Problems in Geometry - Modenov

rr. c .
MOflEi-iOB
3ARAHW nO TEOMETPHH
MOCKBA «H A Y K A *
PS. MODENOV
PROBLEMS
IN GEOMETRY
Translated from the Russian
by George Yankovsky
MIR PUBLISHERS • MOSCOW

First published 1981
Revised from the 1979 Russian edition
H a cmeAUucKOM nabuee
© TjiaBHaA pegaKiuiH $H3HKo-MaTeMaTnqecKoft JurrepaTypu

HsnaTejibCTBa «HayKa», 1979
© English translation, Mir Publishers, 1981
CONTENTS
PREFACE 7
CHAPTER I. VECTOR ALGEBRA

Sec. 1. Vectors in the plane (solved p r o b le m s )................................................ 11
Sec. 2. Vectors in space (solved p r o b lem s)........................................................ 14
Sec. 3. Vectors in the plane and in space (problems with hints and
a n sw ers)........................................................................................................... 30
CHAPTER II. ANALYTIC GEOMETRY

Sec. 1. Application o f analytic geometry(solved p ro b lem s).......................... 44
Sec. 2. Application of analytic geometry(problems with hints and answers) 66
1. Plane g e o m e t r y ...................................................................................... 66
2. Solid g e o m e t r y ....................................................................................... 78
CHAPTER III. THE USE OF COMPLEX NUMBERS

IN PLANE GEOMETRY
Sec. 1. Solved p r o b le m s.......................................................................................... 82
Sec. 2. Problems with hints and a n s w e r s ....................................................... 253
CHAPTER IV. INVERSION

Sec. 1. Iversion defined. Properties of i n v e r s i o n .......................................... 281
Sec. 2. Problems involving inversion ................................................................ 285
Sec. 3. Mapping of regions under in v e r s io n ................................................... 297
Sec. 4. Mechanical inversors: the Peaucelliercell and the Hart cell . . 308
Sec. 5. The geometry o f M a s c h e r o n i............................................................... 309
Sec. 6. Inversion of s p a c e .................................................................................... 313
CHAPTER V. BASIC DEFINITIONS, THEOREMS AND

FORMULAS
Sec. 1. Determinants o f order three 334
Sec. 2. Vector algebra . . 373
6 Contents
Sec. 3. Analytic g e o m e t r y ..................................................................................... 347

Sec. 4. Complex numbers ..................................................................................... 377
LIST OF SYMBOLS .............................................................................................. 387
APPENDIX. LIST OF BASIC FORMULAS FOR REFERENCES 390
B IB LIO G R A PH Y ............................................................................... 394
NAME IN D E X ........................................................................................ 395
SUBJECT IN D E X ....................................................................................396
PREFACE
This text offers certain general methods of solving problems in elemen

tary geometry and is designed for teachers of mathematics in secondary
schools and also for senior students.
The present text includes material that goes beyond the scope of mathe
matics curricula for secondary schools (the use of complex numbers in
plane geometry, inversion, pencils of circles and others).
The book consists of five chapters. The first four chapters deal with the
application of vector algebra, analytic geometry, complex numbers and
the inversion transformation to geometric problems. Chapter V contains
a list of the basic definitions and formulas used in the first four chapters.
Before starting a new chapter, the reader is advised to refresh his memory
with the appropriate material of Chapter V. Some of the derivations of
formulas given in Chapter V are familiar to senior students of secondary
school. More detailed theoretical material can be found in the bibliography
at the end of the book.
I wish here to remark on a supplement to vector algebra that was brought
to my attention in 1930 by Professor Ya. S. Dubnov, my teacher at Moscow
State University. It is that vector algebra in the plane has not been deve
loped to the point that vector algebra in space has, and in order to remedy
this situation in an oriented plane it is necessary to introduce the rotation
of a vector through an angle of +n/2 (designated [a]) and also a pseudo
scalar (or cross) product a x b [or (a, b)] of a vector a by a vector b. Note
that the linear vector function Ax of a vector argument x possessing the
property that ^ x j_ x for any vector x has the form Ax — X[x](X is an
arbitrary number) in the plane, and Ax -- [a, x] (a is an arbitrary vector)
in space. The cross product of vectors in the plane and in space may be
defined as a polylinear scalar function (of two vectors in the plane and of
three vectors in space) which is antisymmetric with respect to any pair of
vectors — in the plane we have
A(x, y) = —A(y, x);
8 Preface
in space we have
A(x, y,z) = —A (y, x, z), A(x, y, z) = - A (z , y, x),
A(x, y, z) = —^(x, z, y)
— and is normed (that is, it becomes +1 for some base).
This product may be defined as the result of two operations (in the
plane and in space)
(a, b) = [a] . b, (a, b, c) = [a, b] . c.
Although a free vector in geometry constitutes the class of all equiva
lent directed line segments, I will permit myself, in this book (in accord
ance with a very solid tradition) to identify a vector and a directed line
segment as equal (as, for example, in arithmetic, where one regards as
equal the fractions pjq and npjnq9 where /?, q, n are natural numbers).
For this reason, in this text, two directed line segments that are collinear,
have the same length, and are in the same direction will be termed equi
valent or equal.
The idea of using complex numbers in plane geometry came to me in
connection with some very interesting lectures on the theory of analytic
functions delivered at Moscow University by Professor A. I. Markushe-
vich, and also with a book on that subject by Markushevich. Also, since
the 1940 s, papers have appeared regularly in mathematical journals in
many countries illustrating how the use of complex numbers in plane
geometry makes for rather simple solutions to complicated problems by
relating the solutions to basic geometric transformations that are nor
mally studied in secondary school (motion, the similarity transformation,
circular transformations, including inversion).
A book by R. Deaux [2] appeared in France devoted specially to the
problems taken up in Chapter III of this book. Since this methodology
is not all represented in Soviet textbooks, I have given detailed explana
tions and calculations of the procedures. In this text I have made use of
the work of R. Deaux, R. Blanchard, Gourmagschieg, V. Jebeau and
others.
I believe that the contents of Chapter III is added proof of how much
elementary mathematics loses if complex numbers are not brought into
the picture. A consideration of the most elementary functions of a complex
variable,
z a i± b _ ^ a d _ bc # z' = QZ ° {ad — be ^ 0),

az + d cz + d
zf = az + b(a # 0), z' = az + b(a ^ 0),
embraces the isometric transformations of the first and second kind

(zf = az + b, z' = az + b9 where \a\ = 1), similarity transformations of
Preface 9
the first and second kind (z' = az + b, z' = az + 6, a ^ 0) and circular

transformations (the case of a linear fractional function; in particular,
the inversion z' = a/z).
Chapter IV gives a survery of the properties of inversion of a plane and
space and various applications (inversors, the geometry of Mascheroni,
and the mapping of regions under inversion). In particular, detailed con
sideration is given to various stereographic projections of a sphere onto
a plane and the construction of conformal maps of a spectrum of meri
dians and parallels of the sphere.
The final chapter, Chapter V, contains a list of basic definitions, for
mulas and the bibliography. The bibliography contains books in which
the reader will find proofs of the formulas used in this text; they include
textbooks on vector algebra, analytic geometry, the theory of geometric
transformation^, and the theory of functions of a complex variable.
The general methods for solving geometric problems described in this
text are closely interrelated: it will be recalled that vector algebra is close
ly related with analytic geometry. The basic formulas used in Chapter III
are derived on the basis of facts taken from analytic geometry; the linear
fractional function of a complex variable contains within it the inversion
transformation; the inversion transformation can be reliably studied by
the methods of analytic geometry, and so forth.
I would like to point out that the drawing on the cover of the book
(it is the same as that in Fig. 114) is a copy of a photograph of a model
that I constructed to illustrate the stereographic projection of a sphere
onto a plane under which the parallels and meridians pass into a hyper
bolic pencil of circles and an associated elliptical pencil of circles. Figures
107 and 108 were done in the same manner.
During the writing of this text I received valuable advice from Professor
V. A. Ilin and Corresponding Member of the USSR Academy of Sciences
S. V. Yablonsky, to whom I express my deep gratitude. Very profound
and valuable advice was obtained from the reviewer of the Nauka Publish
ing House; practically all his suggestions were incorporated in the final
version of the manuscript.
It goes without saying that the general methods of solving elementary-
geometry problems given in this text do not exhaust the range of such
methods. For instance, mention may be made of a very powerful analytic
method for applying trilinear coordinates in the plane, and tetrahedral
coordinates in space (the trilinear coordinates of a point on a projective-Eu-
clidean plane are the projective coordinates of proper points of such a
plane, provided that all four fundamental points of the projective system
10 Preface
of coordinates are also proper points; the same goes for space as well).
The limited scope of this book did not allow for the inclusion of that
method. And there are of course other general methods, which, unfor
tunately, have not been discussed in our textbooks or teaching literature
(for example, synthetic methods of solving problems with the use of iso
metric, similarity, affine and projective transformations). However, I am
sure that this situation will be remedied in time.
P. S. Modenov
The present edition was prepared after the author died. The material
of the book has been re-examined and brought into accord with generally
accepted terminology and notation. A small number of inaccuracies in
the Russian edition have been corrected and the bibliography has been
expanded. •
Chapter I
VECTOR ALGEBRA
Sec. 1. Vectors in the plane (solved problems)
Problem 1. Given the angles B, C of A ABC. Find /_— >
AB(AB + AC) AB2 + A B -A C
cos <p
c yc2 + b2 + 26c cos A
\AB\ \AB + ~AC\
c + 6 cos A
]fb2 + c2 + 2be cos A
and since 6 : c = sin B : sin C, it follows that
sin C + sin B sin
cos
ysin2i? + sin2 C + 2 sin 2? sin C cos A
Problem 2. Given the angles A, B, C of A ABC.
Let M be the midpoint of segment AB, and let Z> be the foot of the bisector
of /_C. Find the ratio (CDM): (ABC) and also (p = /_ DCM.
Solution.
ah + 6a a + b
CD = -------- 9
a+ b 2
where a = CB, b == CA. Consequently

(<tb + bz, a + b)
(C D M )= — (CD, CM)
4(a + b)
(b - a) (a, b) (a - b) (ABC)
4(a + b) 2(a + b)
whence
(CDN) _ a - b
(ABC) ~ 2(a + b)
12 Problems in Geometry
which can also be written as

(CDM) sin A — sin B
{ABC) 2(sin A + sini?)
Furthermore, since CD\\{ab + fca), CA/|j(a + b), it follows that

{ab + ba) (a + b)
►
s tp = --------------------------
| ab + fta | | a + b|
_ ab2+ ba2 + ab(a -f- b) cos C
~ Y2a2b2 + 2a2b2 cos C f a 2 + b2 + lab cos7c
(a + b) cos (C/2) _ (sin A + sin B) cos (C/2)
1fa2 + b2 + lab cos C f sin2 A + sin2 B + 2 sin A sin B cos C
_ {ab + fea, a + b) _ {b — a) (a, b)
|ab + 6aj |a + b| ab /2(1 + cosC) fa 2 + fc2 + 2 cos C
(sin B — sin A) sin (C/2)
l/sin2 A + sin2 B + 2 sin A sin B cos C
Problem 3. Given the interior angles A, B, C of A ABC; M is the mid
point of segment BC, N is the foot of the altitude dropped from point C
to side AB, and 0 is the point of intersection of the straight lines AM and
CN. Find cos cp, where = /_AOC.
Solution. Orientate the plane with the base a, b, where a = CB, b = CA.
Then C/V | j [a — b]. Indeed, the vector [a — b] = [BA] is perpendicular
to the straight line AB and forms acute angles with the vectors a and b
since
[a — b] a = —[b] a = (a, b) > 0,
[a — b] b - [a] b - (a, b) > 0.
Furthermore, AM }} — — b a — 2b.
The desired /_(p is the angle between the vectors AM and CN;
consequently,
cos c p - f r - bH a ~ 2 b ) _ —2[a]b — [b]a
I[a — b]| |a — 2b| a - b| | a - 2b|
_ - 2 (a, b) + (a, b) _ (a, b)
|a — b| |a — 2b| | a - b| |a — 2b|
Vector Algebra 13
ab sin C
Ya2 + b2 — lab cos C Y&2 + 462 —4ab cos C
sin A sin B sin C
]/sin2/4 4-sin2# —2sin A sinB cosC /sin2/t 4-4sin2# —4sim4 sin# cos C
Problem 4. Given vectors a = CB, b ™ CA. Find the vector x = CO,

