rows of a determinant,
equations of 260 — 363;
construction of 324.
,

— , interchanges of — , 376, 380.

Axes of a quadric surface, 533, 549, Compatibility of n+1 equations in
551, 555. n unknowns, 384, 401, 403,
Axes of coordinates, 7, 410. 510.
Auxiliary circle of an ellipse, 120 Compression, simple, cf. Strain,
— of a hyperbola, 143. one-dimensional.
Conchoid, 215.
Bisectors of angles, between two Condition that
lines, 281, 287; the roots of a quadratic equation
-^ between two planes, 522. be equal, 175
Brennpunkt, 97. two sets of numbers be propor-
tional, 385
Cardioid, 201, 209, 215, 286. an equation be homogeneous, 538.
Cassini, ovals of, 215. Condition, in the plane, that
Center of a conic, 102, 125, 247, a point lie on a curve, 19 ;

259; two points be collinear with the

construction of 324. — ,
origin, 25, 402
Center of a quadric surface, 543, three points be collinear, 54,
549, 551. 392 ;
problems, 60-63, 394
607
 ; ; ; ; ; ; ; ; ; ;;; ; ; ; ; ; ; ;;
; ; ; ; ; ; ; ;

608 INDEX
four points lie on a circle, 395, 398 ;
a diameter and a diametral plane
six points lie on a conic, 398 be conjugate, 571
two lines be parallel, 36, 45, 46, three diameters be conjugate,
402; 573;
two lines be perpendicular, 37, three diametral planes be conju-
45, 46 gate, 573
two lines be identical, 46, 402 three numbers be direction cosines,
three lines be concurrent, 53, 189, 422;
393; problems, 54-60, 62, 189, a directed trihedral be right-
394; handed, 443.
a line be orthogonal to a circle, Cones, 536.
76; Confocal conies, 145, 148.
two circles be orthogonal, 77; Confocal parabolas, 95, 146.
a line be tangent to a conic, 183, Confocal paraboloids, 592.
192, 403 Confocal quadrics, 590.
two diameters be conjugate, 291, Conies as sections, of a circular
296; cone, 144
a conic be degenerate, 257, 403 — of a quadric cylinder, 534
a transformation be isogonal, 348. — of a quadric surface, 562.
Condition, in space, that Conies, definition of, 144
a point lie on a surface, 445 equations of —
in polar coordi-
a point lie on a curve, 471 nates, 202, 210, 211;
two points be collinear with the — as loci of equations of the
origin, 441 second degree, 257
three points be collinear, 505 equation of —
through five points,
four points be coplanar, 508 395, 398
five points lie on a sphere, 527 six points on a —
398, 399. ,

two lines be perpendicular, 426, Conies, degenerate, definition of,

435; 257;
two lines be parallel, 420, 427, equations of —
191, 244, 253,
,

430, 432, 436 260, 402

two lines intersect, 512 examples of — 237, 246, 254, 259
— through
,
;

three lines be parallel to a plane, five points, 396, 397

440; condition for — 257, 402
— as intersections
,
;

the normals to four planes be with quadric

parallel to a plane, 508 surfaces, 256, 257, 563, 569.
three lines be mutually perpendic- Conies, similar and similarly placed,
ular, 594 260;
. a line lie in a plane, 506 —— as sections of two quadric
a line be tangent to a quadric, surfaces, 535, 563, 580
567; as parallel sections of a quad-
two planes be perpendicular, par- ric surface, 550, 553, 555, 562-
allel, identical, 455 564, 580.
three planes pass through a line, Conjugate diameters and diametral
504; planes, 571-574, 579.
four planes pass through a point, Conjugate diameters of a conic,
510; see Contents, Ch. XIV.
a plane be orthogonal to a sphere, Conjugate diameters of a quadric
545; surface, 572-574, 579.
a line be orthogonal to a sphere, Conjugate diametral planes, 572-
546: 574, 579.
two spheres be orthogonal, 546 Conjugate hyperbolas, 141, 296.
 ; ; ; ; ; ;; ; ; ; , ; ; ; ; , ;