where O is the centre of a circle circumscribed about A ABC.
Solution. From the relations
x2 = (a — x)2 = (b — x)2
we find
xa - a2/2, xb = b2/2
and, hence, by the Gibbs formula
b2
x = xa •a* + xb •b* = — a* + — b*
2 2
where a*, b* is the reciprocal (or dual) basis of the basis a, b:
[b] [a]
b*
(b, a) (a, b)
Thus,
b2[a] - a2[b]
(», b)
Remark. If we take x in the form x = 2a + /*b, then from the relations
xa = a2j 2, xb = b2/2
we obtain
Xa2 + ju•ab = a2/2,
2-ab +Hb2 = b2l2,
whence
^ _ a2b2 — b2• ab _ a2b2 — a2•ab
~ 2(aW - (ab)2)’ ** ~~ 2(a262 - (ab)2)
and so
x _ 1 a2b2 - b 2- ab & 1 a262 - a 2-ab
~~ 2" a2Z>2 - (ab)2 a T a262 - (ab)2
Problem 5. Given the cross products (a, x) = /?, (b, x) = q of vector x
into the noncollinear vectors a and b. Express the vector x in terms of the
vectors a, b and the numbers p , q.
Solution. Let
. [b] [a]
a* = - ......., b*
(b, a) (a, b)
be the reciprocal basis of a, b. Then
(xa*)a + (xb*)b = ( x ) a + ( - i- J - , x )
V(b, a) ) V(a, b) )
(b,x) a , x) b p b — qa
(b, a) (a, b) (b, a) (a, b)
~ (aT b T
Problem 6. Two forces = {2, 3} and F2 — {4, 1} are specified rela
tive to a general Cartesian system of coordinates. Their points of appli
cation are, respectively, A = (1,1) and B = (2, 4). Find the coordinates
of the resultant and the equation of the straight line / containing it.
Solution. The coordinates of the resultant F are 6 and 4. Now let
M(x, y) be an arbitrary point of /. Then the moment of the resultant about
point M is equal to zero. This moment is equal to the sum of the moments
(AL4, Fj) and (MB, F2) of component forces (the cross product of vectors
is distributive).
Since MA = {1 X, 1 — y}, MB - { 2 x, 4 — >’}, it follows that
-\l - x 2 2 —x 4
( m a , f,) = yg \ (MB, F2) = fg
|i -y 3 4 -y 1
and, hence, the equation of the straight line / is
1- x 2 2 -x 4
+ 0
1 ■y 3 1 4 —y 1
or
4x — 6y + 13 0.
Sec. 2. Vectors in space (solved problems)

Problem 1. The plane angles of a trihedral angle OABC are a = /_BOC,
b — /_ COA, c — / AOB. The interior dihedral angles of the given tri
hedral angle are:
A = B(OA)C, B = C(OB)A, C = A(OC)B*
A trihedral angle OA*B*C* that is the reciprocal of the trihedral angle
OABC is a trihedral angle constructed in the following manner: ray OA*
* The symbol B(OA)C is used to denote a dihedral angle with edge OA, in the half
planes o f which are points B and C.
Vector Algebra 15
is perpendicular to the rays OB and OC and forms an acute angle with

ray OA. The rays OB* and OC* are constructed in similar fashion.
Let a*, b*, c* be the plane angles of the trihedral angle OA*B*C* and
let A*, B*, C*, be its interior dihedral angles.
1°. Knowing, a, b, c, find cos A, cos B, cos C.
2°. Prove that a* — n — A, b* — n — B, c* = n — C.
3°. Prove that A* — n — a, B* = n — b, C* =- 7C— c.
4°. Knowing A, B, C, find cos a, cos b, cos c.
5°. Prove that
sin A sin B. sin C A
sin a sin b sin c sin a sin b sin c
where
1 cos b cos c \ ,/2
A = cos b 1 cos a I
cos c cos a 1 /
— 1 + 2 cos a cos b cos c — cos2a cos 2b — cos2c
(this relation is called the theorem o f sines for a trihedral angle OABC)l\
6°. Prove that the sine theorem for the trihedral angle OABC (see
item 5°) may be written in the form
sin A sin B sin C A*
sin a sin b sin c A
where
A* = ]f\ + 2 cos a* cos b* cos c* — cos2 a* — cos2 b* — cos2 c*
= V I — 2 cos A cos B cos C — cos2 A — cos2 B — cos2 C.
Solution. 1°. Let eA, ea, ea be the direction vectors of the rays CM, OB,
OC (ex t t OA, e2 ft OB, e3tfOC). Then the vectors e1, e2, e3 of the reci
procal basis of the basis eA, e2, e3 are the direction vectors of the rays OA*,
OB*, OC*. We assume the vectors eA, e2, e3 to be unit vectors and lay them
off from point O; then their endpoints Ex, E2, will lie on the ray OA,
— »— > —>
OB, OC respectively. Through O draw a plane %perpendicular to the ray OC.
Let E\ and E2 be orthogonal projections of the points Ex and E2 on the
plane n. Then C =- A(OC)B = /_E° OE$. Consider the vectors
e? = OEl, = OE\.
Or for a spherical triangle cut out o f a sphere with centre O by a trihedral angle.
We have
e? = e2 + 2e3, eg = e2 + /*e3.
Forming the scalar product of both sides of each of these relations by
the vector e3, we obtain
0 = cos b + X9 0 — cos a + /z.
so that
eg = ex — e3 cos b, eg = e2 e3 cos a
and consequently
eV-eS _ (ex — e3 cos b) (e2 — e3 cos a)
cos C =
|Cl| ie 2| V (e i “ e3 COS ^ )2 ]f(e 2 ““ e3 C0S fl) 2
cos c — cos b cos a — cos a cos b + cos b cos a
]f I — 2 cos26 + cos2b ][ 1 — 2 cos2a + cos2 a
cos c — cos a cos b
sin a sin b
In similar fashion we calculate cos A and cos B. Thus,
, cos a — cos b cos c
COS A = ---------------------- ;----------------- 9
sin b sin c
„ cos b — cos c cos a
cos B = ---------------------------
sin c sin a
„ cos c — cos a cos b
cos C = --------------------------
sin a sin b
2°. The formulas obtained in item 1° can be rewritten thus:
cos A =
|[eie2]|![e1e3]|
cos B =
|[e2e1]i|[e2e3]|
[e3eiHe3e2]
cos C = --------------- •
l[e3ei]| |[e3e2]j
Note that in this notation the vectors el5 e2, e3 need not necessarily be
regarded as unit vectors because when e1? e2, e3 are replaced respectively
by Xel9 ê2, ve3, where X > 0, it > 0, v > 0, the right-hand members
of these relations remain unchanged. Thus, el5 e2, e3 may be regarded as
— > — > — >
any direction vectors of the rays OA, OB, OC.

Vector Algebra 17
These last relations can be rewritten as

q2• ^3■^1 gl •e2
cos = -------------5 cos B = — —■:— j cos C = -------------- (a)
|e2| |e3| |e3|(e1! le1)je2|
Indeed,
[e3ei] [eie2]
e2-e3 ®1®2®3 e,e„e. [eiealteiej!]
= cos A
|e2| |e3| |Jf? eiL teie2] ]t©ie3] | |[e1e2]|
i c1e2e3 e!e2e3
and similarly for the other two formulas.
Note also the formulas for cos a*, cos b*9 cos c*:
(e2, e3) (e3, e1) (e1, e2)
cos a* = cos b* = cos c’ — • (b>
*21 |e3|
|e*| |e,>| : |e*| |eA|
From formulas (a) and (b) we conclude that
cos A = — cos a*9 cos B = — cos b*9 cos C = — cos c*
and since all the angles A , B, C, a*, fc*, c* lie in the interval (0, n)9it follows
that
a* = n — A, b* = n — B9 c* = 7i — C.
These relations can also be derived geometrically.
3°. Writing down the formulas obtained above,
a2 • a 3 a 3 . p i p i ■p 2
cos A = — -------, cos B= --------------, cos C= ------------
|e2| |e3j !e3j (e1! le1!|e2|
we have for the trihedral angle OA*B*C*
A ik Qjb 63*6] ^ 1 * ^ 2
cos A* = ------------ , cos B* = ------ — cos C* = ------- ,
|e2| |e3| |e3| le^ |ex| ;e2|
and since 'ej = |e2| = je3; = 1, it follows that
cos A* = — cos a9 cos B* = — cos b9 cos C* = — cos c9
whence
A* = n — a, B* = n — b9 C* = n — c.
4°. Applying the formulas of item 1° to the trihedral angle OA*B*C*>
we obtain
cos a* — cos b* cos c*
cos A* =
sin b* sin c*
2 -8 1 0
or
cos(7r — A) — cos (n — B) cos (n — C)
cos (n — a) =
sin (7r — B) sin (k — C)
or
cos ^4 + cos 5 cos C
cos a =
sin 1? sin C
The formulas for cos b and cos c are derived in similar fashion. Thus
cos A + cos B cos C
cos a
sin B sin C
cos 5 + cos C cos
cos b =
sin C sin A
cos C + cos A cos 2?
cos c =
sin A sin
5°. From the formula
cos a — cos b cos c
cos A =
sin b sin c
we get
sin A = VI — cos2^4
__ ^1 + 2 cos a cos &cos c — cos2a —cos2b—cos2c
sin b sin c sin b sin c
whence
sin A A
sin a sin a sin b sin c
sin B . sin C
The ratios — and have the same value. Thus
sin b sin c
sin A sin B sin C A
sin a sin b sin c sin a sin b sin c
6°. Let us write down the sine theorem for the trihedral angle OA*B*C*,
which is the reciprocal of the trihedral angle OABC:
sin A* sin B* sin C* A*
sin a* sin b* sin c* sin a*sin b* sin c*
or (see items 2° and 3°)
sin a sin b sin c
sin A sin B sin C sin A sin B sin C
Vector Algebra 19
where
1 cos b* cos c* 1/2
A* cos b* i cos a*
cos c* cos a* 1
y 1 + 2 cos a* cos b* cos c* — cos2 a* — cos2 b* - cos2 c*
= ][ 1 — 2 cos A cos B cos C — cos2 >4 — cos2 2? — cos2 C.
From the equations
sin A
------- A
------------------- 9
sin a sin a sin b sin c
sin a __ A*
sin A sin A sin B sin C
we obtain the following by thermwise division:
sin2 A __ A sin A sin B sin C A sin3/4
sin2fl A* sin a sin b sin c A* sin3 a
whence
sin A _ A*
sin a A
Similarly,
sin B __ sin C __ A*
sin b sin c A
To summarize,
sin A sin B sin C __ A*
sin a sin b sin c A
Problem 2. Given, in a parallelepiped, the lengths a, b, c of its edges
OA, OB, OC and the plane angles between them:
/_BOC = a, Z COA = 0, Z AOB = y.
1°. Find the length d of the diagonal OD of the parallelepiped.
2°. Find cosines of the angles <pl5 q>2, cp3 formed by the diagonal OD
and the edges OA, OB, OC.
3°. If the coordinates u, v, w of the projections of OD on the axes OA,
—>
OB, OC are also given, then prove that
d = y au + bv + cw.
4°. Express the length d of the diagonal OD in terms of the angles

a, p, y and the coordinates u, v, w of the projections of the directed line
—> — >— >— >
segment OD on the axes OA, OB, OC.
5°. Express the volume V of the given parallelepiped in terms of
a, b, c, a, P, y.
6°. Two rays OP and OQ emanate from point O; the first ray forms
— >— >— >
with the axes OA, OB, OC the angles a,, Pl9 yx\ the second ray forms the
angles a2, p2, y2. Express the cosine of the angle 6 between the rays OP
and OQ in terms of c l , p, y, otl9 pl9 yl9 cc2, p2, y2.
7°. Find the shortest distance between the straight lines OD and AB
if a, b, c, c l, P, y are given.
8°. Find the distance 3 between point D and straight line AB.
Solution. 1°. Consider the vectors a = OA, b = OB, c — OC, d = OD.
Then
d= a + b + c
whence
d == /d* = /(a + b + c)2
— ][a2 + b2 + c2 | 2be cos a | 2ca cos ji + lab cos y.
2°.
ad a(a + b + c)
cos <px =
la! |d| Ia| |d|
a2 + ab cos y + ac cos P a + b cos y + c cos p
ad d
The formulas for cos <p2 and cos cpz are derived in similar fashion. Thus
a + b cos y + c cos p
cos <p1 = -------------- 73
a cos y + b + c cos a
cos <p2 = ----------------------------- r,
d
a cos P + b cos a + c
cos cp3 = ------- ----------------------
d
3°. From the formulas of item 2° we have
u = d cos cp1 = a + b cos y + c cos p,
v = d cos q>2 = a cos y + b + c cos c l,
w = d cos cpz = a cos P + b cos a + c.
Vector Algebra 21
Multiplying both sides of each of these formulas by a, b, c respectively

and adding the resulting formulas termwise, we get
au + bv + cw = a(a + b cos y + c cos ft)
+ b(a cos y + b + c cos a) + c(a cos P + b cos a + c) = d2.
4°. From the formulas of item 2° it follows that
a + b cos y + c cos P = w,
a cos y + ft + c cos a — v,
a cos P + 6 cos a + c = w.
Solving this system for a, 6, c, we obtain
u cos y cos P
0 v 1 cos a »
\v cos a 1
! 1 u cos P
1 i
~ cos y v cos a *
o \ 0 t
I cos p w 1
1 cos y w
cos y 1 v 9
cos P cos a w
where
1 cos y cos P
5= cos y 1 cos a
cos P cos a 1
= ^ 1 + 2 cos a cos P cos y — cos2 a — cos2 p cos* y,
and, consequently, on the basis of the formula
d2 =r=au + bv + cw
we have
u cos y cos P |j 1 u cos p 1 cos y u
ll v 1 cos a + v j cos y v cos a + w cos y 1 v
w cos a 1 ! COS P w 1 cos P cos a w
1 cos y cos p u
cos y 1 cos a v
~S cos P cos a 1 w
u v w 0
so that
1 cos y cos /? u \ 1/2
d= cos y 1 cos a t?
V* cos P cos a 1 w
u v w 0 /
a2 ab ac
5°. F2= (a ,b , c)2 = ba b2 be
ca cb c2
a2 06 cos y ac cos p | cos
0Crt
1
0
ab cos y b2 cos a 1= cos y 1 cos a
ac cos p be cos a c2 | cos P cos a 1
and so
V = abc ]f 6 = abc f 1 + 2 cos a cos P cos y — cos2 a — cos2 p —cos2 y.
6°. Let us consider the vectors p = OP, q = OQ, assuming that

Ipl = Iql = L
Expand the vector p in terms of the basis a, b, c:
P = Pi* +PJ> +Pz c.
Taking the scalar product of both sides of this equation into a, b, c in
succession, we obtain
cos <*1 = Pi + p2 cos y — p3 cos P,
cos & = Pi cos y + P 2 + P 2 cos a,
cos yi = pi cos P + p2 cos a + p2.
whence
cos a2 cos y cos P !
cos Pi 1 cos a »
cos cos a 1
1 cos oti cos P
cos y cos Pi cos a 5
cos P cos yx 1
1 cos y cos <Xi
cos y 1 cos Pi •
cos P cos a cos y2
Vector Algebra 23
Furthermore, let a*, b*, c* be the reciprocal basis of a, b, c. Expand the

vector q in terms of the basis a*, b*, c*:
q = qxa* + q2b* + q2c*.
Taking the scalar product of both sides of this relation by a, b, c in succes
sion, we obtain
qx = cos a 2, q2 = cos /J2, qz = cos y2.
We get
q = cos a2a* + cos P>2b* -|- cos y2c*.
We now find
cos(p, q) - cos 0 = pq - (pxa + p 2b + pzc) foa* + q2b* + q3c*)
—PlVl + + Pdfi!§
I cos clx cos y cos ft cos ax cos P
cos a21cos px 1 cos a + cos p 2 cos y cos px cos a
I | cos yx cos a 1 cos P cos yt 1
1 cos y cos P cos ax
1 cos y cos ax
1 cos y 1 cos a cos px
+ cos y2 cos y 1 cos px
cos P cos a cos yx
I cos P cos a 1 cos yr
cbs a cos p2 cos y2 0
The formula for cos 0 can be written more compactly:
1 cos y cos P cos olx
cos y 1 cos a cos px _ q
cos p cos a 1 cos yi
cos a2 cos p2cos y2 cos 0
7°. The shortest distance d between two noncollinear straight lines
in space is equal to the length of the projection of any line segment MXM29
whose ends lie on the given lines, onto the common perpendicular to
the given straight lines.
Thus, in item 7°
M1 = 0, A/o = By M1M2 = = b.
Now, the direction of the common perpendicular to the straight lines OD
and AB is given by the vector product of the vector a + b + c = OD
into the vector b — a = AB:
[a + b + c, b — a] = 2[a, b] + [c, b] + [a, c].
Thus,
d= |b(2[a, b] + [c, b] + [a, c])l _ JaLb, c|
|2[a,b] + [c,b] + [a,c]|
where (see item 5°)
|a, b, c| = abc /<5,
T = 4[a, b]2 + [c, b]2 + [a, c]2 + 4([a, b]-[c, b]) + 2([c, b]-[a, c])
+ 4([a, b] •[a, c]) = 4a2b2 sin2 y + c2b2 sin2 a + a2c2 sin2 P
+ 4b2ac(cos P — cos a cos 7) + 4a2bc(cos a — cos P cos y)
—2c2ab(cos y — cos a cos /?).
VS
And so d = abc —— •
\t
8°. Suppose e is the direction vector of the straight line / and B is
an arbitrary point in space. Take some point A on I and let AB = a.
Then the distance d from point B to / is found from the formula d =
- |[a, e]|/|e|. Indeed, |[a, e]| - |a| |e| sin cp |e| d.
In particular, if e is the unit vector, then
d = |[a, e]|.
In determining the distance from point D to the straight line AB, note
that the direction vector of that line is equal to a — b, and since AD =
= b + c, it follows, using the formula
that
j l[(» ~ b), (b + c)]| _ | [a, b] + [a, c] + [c, b]j _ Kj j
la - b| la - b| ]fT2
where
7\ = a2b2 sin2y + b2c2 sin2a + c2a2 sin2P
+ 2a2bc(cos a — cos p cos y) + 2b2ca (cos p — cos y cos a)
— 2c2ab (cos y — cos a cos /?),
T2 = a2 + b2 — 2ab cos y.
Problem 3. Given the plane angles /_ BOC = a, /_ COA = b, /_ AOB=
— c of the trihedral angle OABC. The ray / emanates from point O and
— >— >— >
forms with the edges OA, OB, OC of the given trihedral angle the equal
angles cp. Find tan cp.
Vector Algebra 25
Solution. On the edges of the given trihedral angle, choose points A, B, C

so that OA = OB = OC = 1 and then consider the unit vectors a = OA,
b = OB, c = OC.
Let OP be a directed line segment of length 1 such that /_ POA —
= /_ POB = POC = <p. We now consider the unit vector x = OP.
Then
|xa = xb — xc| — cos C l , b*
------ K* = lC’ ®] , c* [“’ bl •
(a, b, c) (a, b, c) (a, b, c)
Raising both sides of (*) to a scalar square, we obtain
1 = cos2 <p(a* + b* + c*)2
whence
T,
1 + tanV = a*2 + b*2 + c*2 + 2b*c* + 2c*a* + 2a*b* =
where
Tx = sin2# + sin2b + sin2c + 2(cos b cos c — cos a)
+ 2(cos c cos a — cos b) + 2(cos a cos b — cos c),
T2 = (a, b, c)2 = 1 + 2 cos a cos b cos c — cos2a — cos2b — cos2c.
Thus
T, Ti - T2
Vdn2<p = 1
t2 t2
Tx — T2 —- sin2a + sin2fe + sin2c + 2(cos b cos c — cos a)
+ 2(cos c cos a — cos b) + 2(cos a cos b — cos c)
— 1 — 2 cos a cos b cos c + cos2# + cos2b + cos2c
= 2[1 —(cos a + cos b + cos c) + (cos b cos c + cos c cos a + cos acosb)
— cos a cos b cos c]
= 2(1 — cos a) (1 — cos b) (1 — cos c) = 16 sin2 — sin2 — sin2 — >
16 sin2 — sin2 — sin2 —

Tx - T 2
tan2 q> =
1 cos b cos c
cos b 1 cos a
cos c cos a 1
The angle q> may be regarded as an acute angle (because if q> is obtuse
the direction of the line segment OP may be reversed).
Thus
A. a . b . c
4 sm — sin — sm —
2 2 2
tan cp = ------------- ---------------
where
1 cos b cos c
A = cos b 1 cos a
cos c cos a 1
= 1 + 2 cos a cos b cos c — cos2a — cos2& — cos2c.
Problem 4. Given in a tetrahedron OABC the lengths of the edges
OA = a, OB = b, OC = c and the plane angles / BOC = a, /_ COA = fi9
/_AOB = y. Let PQ be the common perpendicular to the straight lines OA
and BC (point P lies online OA, point Q on line BC). Find the ratios
J )P _ = x Jsp =
OA BC
Solution. Consider the vectors a = OA, b = OB, c = OC, p OP,

q — OQ, t = 5 0 , s = P 0 . Then
p = 2a, t = fi(c — b).
Furthermore,
p + s —t —b = 0
whence
s = — 2a + p(c — b) + b.
Since the vector s is perpendicular to the vectors a and b — c, we have
as = 0, (b — c) s = 0
a(— 2a + p(c — b) + b) = 0,
(b - c) ( - 2a + /i(c - b) + b) = 0
or
ka2 — /ia(c — b) = ab,
— 2a(c — b) + ju(c — b)2 = — b(c — b).
Vector Algebra 27
From the resulting system of equations that are linear in A and /i, we
obtain these unknowns:
ft cos y(c2 + ft2 — 2be cos a)—(c cos P—b cos y)(c cos a —ft)
A
a c2 sin2/? + ft2 sin2y — 2ftc(cos a — cos P cos y)
— c cos a + ft + cos y(c cos P — b cos y)

=
c2 sin2/? + ft2 sin2y — 2ftc(cos a — cos /? cos y)
Problem 5. We give here a vectorial derivation of the basic formulas
used in the theory of axonometry (axonometric projections).
Suppose Oxy Oy9Oz are three pairwise perpendicular axes. Let us consider
the plane % that intersects these axes in the points A, B9C respectively.
We assume the space to be oriented by the ordered triple of axes Ox, Oy9Oz.
Denote by O' the orthogonal projection of point O on the plane n — ABC9
by al9 a2, a3 the angles of the axes Ox, Oy9Oz with the plane n9 and by
Pl9 P2 >Pa the respective angles BO'C, CO'A, AO'B.
Prove that
1°. cos /?i = — tan a2 tan a3,
cos /?2 = — tan a3 tan ctl9
cos /?3 = — tan olx tan a2.
T sin Pt __= ___sin p2 _ sin p9 _________ 1_______
sin axcos ax sin a2cos a2 sin a3 cos a3 cos ax 00s a2 cos a3
sin 2Pi _ sin 2/?2 sin 2/?3 _ ^ s^n ai s*n a2 s*n a3
cos2 ax cos2 a2 cos2a3 cos2 ax cos2 a2 cos2 a3
4°. There exists a triangle, the lengths of whose sides are proportional
to cos2ax, cos2a2, cos2a3; one of these triangles is a triangle formed
by the feet of the altitudes of the triangle ABC (SchlomilcKs theorem).
Solution. 1°.
OB OC = {OO' + O'B) (OO' + O'C) = OO'2 + O'B-O'C - 0 <*>
since 00'_L O'C, OO' _L O'B, OB J_ OC; consequently, from the relation
(*) we obtain
O'B tan a2- O'C tan a3 -f O'B• O'C cos /?x = 0
whence
cos Pi = — tan a2 tan a3.
The formulas
cos /?2 = — tan a3 tan ax,
cos /?3 = — tan ax tan a2
28 Porblems in Geometry
are derived in similar fashion.
2°. [ M , OC] = [ 0 0 + OB, OO + ~OC]
= [O'O, OC] + [OB, O'O) + [OB, OC]

whence
[O'B, O C] OOf = [OB, 0 C \-0 0 '. (**)

Furthermore,
to 7!, trc j-O O ' - | [ ^ , (TC]\ 0 0 '

= 0 'i? •0 'C sin •0 0 ' = 0 Z? cos a2•OC cos a3•0 0 ' sin Pl9
[0B,^C] 0 0 ' = Oi!'OC’0 0 ' c o s | - - = OB OC-OO' sin «!
and, hence, by virtue of (**) we have

sin & cos a2 cos a3 = sin ax.
Consequently,
sin Pi _ 1
sin a2 cos ax cos ax cos a2 cos a3
In similar fashion, proof is given that
sin p2 __ sin j?3 1
sin a2cos a2 sin a3 cos a3 cos oq cos a2 cos a3
3°. The solution follows from items 1° and 2°.
4°. Rewriting the relations of item 3° in the form
cos2 oq cos2 a2 cos2 a3
sin 2 ^ ----~ -j sin 2 --j sin 2 |j8 3 —
and noting that
2 ( f t - f ) + 2 ( f c - f ) + 2 ( f c - | ) = «,
Pi ~ n
we conclude that there exists a triangle A with angles 2 |
~2 )’
Indeed>
0i > Jt/2, 02 > jr/2, > n/2
Vector Algebra 29
since all plane angles of the trihedral angle OABC are equal to n/2 and
therefore ft, ft, f t are obtuse angles: a sphere with diameter AB passes
through the point O {/_ AOB = 7i/2), the plane ABC intersects the sphere
along a great circle since AB is the diameter of the sphere, the projection O'
of point O on the plane ABC will be inside that circle and, hence, the
angle AO'B = f t is obtuse (similarly n/2 < f t < n, n/2 < f t < n).
From these relations it follows that cos 2ax, cos2a2, cos 2a3are proportional
to the sines of the angles of the triangle A, and so also to the lengths of
its sides.
Finally, we will now prove that one of these triangles is formed by the
feet of the altitudes of AABC. Indeed, ABC is an acute-angled triangle
because the lengths of its sides
AB = ]jOA2 + OB2, BC = ]f(?B2 + O C \ CA = ]fOC2 + OA2
and, hence, for example, AB2 + BC2 > AC2 (angle B is acute) and so forth.
Furthermore, since OA _L OBC, it follows that OA J_ BC and, hence,
on the basis of the theorem of three perpendiculars AO' ± BC and, simi
larly, BO' _L CA, CO' 1 AB, that is O' is the point of intersection of
the altitudes of A A B C . Let Al9 Bx, Cx be the feet of the altitudes of A ABC.
What we then have is that since / ACxO' — /_ ABxO' = n/2, it follows
that points A, Cx, O', Bx lie on one circle (with diameter AO'). From
this it follows that /_ CxftO' — /_ CxAO' (both angles are intercepted
by the arc CjO'). But / C ^ O ' = 71 - B and so Z C1B1Or = % - B.
2
Similarly, the points Bx, O', Ax, C lie on one circle,, / AiftO ' = /_ AxCO'==
n
= ------ B. Thus, BxO' is the bisector of the interior angle Bx of A A ^ C ^
2
Now, from A 0 'B 1C1 we have
and since
Ax + Bx + Ci — n
it follows that
Similarly,
Problem 6. On the base of the polyhedron OACBPQ lies a rectangle