INDEX 609

Conjugate hyperbolic cylinders, 533. applications of — to plane ana-

Conjugate hyperboloids, 551 lytic geometry, 391-399, 401-
— of revolution, 542 403;
sections of 563. — ,
;

applications of —
to solid ana-
Construction, geometrical, of a pa- lytic geometry, 438, 440, 443,
rabola, 88 453, 463, 467, 477, 489, 496,
— of an ellipse, 104, 114, 119; 504, 507-511, 512-515, 517,
— of a hyperbola, 124, 138, 143 518. 526, 594, 599.
—
;

of center, axes, foci, tangents Diagonals of a determinant, 363.

of a conic, 324, 325 Diameters of a conic, see Contents,
— of poles and polars, 319, 320, Ch. XIV.
325. Diameters of a quadric surface,
Constructions, mechanical, of a 569-574, 579.
parabola, 147 Diametral planes of a quadric sur-'

— of an ellipse, 101, 147 face, 570-574, 579.

— of a hyperbola, 147. Diodes, cissoid of, 214.
Continued equaHty, cf. EquaUty. Directed line-segments, cf. Line-
Coordinate planes, 410. segments.
Coordinates, rectangular, 7, 14 in Directed trihedrals, 430-432
space, 409, 411, 589;
;

rights-handed and left-handed —

polar — —193 , ;
443.
Direction angles of a directed line,
spherical 584, 590, 604
,

—
cylindrical 587, 589, 604 ,
420.
—
ellipsoidal 591. ,
Direction components, see Contents,
Cramer's rule, 381, 464. Ch. XVIII.
Cubic equations, graphs of, 85, Direction cosines, see Contents, Ch.
331, 353, 354. XVIII
Cubic, twisted, 492, 494. — of three mutually perpendic-
Curves, in the plane, definition of, ular lines, 594.
19; Directrices of conies, 88, 116, 137.
plotting of—, 19, 86, 331, 353; Directrix of a cylinder, 532.
in polar coordinates, 198 Discriminant, of a quadratic equa-
symmetry of — , 84-87; in polar
—
tion in X, 175, 402 ;

coordinates, 201. of general equation of second

Curves, in space, definition of, 471 degree in x, y, 248, 402, 403.
parametric representation of — Distance, between two points, 10;
490; in polar coordinates, 210; in

Cylinders, equations of special, 444, space, 414

446 — of a point from a line, 41 ; in

definition and equations of gen- space, 514;

eral —
532. •
— between two parallel lines, 51
,

Cylindrical coordinates, 587, 589, 604. — of a point from a plane, 460

— between two planes,
parallel

Deformation of an elastic body, 357. 466;

Degenerate conies, cf. Conies. — between two in space,
lines

Descartes, 1. 515.
Determinant of a transformation, Division of a line-segment, 17, 416;
358. harmonic — , 309, 359, 575.
Determinants, see Contents, Ch.
XVI; Eccentric angle, of an ellipse, 120,

apphcations of — to linear equa- 306;

—
tions, 381-391,401,403; of a hyperbola, 143.
 ;; ; ;; ; ; ; ; ; ; ; ; ; ; ; ; ,

610 INDEX
Eccentricity of a conic, 102, 116, — of particular transformations,
128, 137, 140. 349;
Element of a determinant, 363. exercises in — , 342, 351, 357.
Ellipse, see Contents, Ch. VII Figures in plane representing space
equations of — not in normal relations, 411.
form, 114, 140; Focal chords, 95, 98, 212.
equations of —
in polar coordi- Focal property of conies, 96, 108,
nates, 203, 211 134, 324.
— as locus of equation of second Focal radii, 88, 106, 128, 212.
degree, 239, 241, 246, 249; Focus, 88, 101, 124;
null or degenerate 245, 256 — origin of name 97 — ,

—
, ;

area of —
341.,
construction of 324. ,

Ellipsoid, 548
sections of — , 548, 550, 563,
Generator, cf. Ruling.
564-566, 569 Graphs of equations, in rectangular
volume of 578 — ,
coordinates, 18-20, 21, 83, 86,
similar —
581 ,
87, 331, 353, 354;
null — 602.
,
— in polar coordinates, 198.
Ellipsoids of revolution, 541, 548.
Elongation, simple, cf. Strain, one-
dimensional. Harmonic division, 309, 575;
Equalitv, continued, representing a — unchanged by an aflfine trans-
line in space, 473, 479, 482 formation, 359.
reduction of —
to normal form, 480 Helix, 491.
—
;

equivalent to representation Homogeneity, 348, 537, 538.