OACB. The edge PQ is parallel to the plane of the rectangle, and the
orthogonal projections Px and Qx of points P and Q on the plane OABC

are such that
PxO = PXA = QXB = QXC
(this kind of polyhedron is called a wedge).
Given the lengths of the sides of the rectangle OACB:
OA — ay OB — b,
the length c of segment PQ and the altitude h = PPX= QQX. Compute
the cosine of the dihedral angle A(CQ) B = <p.
Solution. Introduce a rectangular Cartesian system of coordinates Oxyz>
choosing the directions of the a- and j ’-axes along OA and OB respectively.
The coordinates of the vertices of the wedge in the chosen system are:
O = (0,0,0), A = (a, 0, 0), C = (a, b, 0), B = (0, by0),
(X T -" )-
From this we find the vectors
C A ^ { 0 , - b , 0}ft{0, —1,0} = x,
CQ = | - - • C~ h , AJ t t { - a, c — b, 2 h} = y,
{— a , 0 , 0 } | | { — 1, 0 , 0 } = *.
We also find
[y, x] =
\c - b 2 h\ j2A - a
i - 1 0 | 10 0
fi c — b 2 h {Hi - a
7 "", H “'
a c — b\ 1
1
. ,v
[y, z] =
i 0 o ! ’ jo - 1 1 0 J = *0, ~ 2h’
and, consequently, we have
[y, x]-[y,z] a(c — b)
cos (p =
l[y, *]| l[y. z]| Y4h2 + a2 )[4fi£ + (c — b)2
Sec. 3. Vectors in the plane and in space

(problems with hints and answers)
1. Given two vectors CA = b, CB — a ^ 0. Let P be the orthogonal

projection of point A on the straight line BC. Find the vector CP.
Answer. < ^ - a .
Vector Algebra 3f
2. Four points A, B ,C ,D are located (in the plane or in space) so that

AD ± BC, BD ± CA. Prove that CD _L AB.
Hint. Assuming DA — rl5 DB = r2, DC = r3, we find — r3) = 0,.
r2(r3 — r2) = 0, whence r f a — r2) = 0.
3. An arbitrary point O is joined to the centroid G of A ABC and
a point P is constructed such that OP = 3OG. Let A', B', C be points
symmetric to the point P about the points A, B, C. Prove that O is the
centre of gravity of the four points A', Bf, C', P.
4. Let ABC be an arbitrary triangle. We consider 12 vectors whose
initial points are the centre of an inscribed circle and the centres of escribed
circles, and the end points are the points of contact of circles and the sides
of the triangle. Prove that the sum of these vectors is equal to the vector OH,
where O is the centre of the circle (ABC), and H is the orthocentre of
the triangle ABC.
5. ABC is an arbitrary nondegenerate triangle lying on an oriented plane;
P is an arbitrary point lying in the plane of the triangle. The vectors;
'—>— >— >
PA, PB, PC are coplanar and, hence, are linearly dependent:
otPA + fiPB + yPC = 0

(at least one of the numbers a, /?, y is different from zero). Prove that
a: fi: y = (PBC):(PCA): (PAB).
Hint. Project the zero vector ctPA + fiPB + yPC onto the straight
line BC in the direction of the straight line PA. Let Px be the point into
which are projected the points P and A. Then
PP^B + yP^C - 0 .
6 . Let / be the centre of a circle (/) inscribed in a triangle ABC;
let D, E, F be the points of contact of the circle (I) with the sides BC,
CA, AB; let P be an arbitrary point lying in the plane of A ABC;
let Pa, Ph, Pc be orthogonal projections of point P on the sides BC, CA, AB.
Prove that the circle passing through the centroids of the triangles EPaFy
FPbD, DPCE has a diameter equal to IP.
Hint. Let G be the centroid of the triangle DEF, and let Ga, Gb, Gc be
the centroids of the triangles EPaF, FPhD, DPCE; then
/? = j (7 i+ 7 ? + 7 ? )
= (IE + JF + ID + DPJ == IG DP.

3
Thus, the point Ga is constructed in the following manner: from point G

lay off a directed line segment equal to one third of the projection IP on
the side BC. From this it immediately follows that the points Ga, Gh, Gc lie
on the circle whose diameter is obtained by laying off from point G a
vector determined by the directed segment * IP.
7. ABC is an arbitrary triangle; A l 9 B l 9 C1 are the midpoints of its sides
BC, CA, AB; A2, B2, C2 are the feet of its altitudes; O is the centre of the
circle (ABC); G is the centroid of A ABC (the point of intersection of
its medians); H is the orthocentre (the point of intersection of the altitudes
of A ABC); A 3 ,B 3, C3 are the midpoints of the segments AH,BH, CH;
A a,B a,C 4 are the second points of intersection of the circle (ABC) with
the altitudes/!//, BH, CH (Fig. 1). Prove that:
1°. The points O, G, H are collinear and OG —- — GH.
2
2°. The points A l 9 B l 9 Cl 9 A 29 B2, C2, A 3 , B3 , C3 lie on one circle called
the Euler circle of A ABC or the nine-point circle (0 9) of A ABC.
3°. The points Aa, B4, C4 are symmetric to the orthocentre H with
respect to the straight lines BC, CA, AB.
Hint. This problem may be solved vectorially. First prove that if O
is the centre of the circle (O) circumscribed about A ABC, then OA 4-
+ OB 4 OC — OH. (The vectorial solution is left to the reader to carry
out; below we give another proof of all these propositions by resorting
to complex numbers.)
For the present, note the following simple synthetic solution.
1°. In the case of a homothetic transformation ( G, — , the points
A, B, C go into the points A l9 Bl 9 C1 and, hence, the altitudes of the triangle
ABC (which are now regarded as straight lines) go into the midperpendic
ulars of the sides BC, CA, AB (the midperpendicular of a line segment
is a straight line perpendicular to the line segment at its midpoint). From
this it follows that the point H of intersection of the altitudes of A ABC
goes into the point O of intersection of the midperpendiculars of its sides.
Thus, in the case of a homothetic transformation , point H goes
into point O, which means the points G, H, and O are collinear and
OG = GH/2.
2°. In the case of a homothetic transformation
goes into the circle (A ^ C x ) and, hence, the centre of the circle ( A ^ C j)
is the image of the centre O of the circle (ABC) that is, the midpoint 0 9
Vector Algebra 33
C«
Fig. 1
of segment OH. Since 0 9 is the midpoint of segment OH, and the points O
and H are projected respectively onto the side BC into the points Ax
and A2, it follows that the point 0 9 is projected into the midpoint of the
line segment AXA 2 and, hence, 0 '9AX= 0'9 A2, where 0 9 is the projection
of 0 9 on the straight line BC. But if the projections of inclined lines are
equal so also are equal the inclined lines themselves, hence, 0 9 AX= 0 9 A2.
Similarly, proof is given that 0 9 BX= 0 9 B2, 0 9 CX= 0 9 C2 and since
0 9 AX— 0 9 BX= 0 9 CX [0 9 is the centre of the circle (AXBXCX)], it follows
that the circle (AXBXCx) also passes through the points A2, B2, C2. Note
that the radius of the circle (AXBXCX) is half the radius R of the circle (ABC).
Now let us consider the homothetic transformation Under
this transformation, the point O goes into the point 0 9, and the circle (ABC)
goes into a circle with centre 0 9 and radius Rj2, that is, into the circle
(AxBxCj). But under the homothetic transformation —- j , the points
A, B, C go into the points A3, B3, C3 and since the points A, B, C lie on
the circle (ABC), it follows that the points A3, B3, C3 lie on the Euler circle
(AXBXCX) = (0 9).
3°. Finally, let us consider the homothetic transformation (H, 2). Under
this transformation, the circle (AXBXCX) goes into a circle with centre O
and radius R, that is, into the circle (ABC). On the other hand, under
the transformation (H, 2), the points A2, B2, C2go into the points Ax, Biy C4,
which are symmetric to the point H with respect to the straight lines
BC, CA, AB. But the points A2, B2, C2 lie on the Euler circle and, hence,
the points Ait B4, C4 lie on the circle (ABC).
ii hl()
8. Suppose H, G, O, and 0 9 are, respectively, the orthocentre of A ABC,

its centroid the centre of the circumscribed circle (ABC), and the centre
of the Euler circle; let A l 9 Bu Cx be the midpoints of the sides BC, CA, AB;
let AfI, Bn, Cn be the feet of the altitudes; let A', B', C be points symmetric
to the vertices A, B, C of A ABC with respect to its sides BC, CA, AB;
let A ", B", C" be points in which the altitudes AH, BH, CH intersect the
circumscribed circle (ABC); let O' be the centre of the circle (A'B'C') =
= (O'); let a, /?, y be the projections of the point 0 9 on the sides BC, CA, AB
of A ABC. Then:
1°. The triangle A'B'C' is an image of the triangle afiy under the homo-
thetic transformation (G, 4).
2°. 4 0 9a = AA", 40 9fi = BB", 40 9y = CC ".
3°. For the circumscribed circle (ABC) to be tangent with the altitude
dropped from A onto BC, it is necessary and sufficient that the centre 0 9
of the Euler circle lie on the side BC.
4°. Let 0 9 be a point symmetric to the point 0 9 with respect to the
centre co of the circle (oc/fy); then the centres O and O' of (ABC) and
(A'B'C') are symmetric with respect to the point 0 9.
5°. If the centre O' of the circle (A'B'C') lies inside A ABC, then it
is a point that has the property that the shortest distances from O' to
each vertex A, B, C with a preliminary shift along the straight line to
the opposite side are equal.
6 °. If the point O' lies on the side BC, then the point lies on the.4
circle (A'B'C').
Hint. 1°. 4G0 9 = GH, 4 0 9x = 2(HAU + OAx) = HA" + AH =
= HA" + A" A' = HA'.
2°. 4 0 9a = HA' = AA".
3°. This is a consequence of item 2°.
4°. On the basis of item 1° we have Geo = — GO' and besides,
4
but we also have
5°. If P is an arbitrary point lying within A ABC, then, denoting by

A2, B2, C2 the points of intersection of the straight lines PA', PB', PC'
with the sides BC, CA, AB, we conclude that the lengths of the polygonal
lines PA2A = PA', PB2B = PB', PC2C = PC' will be the shortest routes
from point P to the vertices A, B, C with preliminary displacements to
Vector Algebra 35
the opposite sides. For these minimal distances to be equal, it is necessary

and sufficient that PA' = PB' = PC', that is, that the point P coincide
with the point O'.
6° This is a consequence of item 5°.
9. Given three noncoplanar vectors
OA = a, OB = b, OC = c.
Let S be the centre of a sphere passing through the points 0 9 A, By C.
Find the vector OS = x.
a2 [b, c] + b2[c, a] + c2[a, b]
Answer, x =
2(a, b, c)
10. Given the vectors
OA = a, OB = b, OC = c.
The vectors b and c are noncollinear. Let H be the orthogonal projection
of point A on the plane OBC. Find the vector OH = h.
Answer, h = a — -?)_ [b, c].
[b, c?
11. Given four vectors
OA = a, d.
It is also given that the vectors a, b, c are noncoplanar and that the straight
line OD intersects the plane ABC at some point M. Find the vector OM = m.
(a, b, c) ,
Answer. ------------------------------------- d.
(d, b, c) + (d, c, a) + (d, a, b)
12. Given three vectors
OA a, c.
The vectors a and b are noncollinear. Let H be the orthogonal projection
of the point C on the plane OAB. Find the vector CH = x.
Answer, x = ----— [a, bl.
[a, b f
13. Find the vector x if three noncoplanar vectors a, b, c and their
scalar products into x are known:
xa = p, xb = q, xc = r.
p[b, c] + q[c, a] + r[a, b]
Answer.
(a, b, c)
14. Find the vectors x and y if we know their sum x + y = a, the

scalar product (x, y) = p and the vector product [x, y] = b.
Answer. If a 4 < 4(b2 + pa2), then there are no solutions. If a 4 =
= 4(b2 + pa2), then there is one solution:
a , [a,b] a [a, b]
x —t- , y
2 a2 2 a2
If a4 > 4(b2 + pa2), then there are two solutions:
_ a2 + Ya* — 4(b2 +pa2) , [a, b]
X l~ 2a 2 *+
_ a2 — ]/a4— 4(b2 +pa2) _ [a, b]
yi “ 2a2 “ c f~
a2 — j/a4 — 4(b2 +pa2) [a, b]
2a2 8 ~ ~ a2
a2 + 1[a* — 4(b2 -\-pa2) n , [a, b]
a *
15. ABCD is an arbitrary tetrahedron located in oriented space; P is
— > — > — > — >
an arbitrary point. The vectors PA, PB, PC, PD are linearly dependent:
ocPA+PPB + yPC + 8 PD = 0
(at least one of the numbers a, /?, y, 8 is different from zero). Prove that
a: P: y: 5 = (PBCD): (APCD): (ABPD): (ABCP).
Hint. Project the zero vector olPA + fiPB + yPC + 8 PD on the straight
line AB by planes parallel to the plane PCD. The points P, C, D are pro
jected into the single point Pj and we obtain aP±A + PP±B = 0, and
so on.
16. Given a trihedral angle OABC such that among its plane angles
BOC, COA, AOB there is not more than one angle equal to n/2. Prove
vectorially that the three straight lines that pass through the vertex O,
lie in the planes of the faces BOC, COA, AOB, and are perpendicular to
the edges OA, OB, OC, respectively, lie in one plane.
Hint. If a, b, c are the direction vectors of the edges, then the direction
vectors of the indicated straight lines are [a, [b, c]], [b, [c, a]], [c, [a, b]].
Their sum is zero.
17. Prove vectorially that if all edges of the trihedral angle OABC,
all plane angles of which are right angles, are cut by a plane that does not
pass through its vertex O, then the point of intersection of the altitudes
Vector Algebra 37
of the triangle obtained in the section coincides with the projection of

the point O on the plane of that triangle.
18. ABCD is an arbitrary tetrahedron: A', B', C', D' are the orthogonal
projections of its vertices on the planes of the opposite faces. Given the
radius vectors
DA = r„ DB = r2, DC = r3.
1°. Find the radius vectors DA' = t[, DB' = r2, DC' = r^, DD' =
and prove that if the straight lines AA' and BB' lie in one plane, then
AB J_ CD, and conversely.
2°. If it is also given that AC — AD = BC = BD, then prove that the
straight lines AA' and BB' lie in one plane. Let H be the point of intersection
of these straight lines, and let K be the point of intersection of the straight
lines CC' and DD'. Prove that the plane AHB intersects the segment CD
at its midpoint / and the plane CKD intersects the segment AB at its mid
point J. Prove that the point H is located on the straight line IJ.
19. Given, in a pyramid OABC, the length of the edge OA = a and
the plane angles /_ BOC = a, /_ COA = P, /_ AOB = y. A sphere (5)
is tangent to the face BOC at the point O and passes through the point A.
Find its radius x.
Solution. Consider the vector x = OS. Since the vector x is perpendic
ular to the plane BOC, it follows that x = 2[b, c], where b = OB, c = OC.
From the equality OS = SA, we find x2 = (a — x)2, where a = OA,
whence xa = a2 /2. Substituting 2[b, c] for x in this equation, we obtain
2(a, b, c) = a2 j 2 , whence X = and, consequently, OS = x =
2(a, b, c)
= a2 [b, c] . The length x of the vector x is equal to the radius R (of the
2(a, b, c)
sphere S):
a2 \[bc]\ a2bc sin a
R \x\ = x =
2|(a,b,c)| 2|(a, b, c)|
a2bc sin a
2abc 1^1 + 2 cos a cos p cos y — cos2a —cos2/? —cos2y
'finis,
a sin a
R —- ■ - ■ - ----- --- —:— ■. .v.— ' = •
2 ]/1 + 2 cos a cos /? cos y — cos2a — cos2/? — cos2 y
20. Given, in a tetrahedron OABC, the lengths of the edges OA = a,
OB — b, OC — c and the plane angles at the vertex O: /_ BOC = a,
/_ COA = P, /_ AOB = y. Find the radius x of the sphere (5) circum

scribed about the given pyramid.
Solution. Let x = OS, a = OA, b = OB, c = OC; then
x2 = (x — a)2 = (x — b)2 = (x — c)2
whence
xa = a2 j 2 , xb = b2 j 2 , xc = c2/ 2 .
By the Gibbs formula,
x = xa-a* + xb-b* + xc-c*,
where a*, b*, c* is the reciprocal triple of the triple a, b, c, we find
x = ^ a* + ^ b* + ^ c * = * » . . P t e j L + c2ta>■»]
2 2 ‘ 2 2(a, b, c) 2(a, b, c) 2(a, b, c)
and, hence,
„ _ ife _ V c] + Z>2[c, a] + c2[a, b]}2 _ ff
V 21(a, b, c)| 2 j(a, b, c)| ’
where
T = a \ b, c]2 + b*[c, a]2 + c4[a, b]2
+ 2a 262[b, c]*[c, a] + 262c2[c, a]*[a, b] + 2cW[a, b]-[b, c]
= axb2 c2 sin2a + bxc2 c? sin2/? + c*a2 b2 sin2y
+ 2a2 b2(bc cos a- ac cos ft — c2ab cos y)
+ 2b2 c2(ca cos /?• ba cos y — a2bc cos a)
+ 2c2 a2{ab cos y • cb cos a — b2ca cos /?)
= <fib2 c2 sin2a -f b*c2 a2 sin2/? + c*a2 b2 sin2y
+ 2 a2 b2 c2(cos a cos /? — cos y) + 2 b2 c2 a2(cos p cos y — cos cl\
+ 2c2 a2 b2(cos y cos a — cos P)
= a2 b2 c2 {a2 sin2a + b2 sin2/? + c2 sin2y
+ 2a&(cos a cos P — cos y) + 26c(cos P cos y — cos a)
+ 2ca(cos y cos a — cos /?)} = a2 b2 c2 M,
|0 »>b. c)|
= abc ][ 1 + 2 cos a cos P cos y — cos2a — cos2/? — cos2y = abc • d.
To summarize:
1fM
X~ 2A
21. Given the plane angles a — Z_ BOC, b = Z. COA, c = /_ AOB
of a trihedral angle OABC and the interior dihedral angles A, B ,C that
Vector Algebra 39
are respectively opposite the plane angles. In item 6 ° of example 1, Sec. 2,

proof was given that
sin a sin b sin c A
-------- - = --------- = ---------= ------9
sin A sin B sin C A*
where A is the volume of the tetrahedron constructed on the unit vectors
OA, OB, OC, that is,
A = (1 + 2 cos a cos b cos c — cos2a — cos26 — cos2 c) 1/2
and
A* = (1 — 2 cos A cos B cos C — cos2A — cos2B — cos2C)1/2
Consider the following special cases:
1°. a = A, b = B, c = C. Prove that in this case at least one of the
angles A, B, C is equal to n/2 {Hint: A 2 — A * 2 = 0); furthermore, two
edges are perpendicular to a third; two dihedral angles of the trihedral
angle OABC are equal to n/2, and the third one is arbitrary.
2°. a — A, b = n — B, c = n — C. The reasoning is the same as in
item 1°.
3°. a = A, b = B, c = n — C. Prove that in this case cos a cos b cos c
and cos A cos B cos C have opposite signs. The plane angle c is found
from the relation
cos A cos B
cos c -----------------------
1 + sin A sin B
4°. a = n — A, b = n — B, c — n — C. Piove that the edges of the
trihedral angle OABC are pairwise perpendicular.
22. ABCD is an arbitrary tetrahedron. Let
x - [DB, DC], y = [DC, DA], z = [DA, DB], t = [AC, AB].

1 °. Prove that
x + y + z + t = 0.
Derive from this fact that with any tetrahedron ABCD it is possible to
associate three spatial quadrangles whose sides are perpendicular to the
faces of the tetrahedron and the lengths of the sides of each of the spatial
quadrangles are proportional to the areas of the faces of the tetrahedron.
Let lia, hb, hc, hd be the altitudes of the tetrahedron ABCD, let (AB),
(AC), (AD), (BC), (BD), (CD) be the interior dihedral angles of the tetra
hedron, which angles are adjacent to the edges AB, AC, AD, BC, BD, CD.
Prove that
0O 1 cos (CD) cos (DB) cos (BC)
K ~ K ~ ” /T '
y 1 J ____ 2 cos{CD) _ 1 1 2 cos(AB)
hi h% hahb hi hi hchd
4°.
2 V« Af «A? ’ hi J
cos(/L5) cos(AD) cos (BC) cos(2?Z>) cos (CD)
hji* + KK hA hA h jic hahh
- 1 cos (CD) cos (BD) cos (BC)
5°. cos (CD) - 1 cos (AD) cos (AC) = 0 .
cos (BD) cos (AD) -1 cos (AB)
cos (BC) cos (AC) cos (AB) -1
Hint. Form the scalar product of the equation x + y + z + t = 0 b y x ;
write down the equation x + y + z + t = 0 as x + y = — z — t and
square both sides; square the equation x + y + z + t = 0 ; eliminate h
in the equations obtained in item 1°.
23. Prove that the six planes passing through the midpoints of the edges
of the tetrahedron ABCD and perpendicular to the opposite edges of the
tetrahedron pass through the single point M (Mongers point). Prove that
the Monge point is symmetric to the centre O of the sphere (O) = (ABCD)
with respect to the centroid G of the tetrahedron ABCD.
Hint. Let OA = rl 9 OB = r2, OC = r3, OD = r4. Then the equations
of the planes indicated in the statement of the problem may be written
thus:
+ r, + r , + r 4Lj(r<
l ,_ ~ _r*), =_ A / ^ y (/J = 1; 2, 3,4).
24. Given a tetrahedron ABCD and a point M. Let the straight line
passing through M parallel to the straight line AB intersect the faces CD A
and CDB at the points P and Q. Prove that the sum of the scalar products
M P • MQy the terms of which are made up as indicated for all six edges
of the tetrahedron ABCD, is equal to the power of the point M with respect
to the sphere (ABCD).
25. Given a tetrahedron ABCD and a point M. Prove that the sum of
the powers of the vertex A with respect to the spheres with diameters MC,
MB 9 MD is equal to the sum of the squares of the lengths of all edges of
the tetrahedron ABCD (and similarly for the vertices B, C, D).
26. Find the tetrahedron if we know the areas of its faces and also that
the altitudes intersect in a single point (such a tetrahedron is said to be
orthocentric).
Solution. Lagrange investigated this problem analytically (J. L. Lagrange
Mem. Acad. Berlin, 1773, p. 160; OEuvres, t. Ill, p. 662) and formulated
it slightly differently:
“Find a tetrahedron of greatest volume for which the areas of all its faces
are given”. From the formulas he obtained it follows that the desired
tetrahedron is an ortho centric tetrahedron, as was also pointed out by
Vector Algebra 41
Serret (P. Serret.—J. de Liouville, 1862, p. 377), who proved this fact
geometrically (without, however, going into the definition of such a tetra
hedron). In connection with this problem, Lagrange obtained a fourth-
degree equation and proved that it has at least one positive root; he com
puted the lengths of the edges of the tetrahedron as functions of that root,
but did not consider the conditions under which a tetrahedron with such
lengths of edges exists. Finally, an English mathematician Iyenger (Iyenger,
The Mathematics Student, 1947, p. 104) solved the same problem by reduc
ing it to a fourth-degree equation; he carried out a full investigation and
proved that for such a tetrahedron to exist it is necessary and sufficient
that the sum of the areas of any three faces be greater than the area of
the fourth face. We now give the solution proposed by Marmion (A. Mar-
mion, Mathesis, 1953, p. 69; problem No. 3253 proposed by V. Th6bault).
As we know, the sum of four vectors x, y, z, t perpendicular to the edges
of the tetrahedron ABCD and directed outwards (the lengths of the vectors
are equal to the areas of the faces of the tetrahedron) is equal to zero
(see example 22). From this it follows that there exists a closed spatial
quadrangle A 1 B1 C1 D1 whose sides are such that AXBX = x, B1 C1 = y,
C1 D1 = z, D1 A 1 = t and, consequently, the sum of the areas of any three
faces of the tetrahedron ABCD exceeds the area of the fourth face.
Putting DA = a, DB = b, DC = c, we conclude that for the vectors
x, y, z we can take the following vectors:
x= [b, c], y = [c, a], z = ^ -[a , b].