by projecting planes, 502. Homogeneous linear equations, 387,
Equation, general, of the second 401.
degree in x, y, see Contents, Hyperbola, see Contents, Ch. VIII
Ch. XII. equations of not in normal —
Equation, general, of the second de- form, 140
gree in X, y, z, 599 equations of in polar coordi-—
reduction of —
in special cases, nates, 203, 211;
592, 595-597 — as locus of equations of the
reduction of —
in general case, second degree, 239, 241, 245,
597, 600-603, 605. 248, 250
Equation of a curve, 20. — with foci on the axis of y,
Equation of a surface, 445. 141, 221.
Equations, linear, cf. Linear equa- Hyperboloid, biparted or of two
tions. sheets, 551
Equations, simultaneous, of a space sections of — . 551, 553, 561-563,
curve, 472. 565;
Equation u-\-kv = 0, 165, 188-190; similar ^

— . -581.

in space, 520. Hyperboloid, unparted or of one

Equation uv = 0, 173, 190; in space, sheet, 550;
521. sections of —
550, 553, 561- ,

Equiangular hyperbola, cf. Rec- 563, 565, 566

tangular hyperbola. as^inptotic cone of 551, 557 — ,

Equiangiilar transformations, cf. rulings of —

555-559, 568, 581-
,

Isogonal transformations. 583;

parametric representation of —
Factorization of a transformation, 559;
343, 355 similar — , 581.
 ;; ;; ; ;; ; ; ; ; ; ; ; ;; ; ; ; ; ,

INDEX 611

Hyperboloids of revolution, 541, Line, in space, see Contents, Chs.

550. 551. XVIII, XX, XXI.
Line-segments, see Contents, Intro-
Identical transformation, 342. duction, Chs. I, XVII.
Inequalities, loci of, 277, 279, 522. Linear combination, of two lines,
Infinity, 13. 165, 168, 169
Initial ray, 193. — of two curves, 167, 168, 170;
;

Intercepts, of a line, 33 — of two planes, 498, 504, 506

— of a plane, 450.
;

— of two surfaces, 520

Intersection, of two lines, 22 — of three planes, 521
— of two curves, 23 — of of numbers, 400.
sets
— of three planes, 463
;

Linear equation, in 31 x, y,
— of two planes, 470, 476
;

— in 448.
X, y, z,

— of a line and a plane, 488, 492 Linear equation, simultaneous, see

— of a plane and a sphere, 529 Contents, Ch. XVI, §§8-10;
— of two spheres, 530 also 401, 403, 464, 510.
— of a curve and a surface, 488, Loci of inequaUties, 277, 279, 522.
492; Loci problems in plane, see Con-
— of two surfaces, 472 tents, Chs. V, XIII
—
— of three surfaces, 463. further exercises in 100, 123, ,

Invariants of general equation of 152, 214, 328;

second degree, 257, 599. — in polar coordinates, 214.
Inverse of a transformation, 335. Loci problems in space, 497, 546,
Involutory transformations, 337. 582.
Isogonal transformations, 336, 347, Locus of an equation, 19 in space, ;

359. 445
— of two simultaneous equations,
Kepler, 97. 471.
Longitude, 584.
Latus rectum, 94, 112, 135.
Lemniscate, 201, 207, 215, 286. Mid-point of a line-segment, 16,

Limapon, 202, 215. 415.

Line, equations of, Minimum ellipse of an unparted
perpendicular to a given plane, hyperboloid, 550.
483; Minor of a determinant, 367 ;

parallel to two planes, 484 use of —

to evaluate a determi-
parallel to a given line, 485 nant, 367, 370, 380.
perpendicular to two lines, 485.
Line, in plane, see Contents, Ch. Nappe of a cone, 537.
Nicomedes, conchoid of, 215.
equation of — in polar coordi- Nil-segment, 5.

nates, 196, 210; Normal to an arbitrary curve, 158

equation of — in normal form, exercises in finding equations of
286, 287 — 160, 165.
,

equations of — determinant
in Normal, to a parabola, 95
form, 391, 393, 401 — to an ellipse, 111, 112;
three — through 'a point, 53, — to a hyperbola, 135.
189, 392 Normal, to a plane, 447
— —
;

problems three — through a

in direction components of •

point, 54-60, 62, 189, 394; 449;

single equation for two — 173, ,
— to three planes parallel to a
190, 191. plane, 463
 ; ; ;;; ; ; ;; ;; ;;; ; , ; ; ; ; ; ; ; ,