2 2 2
From the vectors x, y, z, t we can generate 6 spatial quadrangles of the
type A 1 B1 C1 D1; all tetrahedrons A 1 B 1 C 1 D1 will have equal volume V1:
y , = 4 “ (x >y> z) = — ([b, C], [c. a], [a, b])

6 48
48 ( a>b>c)2 = i48( 6 F)2 = i4 - v s •

= ^
From this it follows that the tetrahedrons ABCD and A 1 B 1 C1 D1 assume
the maximum value simultaneously. On the other hand, the volume of
any tetrahedron is equal to 2/3 of the product of the areas of two of its
faces into the sine of the dihedral angle <p between them divided by the
length of the edge of that dihedral angle, and so
Tr 2 area A Â C V area A A ^ Q .
V1 = --------------- -- - ----------------—-—- sin cp,
3 AXCX
where cp is the size of the dihedral angle with edge AiCi. Since the factor
in front of sin cp does not depend on cp, it follows that Vx is a maximum
when cp = n/2. Similar reasoning carried out with respect to the other
faces A1B1D1 and C1B1D1 leads to the fact that the dihedral angle formed
by these faces must also be a right angle. But since
A±BXLDBC, BXCX1.ACD, C A 1 BDA, DXAX 1 CAB
it follows that
AB 1 A&Dto BC _L B& Ato CD 1 C ^ A , DA 1
and, consequently, AB _L CD, BC 1 AD, that is, the tetrahedron ABCD
of maximum volume for which the areas of its faces are given is ortho
centric.
The coefficient of sin <p in the expression for Vx depends solely on one
variable AXCX= A. Applying the Heron formula for the areas of A AxByCx
and AAxDxCx and noting that AXBX= x, B1C1 = y, CXDX= z, DXAX = t,
we find that the square of the expression
area A ^ A C ra re a A A1C1Dl
A&
multiplied by 28 is equal to
m 2 ) = [V - ( x + y f] [A2 - (x - y f] [A2 - (t + z f ][A2 - (t - zf]
{A) A2
Putting A2 = n, we find the following expression for the logarithmic deri
vative of the function F(p):
fM . = 1 , 1
F(n) V -(x + y f II — (x — y f
1 1 1
n — (t + z f n - (t - z f n
Now let (x — y)2 < (t — z)2 < (x + j ')2 ^ (I + z f since it must be
true that
(t — z f < AXC\ = H, (x + y f > AXC\ = n
(only under this condition do triangles A1B1C1and A±DXCXexist with lengths
of sides y, x, A and t, z, A). The derivative F \n) has only one root in the
interval ((t — z)2, (x + y f ) and therefore there is only one value of A for
which the function F (A 2) attains an extremum, a maximum. Knowing A,
we can construct A Ai^ C j and A A ^ D ^ Putting one against the other
so that the dihedral angle B1{A1C-j)D1 is equal to n/2, we construct the tetra
hedron (and there is only one such tetrahedron to within isometry) that cor
responds to the tetrahedron ABCD with greatest volume. Furthermore,
by virtue of the relations
AB 1 AXCXDX, BC JL B & A ^ CD 1 DA 1 D1B1Cl
Vector Algebra 43
and
AB + BC + CD + DA = 0,
we conclude that AB, BC, CD, DA must be proportional to the areas of
the faces A ^ D ^ A & D ^ B j C ^ , C1D1B1 so that, denoting propor
tionality factor by k , we have
AB = k iA & D J , BC = k iA & D J ,
CD = k iB & A J , DA = k iC & B J ,
where, for instance, (A ^ C ^) is the nonoriented area of face A ^ C ^
and on the basis of the foregoing, we have
3 27
V = — k*Vf = -----k? V4,
4 64
whence we find the proportionality factor
k = _ i_ = _ 4 _
3V )[3Fi
The lengths of the edges AB, BC, CD, DA have been determined and
they must be perpendicular to the planes A ^ D ^ A ^ D ^ B ^ C ^ , C x D ^
respectively. To summarize: up to isometric transformations (motions),
there is only one tetrahedron that satisfies the statement of the problem.
CHAPTER II
ANALYTIC GEOMETRY
Sec. 1. Application of analytic geometry (solved problems)
Problem 1. The sides BC, CA, AB of a triangle ABC are divided by the
points P, Q, R in the ratios
PP __ ^ Cg AR
V, = v*
PC RB
Find (PQR)
(ABC)
Solution. We introduce a general Cartesian coordinate system in
the plane, putting C = (0, 0), A = (1, 0), B = (0, 1). Then P =
= ( 0, - L - V q = (-J L -; (A , * = - - - - - ) and, con-
I l+ X ) U+Ai ) ll+v 1 + vj
sequently,
0 — — 1
1+ X
(PQR) n 1 + Xp v
0 1
{ABC) 1+ n (1 + A) (1 + //) (1 + v)
1
— v- 1
1+ V 1+ V
Corollary. For the points P, Q, R, lying on the sides BC, CA, AB of
A ABC, to be collinear, it is necessary and sufficient that the following equa
tion hold:
BP _C Q _AR__ BP CQ AR
PC QA RB CP AQ BR
(Menelaus’ theorem).
Problem 2. A triangle ABC is inscribed in a circle. Prove that the points
P, Q, R of intersection of tangents to the circle at the points A, B, C are
respectively collinear with the sides BC, CA, AB (this is special case o f
the Brianchon theorem).
Analytic Geometry 45
Proof‘ Note that all the ratios

CQ AR
P= — * v=
PC PP
are negative. Furthermore,
BP _ sin C PC sin P
sin B P^4 sin C
consequently,
PP sin2 C
sin2 P 6*'
PC
Z>2
Similarly, /* = — —, v = ----- and so An v = — 1 (see problem 1).
c2 fl2
Problem 3. Three circles with centres A, B, C and radii Rl9 R2, P 3 are
located so that each of them lies outside the other two. Let P, Q, R be
the centres of exterior similitude of these circles taken in pairs. Prove
that the points P, Q, R are collinear.
D D
*2 _ Ri
Proof A = —- = h __ ^ _ ^3 v - AR -
PC QA V Tb
whence Afi v = —1,
Problem 4. The straight lines AAl9 BBl9 CCX belong to one pencil (in
particular, they pass through one point 0). Prove that the points P, Q, R
of intersection of the corresponding sides PC and P 1C1, CA and CxAl9 AB
and AXBX of the triangles ABC and AXBXCX are collinear ( Desargues’
theorem).
Solution. Consider A OBC and the transversal BXPCX (the transversal
o f a triangle is any straight line lying in the plane of the triangle). On the
basis of Menelaus’ theorem we have
OBx BP CCX
— >— > — >
BXB PC CxO
Similarly,
OCx CQ AAX
— > — >— > = - l ,
CXCQA AxO
— >— > — >
OAx AR BBX
AXA RB BxO
Multiplying these equalities term by term, we get

■>— >— >
BP CQ AR
— >— > — >
PC QA RB
It is now left to the reader to prove Desargues’ theorem for the case where
AAX, BBl9 CCx belong to one and the same improper pencil, that is to say,
are parallel.
— > — >— >
Problem 5.1°. The sides BC, CA, AB of A ABC are divided in the ratios
PC QA RB
Suppose the straight lines BQ and CR intersect in the point Al9 the straigh
lines CR and AP in the point Bl9 and the straight lines AP and BQ in the
point Ci. Find the ratio *—
(ABC)
2°. Consider the special case A = p = v = 2.
3°. Under what necessary and sufficient condition do the straight lines
AP, BQ, CR belong to the same proper pencil?
4°. Under what necessary and sufficient condition do the straight lines
AP, BQ, CR belong to the same improper pencil?
Solution. 1°. Introduce a general Cartesian coordinate system and set
C = (0, 0), A = (\, 0), B = (0, 1). Then
The equations of the straight lines AP, BQ, CR are:

x y 1
= 0 or jc + (1 + X)y — 1 = 0 (straight line B1C1 or AP)

0 ------- 1
1 +A
.v y 1
= 0 or (1 + /i) x + py — n — 0 (straight line CLA±or BQ),
1 + li
x y 1
—- — — V— 1 = 0 or vx — y = 0 (straight line AXBX or CR)

1+ v 1+ v
0 0 1
whence
1 1 + A —1 2
1 +A* —J“
(A 1 B1 C1) _ V -1 0
(.ABC) (1 -J- [i -f- f i \ ) (1 -f- v -f- A v) (1 A -f- /iA)
= _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ H v - I ) 2_ _ _ _ _ _ _ _ _ _ _ _ _
(1 + n + fiv) (1 + v + Av) (1 + A + /«A)
Remark. By solving the systems of the straight lines BxCl9 CxAl9 AXBU
we could have found the vertices of the triangle AXB^CX\
vti
1 + n + nv 1 + JJ +/iV
1 V
1 + v + vA 1 + v + vA,
A/z 1
1 -f- A -f- AjU 1 + A + A/i
and then
\i fiv 1 + /I + p v
1 V 1 + V+ Av
(A1 B1 C1) Aî 1 1 -j- A -f- A /i
{ABC) (1 + A + A /i) (1 + n + /tv) (1 + v + vA)
By subtracting from the last column the sum of the first two and simpli
fying, we obtain the very same result.
2°. For A = n = v = 2, we find — - - 1— = — •
(ABC) 1
3°. The straight lines AP, BQ, CR belong to a single proper pencil if
and only if A/z v = 1,1 + A+ A/z ^ 0 (by multiplying both sides of the in
equality 1 + A + X\x # 0 by v ^ 0, we obtain 1 + v + vA # 0; then,
multiplying by n ^ 0 , we obtain 1 + ^ + / i v ^ 0).
4°. The straight lines AP, jB<2, CR belong to the same improper pencil,
if and only if A// v = 1, 1 + A + A^ = 0 (in this case, the lines
CR are pairwise parallel, and there are no coincident lines).
Combining items 3° and 4°, we obtain a theorem: if the sides BC, CA,
AB of A arc divided in the ratios
then the straight lines AP, BQ, CR belong to a single pencil (proper or im
proper) if and only if X p v = 1 ( Ceva's theorem).
Problem 6 . Prove that if a circle inscribed in A ABC is tangent to the
sides BC, CA, AB at the points, respectively, P, Q, jR, then the straight
lines AP, BQ, CR intersect in one point.
Solution.
BP p -b P -a
X
p -b
(p = (a + b + c)/2, a, b, c are the lengths of the sides of A ABC), whence

X p v = \, 1 + X + X p ^ O .
Problem 7. Two distinct points A and B are fixed in a plane. Also fixed
is a real number k. Find the locus of points M for each of which
M A2 + MB2 = k.
Solution. We introduce a rectangular Cartesian system of coordinates.
Let
A = (xl5 y j , B = (x2, y 2), M = (x, y).
Using the formula for the distance between two points in a plane, we get
(x - xxy + ( y - J ,)2 + (x - x 2)2 + ( y - y 2)2 = k
or
or
MO 2 = — _ — A B \
2 4
where O is the midpoint of segment AB. From the last relation, it follows
that if k < AB2/2, then the specified locus is empty. If k = AB2j2, then
the specified locus consists of the single point O, the midpoint of AB.
If k > AB2! 2, then the given locus of points M is a circle with centre O and
radius ]f2k — AB2/2.
Problem 8 . Three points A, B, C are fixed in a plane. Also fixed is a real
number k. Find the locus of points M for each of which
M A2 + MB2 + MC2 = k.
Solution. Introduce in the plane of A ABC a rectangular Cartesian coor
dinate system. In this system, let
A = (xl5 Ji), B = (x2, y 2), C = (x3, j 3), M = (x, y).
Applying the formula for the distance between two points, we obtain
(x - Xj)2 + o - yyf + (x - x 2f + ( y - y 2 ) 2 + (x - x3)2 + ( y - y 3f = k
or
yi + y 2 + j 3 j 2j
+ [(x3 - x 2)2 + (y3 — jo )2 + (xa - x3)2

+ (Ji - Js )2 + (*2 - xi)2 + (J 2 - J 1)2] = k.
Since Xl X2 ^ 2 are the coordinates of G, the