612 INDEX
— to four planes parallel to a — of a helix, 491
•

plane, 508. — of a twisted cubic, 492

— of an unparted hyperboloid,
Octants, 411. 559;
Optical property of conies, cf. Focal
— a hyperbolic paraboloid,
of
property. 560.
Orbits of planets and comets, 120, Pascal, HmaQon of, 215.
211. • Pencil, of lines, 169, 172, 188;
Origin of rectangular coordinates, — of curves, 170, 172.
7,410; Plane, see Contents, Chs- XIX, XX,
— of polar coordinates, 193. XXI.
Orthogonality, of a line and a circle, Plane, equation of,
76; parallel or perpendicular to a
— of two 76
circles, given plane, 456, 466, 467
— of two confocal conies, 95, perpendicular to a given line,
146; 483;
— of a sphere and a plane, 545 parallel to two Hues, 484, 496;
— of a sphere and a 546 line, through a hne and a point, 499,
— of two spheres, 546 504;
— of confocal quadrics, 591.
;

through a line parallel to a second

Ovals of Cassini, 215. line, 513
determined by two non-skew lines,
Parabola, see Contents, Ch. VI 513.
equations of —
not in normal Planes, three through a line, etc., cf.
form, 92, 97, 140, 230 Condition, in space, that.
equations of —
in polar coordi- Plucker, abridged notation of, 171.
nates, 203, 210, 211 Points, three on a line, etc., cf.
equations of —
through four Condition, in the plane, that.
points, 397, 399 Polar coordinates, see Contents,
eccentricity of 140 — ,
Ch. X
—
— as limit of ellipses, 118 rotation of prime direction of
— as limit of hyperbolas, 139 212;
—
— as locus of equation of the sec- pole of in an arbitrary point,
ond degree, 239, 254 213.
confocal — 95, 146.
,
Polar lines with respect to a quad-
Paraboloid, elliptic, 553 ric svu"face, 576.

sections of —
553, 555, 563, 566
,
Pole of polar coordinates, 193. ;

similar —
581.,
Poles and polars, with respect to a
Paraboloid, hyperbolic, 553 conic, see Contents, Ch. XIV
sections of —
554, 555, 564, 580;
,
with respect to a degenerate —
ruhngs of — , 559-561, 568, 582, conic, 402
— with respect to a quadric
583;
directrix planes of — 559 surface, 575.
—
,

parametric representation of Prime direction, 193,

560; Prime meridian, 584.
similar —
581. ,
Principal planes of a quadric sur-
Paraboloid of revolution, 543, 553. face, 549, 551, 555.
Parabola, semi-cubical, 231, 285. Principal sections of a quadric sur-
Parametric representation, of an face, 549, 551, 555.
ellipse, 119, 306; Product of two transformations,
— of a hyperbola, 143, 308 338;
—
— of a line in space, 490 exercises in , 341, 342, 358.
 ; ; ; ; ; ; ; ; ; , ; ;;; ;; ; ; ;; ; ; ; ; ,

INDEX 613

Projecting planes of a line, 501, Rotations of the plane, 332.

503. Rows and columns of a determinant,
Projection, of a point on a line, 5, 363;
405; interchanges of 376, 380. — ,

—of a directed line-segment on a Rulings, of a cylinder, 532

line, 5, 405, 408, 441 of a cone, 536 —
— of a broken line on a line, 5, of an unparted hyperboloid, —
406; 555-559, 568, 581-583
— of a directed line-segment on of a hyperbolic paraboloid, —
the axes, 9, 413 559-561, 568, 582, 583.
—of a point on a plane, 406
;

—of a line on a plane, 406, 501- Sections, of quadric cylinders, 533

503; of quadric cones, 144, 539, —
— of an area on a plane, 516, 561-563, 565;
517; of a quadric surface, defined, —
— of a plane curve on a plane, 549;
534. parallel 549, 561 —
——
,

circular 564 ,
;

Quadrants, 8. cf. also Ellipsoid, etc.

Quadratic equation, discriminant of, Shears, simple, 351

175, 402 factorization of 355, 359. — ,

sum of roots of 274. —

Shrinkings of the plane, 334, 337.
,

Quadric cones, 539 Similar and similarly placed conies,

sections of —
144, 539, 561-563,
, cf. Conies.

565. Similarity, of two parabolas, 89

Quadric cylinders, 532 of two ellipses, 103, 121 —
sections of 533 — of two hyperbolas, 128, 149 —
— —
,

degenerate 602. , of two quadric surfaces, 581. •

Quadric surfaces, see Contents, Ch. Similitude, transformations of, 334,

- XXIII 358.
confocal —590, 592 Slope angle, 13.
—
,

of revolution, 540 Slope of a curve, definition of,

—
;

similar 581. 154;

,

general method of finding 154- — ,

Radical plane of two spheres, 530. 158, 160-163;

Radius vector, 193, 584. exercises in — 159, 164, 188. ,

Ray, equation of, in polar coordi- Slope, of a 12 line,

nates, 196. — a of 72 circle,

Rectangular hyperbola, 132 — a parabola, 94, 161

of
problems concerning 151, 152, — — of an 162
ellipse,
— of a hyperbola, 163
,

157;
equation of — in polar coordinates, — of a general conic, 188.
207; Sphere, see Contents, Ch. XXII,
equation of — in form 2xy = a^, §§1-5;
221; null —
524 ,

—
;

more general equations of — further problems concerning

239, 258, 259. 544-546
Reflection in the origin, 342. — transformed by one-dimen-
Reflections in the axes, 336. sional strains, 578, 579.
Rigid motion of the plane, 358. Spherical coordinates, 584, 590, 604.
Rotation of axes, 219, 239, 241; Spiral, of Archimedes, 202
—
;

in space, 593. , hyperbolic, 202.

 ; ; ;; ; ; ; ; ;; ;

614 INDEX
Square array of a determinant, 361, general method of finding slope
364. of —154-158, 160-163
,

Square root sign, 11. exercises in finding equations of

Strains, see Contents, Ch. XV. — , 158, 160, 165.
Strains, one-dimensional, 304, 307, Term of a determinant,
366
309, 337; in space, 577. determination of sign of
;

— , 365,
Strains, homogeneous, 357. 366, 378.
Strains, simple, 357. Tetrahedron, center of gravity of,
Stretchings of the plane, 334, 337. 419;
Supplemental chords, 327. volume of — , 518.
Surface, definition of, 444 Torus, 544.
symmetry of 468. — ,
;

»
Transformation of coordinates, polar
Surfaces of revolution, 540. to rectangular, 206, 214
Symmetry, in a line, 83 — spherical to rectangular, 585
;

— in a point, 84 —
,
;

cylindrical to rectangular, 587.

— of curves, 84-87 in polar co- ; ,

; Transformation of rectangular co-

ordinates, 201 ordinates, see Contents, Chs.
algebraic — 267 XI, XXIV, §§ 5, 6.
— of surfaces, 468
,

Transformation to parallel axes,

— of quadric surfaces, 600, 605, ;

216, 235 in space, 592. ;

606. Transformations of similitude, 334.

Transformations of the plane, see
Tangent at a given point, to a Contents, Ch. XV.
69
circle, Translations, 330.
— to a parabola, 93, 164; Triangle, area of, 43 in determi-
—
;

to an ellipse, 111, 163 nant form, 401 in space, 517;

— —
;
;

to a hyperbola, 133, 164, 180; medians of 54, 56, 419

—
,

to a general conic, 188 altitudes of —

•
58
—
,
;

construction of 324. , perpendicular bisectors of sides

Tangent line to a quadric surface, of — , 60
567. circle — 74
circumscribing ,

Tangent plane, to a sphere, 527 circle inscribed in — 283 ;

—
,

to a quadric surface, 568, 580. nine-point — 78 circle of ,

;

bisectors of angles of — 283, 287.

;

Tangent to a conic, with a given ,

slope, 174, 179; Triply orthogonal systems of sur-

— from an external point, 185 faces, 589-592.
construction of the latter, 325
condition that a line be — , 183, Umbilics, 580.
192, 403
common —185, 191.
, Vertex of a cone, 536.
Tangent to an arbitrary curve, Vertex of a paraboloid, 555.
definition of, 154, 176; Vertices of a conic, 88, 102, 125.
 BOSTON PUBLIC LIBRARY

3 9999 05987 596 1

•n:HJ:iHti'!:r''

.,1, 1 - -^ - ji i»j