3 3
point of intersection of the medians of the triangle ABC, and
0 3 — Xo)2 + ( j 3 — j 2)2 = a2,
(xi - x3)2 + (yx — j 3)2 = b \
(x 2 - Xi)2 + 0 2 - >’l)2 = c2,
where 0 , ft, c are the lengths of the sides BC, C>4, >45 of the triangle, it
follows that the last relation may be rewritten thus:
MG 2 = — [3k - (a2 + b2 + c2)].
From this it follows that if k < (a2 + ft2 + c2 )j3, then the desired locus
of points M is empty. If k = (<a2 + ft2 + c2)/3 then the locus of points M
consists of the single point G, which is the point of intersection of the me
dians of the triangle ABC. If k > (a2 + ft2 + c2)/3, then the locus is a circle
with centre G and radius j/3k — {a2 + ft2 + c2)/3.
Problem 9. Fixed in the plane are two distinct points Cx and C2 and also
a real number k . Find the locus of points Af for which
MC\ - MCI = k.
Solution. Let us introduce a rectangular Cartesian coordinate system
and take C±C2 for the x-axis and the midpoint O of segment CXC2 for the co
ordinate origin. In that coordinate system, let
Ci = (—a, 0), C2 = (a, 0), M = (x, y).
Then the given relation takes the form
k
(x + a) 2 + y 2 — (x — a) 2 — y 2 = k or x = — 9
4 a
which is the equation of a straight line perpendicular to CXC2 (the x-axis).
4-810
Problem 10. The radical axis of two nonconcentric circles (Cl9 r±) and
(C2, r2) is the locus of points M for each of which its powers with respect
to the circles are equal. Set up the equation of the radical axis of the cir
cles (Cj, r2) and (C2? r2) assuming CXC2 to be the ;c-axis and taking for the
coordinate origin O the midpoint of the line segment CXC2.
Solution. In our coordinate system, C2= ( —a, 0), C2=(a, 0), M = (x, y).
The relation
or
MCI - r\ = MCI - r\
or
MCI - MCI = r\ - r\
takes the form
(x + a)2 + y2 — (x — a f — y2 = r\ — r\
or
o o
x = n ^ l.
4 a
This is a straight line perpendicular to the straight line C1C2 (also see the
preceding problem).
Remark. Consider the construction of a radical axis of two circles depend
ing on their different mutual positions.
(1) The circles (Cx) and (C2) intersect. Their radical axis is [the straight
line / that passes through the points P and Q (their points of intersection).
Indeed, the powers of both points P and Q with respect to (C2) and (C2)
are zero and hence the points P and Q belong to the radical axis of those
circles; therefore the radical axis itself is the straight line PQ.
(2) The circles (C2) and (C2) are located so that each lies outside the
other. Draw one of four straight lines tangent to the given circles; Pk, Qk
(k = 1, 2, 3, 4) are points of contact; Pk Qk are segments of the tangents.
Let the point M k bisect the segment Pk Qk. Then the point M k lies on the
radical axis of the given circles. Indeed, M k Pi = M k Q% = k 9 where k
is the power of the point M k with respect to both circles. Thus, the mid
points of the four segments Pk Qk of the common tangents to the circles
(Cx) and (C2) bounded by the points of tangency Pk and Qk lie on the ra
dical axis of the circles. To construct the radical axis, it suffices to construct
two of the points Ml9 M 2, M3, A/ 4 or even one (and then drop from it a
perpendicular to the line of centres CXC2).
Remark. If the nonconcentric circles (Cj) and (C2) do not intersect, their
radical axis does not have any points in common with either of them.
Indeed, suppose that the radical axis / of the circles (Cx) and (C2) has a
common point A with circle (Cj). Then the power of point A with respect
to the circle (Cx) is equal to zero, and since point A lies on the radical axis
of (Cj) and (C2), it follows that the power of A with respect to the circle (C2)
is also zero and therefore A also lies on the circle (C2). This is a contradic
tion: the circles (Cx) and (C2) have a common point A , which runs counter
to the assumption.
It is left to the reader to prove a stronger statement: if each of the cir
cles (Cj), (C2) lies outside the other, then (Cj) and (C2) lie on different
sides of their radical axis, and if one of the circles (Q), (C2) lies inside the
other (see below), then (Cx) and (C2) lie on one side of their radical axis
[in all instances we are dealing with two nonconcentric circles (Cx) and
(C2)].
(3) The circle (Cx) is a zero circle (point Cx) and lies outside the circle
(C2). Construct the tangent lines CxTx and C2T2 drawn from the point Cx
to the circle (C2) (7\ and T2 are the points of tangency). Suppose Mx
and M 2 are the midpoints of segments CXTX and C2T2; the straight line
MXM2 is the radical axis of the circles (C2) and (CV). One could construct
a single tangent line CxTly the radical axis is a straight line passing through
the midpoint M x of segment CXTXperpendicularly to the straight line CXC2.
(4) The circles (Q) and (C2) are zero circles. Their radical axis is the per
pendicular bisector of the line segment CXC2.
(5) The circles (Cx) and (C2) are nonzero and are tangent to one another
externally or internally. Their radical axis is the tangent line / at their
common point T. Indeed, the powers of any point M lying on the tangent
line / with respect to (CA) and (C2) are equal to M T2, which means they
are equal.
(6) The zero circle (Cj) “lies” on the nonzero circle (C2). The radical
axis of these circles is the tangent line to the circle (C2) at the point Cv
Indeed, the point Cx belongs to the radical axis since its powers with
respect to (C2) and (C2) are zero, and, besides, the radical axis is perpen
dicular to the straight line CXC2.
(7) The circle (Cj) is a zero circle, the circle (C2) is nonzero and the point
Cj lies inside (C2); also the circles are nonconcentric (Cx ^ C2). Let us
construct some circle (C) tangent to the straight line CXC2 at the point Cx
and such that it intersects (C2) at the points A and B. Then the straight
line AB intersects the straight line CXC2 at the point P lying on the radical
axis of the circles (C2) and (C2). Indeed, the power of the point P with re
spect to the circle (C2) is equal to PA-PB. But this product is equal to PC\>
that is, to the power of the point P with respect to the zero circle (Cx).
The radical axis of the circles (Cx) and (C2) is a straight line passing through
P and perpendicular to the straight line CXC2.
(8) The nonconcentric circles (Cx) and (C2) are nonzero and the circle
(C,) lies inside the circle (C2). Let us construct some circle (C) which inter
sects (C2) at the points A and B, and the circle (C2) at the points A ' and B'.
Let P be the point of intersection of the straight lines AB and A'Bf. Then
P belongs to the radical axis of the circles (Q) and (C2), since its powers
c = (PA) • (PB) and o' = (PA') • (PB') with respect to the circles (Ci)
and (C2) are equal: (PA) • (PB) = (PA') • (PB'); this is the power of the
point P with respect to the circle (C). The radical axis / of the circles (Q)
and (C2) is a straight line passing through the point P perpendicularly to
the straight line CXC2.
Problem 11. Fixed in a plane are two distinct points A and B and a
positive number k not equal to unity.
1°. Prove that the locus of points M for which the following equation
holds,
MA ,
-----= k ,
MB
is a circle (Cfc)*, the centre Ck of which lies on the straight line AB. Intro
duce a rectangular Cartesian coordinate system taking for the origin the
midpoint O of AB, and for the x-axis the straight line AB. On the x-axis
find the coordinate x k of the centre Ck of the circle (Ck) if we know the
coordinates —a and a of the points A and B (we assume a > 0, that is, the
positive direction of the coordinate axis is chosen from A to B). Also
find the radius R k of the circle (Ck).
2°. Prove that the centre Ck of the circle (Ck) is an external point of
the line segment AB.
3°. Prove that for any k (0 < k ^ 1) the circle (Ck) does not have com
mon points with the perpendicular bisector (or midperpendicular) s of
A B : for 0 < k < 1, the circle (Ck) and the point A lie to one side of the
midperpendicular ^ of AB [and, consequently, B and the circle (Ck) lie on
different sides of s], and for k > 1, the point A and the circle (Ck) lie on
different sides of s [and, hence, B and the circle (C*) lie on the same side
of s].
4°. Prove that if 0 < k < 1, then point A lies inside the circle (Ck)
and point B outside it. If k > 1, then, conversely, point A lies outside
(Ck) and B lies inside (Ck).
5°. Prove that the circles (Ckl) and (Ck2) (0 < # 1, 0 < fc2 # 1)
are symmetric with respect to the midperpendicular s of segment AB if
and only if kxk 2 = 1.
6 °. Let Pk and Qk be points of intersection of the circle (Ck) with the
straight line AB. Prove that the ordered quadruplet of points A, B, Pk,
Qk is harmonic, that is,
APk. AQk j
K b o i
* A circle thus defined is called a circle o f Apollonius. The points A and B are termed
limit points , or Poncelet points o f the circles (Ck) (if k varies from 0 to + o o , with the
exception o f k = 1, then we obtain a family o f circles (Ck); the case of k = 1 is clearly
associated with the straight line: the midperpendicular o f points A and B).
7°. Prove that if S is an arbitrary point of the midperpendicular s of

AB9 and T is the point of contact of any tangent (any one of two) drawn
from point S to the circle (Ck), then
S T = SA = SB.
In other words, the midperpendicular s of AB is the radical axis of any one
of the circles (Ck) (0 < k ^ 1), including the zero circles A and B.
8°. Prove that any one of the circles (S) passing through points A
and B intersects all circles (Ck) (0 < k ^ 1) orthogonally, that is, the tan
gents to the circles (S) and (Ck) at any one of two points of their intersec
tion are mutually perpendicular.
The set (Cfc) of all circles of Apollonius given by
MA
where k takes on all positive values (this set includes the midperpendicular
of AB as well) is termed a hyperbolic pencil o f circles. Points A and B are
called the limit points of the pencil, or Poncelet points.
The set of all circles passing through points A and B (this set includes
the straight line AB as well) is called an elliptic pencil o f circles which
is conjugate to the hyperbolic pencil. The points A and B are termed base
points of the elliptic pencil.
Remark. A hyperbolic pencil of circles may be defined in a variety of
ways (all the definitions given below are equivalent).
(1) A hyperbolic pencil r of circles is the image of a family of all con
centric circles (0) under inversion. Here, the pencil of straight lines with
centre O is transformed into an elliptic pencil E of circles which is conjugate
to the hyperbolic pencil r .
(2) A hyperbolic pencil r of circles is a set of all circles with a common
radical axis (that is, a set such that the radical axes of any two circles coin
side); in this case the radical axis does not intersect the circles.
(3) A hyperbolic pencil r of circles is a set of all circles (PQ) with dia
meter PQ, where the points P and Q are harmonic conjugates of the limit
points A and B (any two distinct points may be taken for A and B). The
set of all circles passing through A and B then form an elliptic pencil E of
circles conjugate to the hyperbolic pencil r.
(4) A hyperbolic pencil r of circles is a stereographic projection (see
Chapter IV) from an arbitrary points of the sphere of the set of all circles
of the sphere, the planes of which are perpendicular to any diameter N S
of the sphere. In this case, the projections A and B of points A and S are
limit points of the pencil r , and the stereographic projections of the set of
great circles of the sphere that pass through the points N and S form an
elliptic pencil E of circles conjugate to the pencil r .
(5) A hyperbolic pencil r of circles with limit points A (—a90) and
B (a9 0) is a family of circles given by the equation
A / + pep = 0 ,
where
/ = x2 + y2 — 2px + a2 = 0,
<p = x2 + y 2 — 2qx + a2 = 0, p ^ q
are any two circles of the family (at least one of the numbers A and p is
nonzero).
(6) A hyperbolic pencil r of circles is a set of circles specified by the
equation
x2 + y 2 — 2px + 1 = 0 ,
where p is a parameter that takes on all values that do not exceed 1 in abso
lute value. Then the conjugate elliptic pencil is given by the equation
x 2 + ^2 _ 2qx — 1 = 0 ,
where the parameter q takes on all real values; the fundamental points of
the pencils r and E are (± 1,0).
There are other definitions as well.
9°. Let 0 < k x < k 2 < 1. Prove that the circle (Ckl) is inside the circle
(Cki). But if 1 < kx < k2, then the circle (Cki) is inside the circle (Ckl).
10°. Prove the validity of the following method of constructing a circle
of Apollonius: given Poncelet points A and B and one of the points Md
of the circle of Apollonius not lying either on the straight line AB or on
the midperpendicular of the segment AB; at the point M0 draw a tangent
to the circle (ABM0). If C is the point of intersection of that tangent with
AB, then the circle of Apollonius passing through M0 is a circle with centre
C and radius CM0. Consider the case where the point M0lies on the straight
line AB but is distinct from the points A, B and the midpoint O of AB.
11 °. Find the locus of points M for each of which
„, 0 MA , ^
lli. ------> where 0 < k # 1,
MB
112- ~ ~ < k, where 0 < k ^ 1.
MB
12°. Prove that two nonconcentric circles (C!) and (C") without com
mon points define uniquely a hyperbolic pencil of circles to which they
belong. How are the limit points of the pencil constructed? Consider the
case where one or both circles are zero circles.
13°. Fixed in the plane are two distinct points A and B. Given a straight
line /. On / find points M such that the ratio
MA
MB
assumes maximum and minimum values. Investigate the question depend
ing on the position of the straight line / relative to the points A and B.
14°. Given: a plane n and a straight line / intersecting the plane in the
point A. Given an angle cp that the straight line / forms with the plane n
(if / is not perpendicular to the plane n, then <p is an acute angle between
/ and its projection /' on the plane n; if / is perpendicular to the plane n,
ip = 90°). A point B different from A is fixed on /.
Let PQ be an arbitrary line segment lying in space. Draw through points
P and Q straight lines collinear with / and let P' and Q' be the points of
intersection of these lines with the 7r-plane. The line segment P fQ' is called
the parallel projection of PQ on the rc-plane in the direction of /. The ratio
is called the deformation ratio of segment PQ under its parallel projection

on the 7r-plane in the direction of the straight line / (we will simply write
deformation ratio because in this problem we consider the parallel pro
jection only on the ;r-plane in the direction of the straight line /).
Let there be a fixed positive number k. In the 7r-plane, find the locus of
points M each of which has the following property: the deformation ratio
of any nonzero segment lying on the straight line BM is equal to k .
Solution. 1°. A = (—a, 0), B = (<a, 0), M = (x, y). Employing the for
mula for the distance between two points, we obtain
K(* + a f + /
—a f + y 2
or
)t2 + l \ 2 -f y2
--------- , = ------------
4a2 k* ,
A:2 - 1 J (k 2 — l ) 2
this is the equation of a circle with centre at the point Ck(xk, 0), where
k2 + 1
xk= a
k2 - 1
and with radius

2ak
Rk
I*2 - i f
2°. From the equation
k2 + 1
xk
k2 - 1
it follows that |xk| > a.

3°. If the circle (Ck) had at least one point M0 in common with the mid
perpendicular of AB, then for that point we would have
M0A
= k
M 0B
£2 i i
but this contradicts the condition 1. If 0<A:< 1, then x k= a ------- <0;
k2- 1
hence, the points Ck and A lie on one side of the midperpendicular of AB
and thus the point A and (Ck) lie on one side of the midperpendicular. If
k > 1, then x k > 0; hence, the points B and Ck lie on one side of the
midperpendicular of segment AB, and since the circle (Ck) does not
have any points in common with that midperpendicular, it follows that
the point B and (Ck) lie on one side of the midperpendicular of AB.
4°. The power of the point A with respect to the circle (Ck) is equal to
k2 + n 2 4a2 k 2 _ 4a2 k 2
k2- i ) (k2 - l)2 ~ k 2 - 1
If 0 < k < 1, then ga < 0, that is, point A lies inside (Ck) and, hence,
point B lies outside (Ck) since the points A and B lie on different sides of
the midperpendicular s of segment AB, and the point A and (Ck) lie
(for 0 < k < 1) on one side of the midperpendicular of AB; if k > 1,
then g a > 0 and A lies outside (Cfc).
The power g b of point B with respect to (Ck) is
k2+ 1 \ 2 4a2 k2 _ 4a2
k2 - 1 J (k2 — l)2 ~ 1 — k2
If 0 < k < 1, then g b > 0, and if k > 1, then g b < 0.
5°. Prove that if 0 < k x < k 2 < 1, then R kl < Rk2, and if 1 < k x < k 2,
then Rkl > Rki.
Indeed, if 0 < k x < k 2 < 1, then
2ak1_____ 2ak2 ^ {k1 k 2) (1 -|- k1k 2) ^ ^
-ftjfci ” Rk2
Y ^ k 2" i - k \ ~ (i - k\) (i - k2)
and if 1 < k x < k 2, then
R -R 2akl 2ak'2 = 2a ~ kl^ 1 + k l k ^ > Q

k2 “ k \ - 1 * i - i (*i - 1) ( * | - 1)
Thus, the circles (Ckl) and (Ck2) may be symmetric with respect to the mid
perpendicular s of segment AB in one of the following cases:
0 < k x < 1 < k 2, 0 < k2 < 1 < kv
Both cases are investigated in the same way. For example, let 0 < k ± < 1,
k 2 > 1; then
2ak1 2ak2
Rkt —
1~k\ k \-\'
A necessary condition for the circles (Cftl) and (Ckt) (in the case of 0 <
< k x < 1 < k 2) to be symmetric with respect to the midperpendicular s
of segment AB is that their radii be equal:
Rki —
2akx _ 2ak2
1 - k \ ~ ~ k f^ I 9
k x k\ — k x — k 2 + k\ k 2 = 0,
(ki + k 2) (kk k 2 — 1) = 0,
and since k x + k 2 > 0, it follows that k xk 2 — 1 = 0 , whence kjc2 = 1.
This condition is also sufficient for the circles (Ckl) and (Cki) to be sym
metric with respect to the midperpendicular s of segment AB. Indeed,
assuming that k xk 2 = 1, we have
k l + 1 = 2a w + i 1 +kl
x kl = 2a ~ x kt.
l-k\
kt-l
«- 1
Consequently the points Ch and Ckt are symmetric with respect to the point
O or, what is the same thing, with respect to the midperpendicular s of
segment AB. But from the condition k xk 2 = 1 it follows that Rkl = R kt,
which means the circles (Cfcl) and (Ckt) themselves are symmetric with
respect to the straight line s.
Remark. Everything stated in this item also follows from the fact that
the given equation is equivalent to the following equation:
MB _ 1
M A ~ lc
6°. One of the points A , B lies inside the circle (Ck) and other outside
the circle; to put it differently, one of the points A, B is an interior point
of the diameter PkQk of the circle (PkQk) and the other is an exterior point.
From this it follows that one of the points P k9 Qk is an interior point of
segment AB and the other is an exterior point. Therefore the numbers
PkB QkB
have different signs. But the absolute values of these ratios are equal to k>
since the points Pk and Qk lie on the circle (Ck) and, hence, belong to the
given locus. Thus,
APu.AQt = - 1.
PkB QkB
To summarize: the points Pk and Qk are harmonic conjugates of the points

A and B.
Conversely, if the points Pk and Qk are harmonic conjugates of the pair
A, By then
AP- = ( = k ),
PkB QkB
that is, the points Pk and Qk belong to the circle (Cfc) of the hyperbolic
pencil of circles with limit points A and B (for the indicated value of k).
From this it follows that the circle constructed on the segment PkQk as
diameter is the circle (Ck).
The proposition just proven that A, B, Pk, Qk is a harmonic set of four
points permits the following elegant construction of circles of Apollonius:
through points A and B draw two straight lines sx and s2 perpendicular
to the straight line AB (Fig. 2) and on them fix points 5Xand S2 that are
distinct from points A and B respectively. On the straight line ^2 lay off,
on different sides of the straight line AB, congruent line segments BPk and
BQ'k. The straight lines 5X/* and 51£>* intersect AB at the points Pk and
Qk. Then the straight lines
S,A, SXB, SLPk, SxQk
form a harmonic set of four points. The circle with diameter PkQk is the
circle of a hyperbolic pencil with limit points A and B.
Figure 3 gives yet another method for constructing a circle of Apollo
nius: let / be a tangent to the circle Q at the point S; N is a point on the
circle Q diametrically opposite to point 5.
Construct a family of chords PkQk of the circle Q parallel to its diameter
SN. Let A'B' be the diameter of the circle Q that is perpendicular to the
diameter SN, and let A and B be the projections of the points A ’ and B'
on the straight line / from the point N. Denote by Pk and Qk the projec
tions of points Pk and Qk on the straight line I from point N . The set of
four points
B, Pk, Qk
is a harmonic set since the quadruplet of straight lines
NA', NB', NPk, NQ'k
is harmonic (/_A'NB'=nl2 and NBf and NA' are bisectors of Z.Pk^Q'k)-
The circle with diameter PkQk is a circle of Apollonius with limit points
A and B.
This method of constructing circles of Apollonius is connected with the
so-called stereographic projection of a sphere on a plane (this projection
is used in the preparation of geographical maps).
7°. Let 5(0, b) be an arbitrary point of the midperpendicular s of seg
ment AB. Since all points of the midperpendicular s of AB are external
points of the circle (Ck), it follows that from the point 5 it is possible to
draw two tangents to the circle (Ck). Let S T be one of them and let T be
the point of tangency.
The power of point S with respect to the circle (Ck) is equal to ST 2:
k2 + 1 \ 2 4a2k 2
a = ST 2 =
(' k2 — I j
+ b2 -
(k2 - 1 ) *
= a2 + b2
and, consequently,
ST = ]fa2 + b2 = )[0 A2 + OS2 = SA = SB.
We see that the length of segment S T is independent of k, and, hence,
all circles (Ck) and the zero circles A and B have a common radical axis,
the midperpendicular s of AB.
8°. From what was proved in item 7° it follows that the circle whose
centre is any point S of the midperpendicular s of AB and whose radius
is equal to SA = SB orthogonally intersects the circle (Ck), since SA =
= SB = ST, where T is the point of contact of the tangent (either one of
the two) to (Ck) drawn from point S. Any circle passing through the
points A and B has its centre on the midperpendicular of segment AB and
its radius is equal to SA = SB.
9°. Let 0 < k x < k 2 < 1. The circles (Ckl) and (Ck%) do not have any
points in common for k x ^ k 2 because if they did have a common point
M 0, then the following relations would hold true:
. _ m 0a _
1 — —
M0B
but this is not true (k± ^ k 2). In the case of 0 < k x < k 2 < 1, point A
lies inside both (Ckl) and (Ck2), but since these circles do not have any points
in common, one of them lies inside the other. But if 0 < k x < k 2 < 1 we
have R ^ < R ki (see item 5°); this means the circle (Ckl) lies inside (Ck%).
But if 1 < k x < k 2, then the point B lies inside (Cfcl) and (Ck2) (see
item 4°), the circles (Cfcl) and (Cki) do not have any points in common
and R kl > R k2 (see item 5°); which means (Cfc2) lies inside (Ckl).
10°. (1) Construct a circle (ABM0). By what was proved in item 8°,
the circle (Ck) of Apollonius must intersect (at point M 0) (ABM0) ortho
gonally. Therefore the radius of (Ck) will lie on the tangent to (ABM0)
at the point M0. On the other hand, the centre Ck of the circle (Ck) lies
on line AB and therefore it is the point of intersection of the straight line
AB with the tangent to (ABM0) at the point M0. The value of k that cor
responds to the circle (Ck) is:
M 0A
m 0b
because the point M 0 lies on (Ck).

(2) If point M0 lies on the straight line AB but differs from the points
A, B, O, then it is one of the points Pk, Qk of intersection of (Ck) with that
straight line. The other point Ar0 of intersection of (Ck) with AB is found
from the relation
(ABM qN0) = —1,
whence
(OM0)-(ON0) = a \
where O is the midpoint of AB,, the point N0 is the image of point M 0
under inversion with respect to the circle constructed on AB as a diameter.
The construction of point N0 is given in Figs. 4 and 5; the point N0
belongs to the circle of Apollonius passing through point M0 because
from the relation
(OM0y(O N 0) = a2
it follows that
(ABM0 N0) = - 1 ,
AM 0 _ A N 0
BM0 ~ B N 0(~
The construction of point A^0, which is the harmonic conjugate of point M 0
with respect to A, B may be carried out in a great variety of ways (without
recourse to inversion).
If a straight line parallel to AB is given in a plane, then that construction
may be performed even with a straightedge alone.
11°. The inequality
MA ,
------> k
MB
or
MA2 - k2 MB2 > 0
is equivalent to the following:
(1 - k 2) x2 + (1 - k 2) y 2 + 2a{\ + k 2) x + a2 (1 - k2) > 0.
Analytic Geometry
If 0 < k < 1, then by dividing the left member of this inequality by

1 — k 2 and simplifying we obtain
This inequality is satisfied by the coordinates of all points (x, y) lying

outside the circle (Ck).
If k > 1, we get
This inequality is satisfied by the coordinates of all points lying within (Ck).
In similar fashion, proof is given that for 0 < k < 1 the inequality
AIA
MB
is satisfied by all points M lying inside (Ck), and for k > 1, by all points
lying outside (Ck)
12°. Let s be the radical axis of the circles (C') and (C") and let O be
the point at which the radical axis intersects the straight line C'C" of cen
tres of the given circles. Denote by Pk, Qk, Pk\ Qk respectively the points
of intersection of the circles (C') and (C") with the straight line C'C".
Since the point O lies on the radical axis of (C') and (C"), it follows that
its powers with respect to these circles are equal :
a = {OP'k)(OQ'k) = (OP;;)-(OQ'k')
(these powers are positive since the radical axis does not intersect the
given circles and, hence, point O lies outside the circles).
Let us now construct on the straight line C'C" the points A and B dis
tant ][a from O. Then
OA‘»= OB2 = (OPk) ■(OQk) = (OP’k )-(OQk’) = a.
From these relations it follows that
{ABP'k Q'k) = - 1, (ASP? Q'k' ) = - 1,
that is, (C') and (C") are circles of Apollonius for values k equal to
P U _ QjA k „ = PL'A _ Qk A
PLB Q’kB Pk B Qk B
13°. Construct the midperpendicular s of segment AB. Suppose that a
straight line / intersects this midperpendicular but does not pass through
the points A and B [case (1) see Fig. 6]. Let S be the point of intersection
of straight lines / and s. Now construct a circle with centre S and radius
S A ; it intersects the straight line / in two points M± and M2. Draw a tan-
gent line to the circle at point Mv Let P1 be the point of intersection of

this tangent and the straight line AB. Construct a circle with centre at the
point Px and with radius PxM^ This is a circle of Apollonius with the
limit points A and B. Thus
M XA
— k x > 1.
M XB
All the points of the straight line I lie outside the circle (Ckl) and so
MA
for all points of / different from Ml9 the ratio -----is less than kl9 that
MB
MA
is, at the point Mx the ra tio -----is a maximum. Similarly it can be proved
MB
MA
that this ra tio ----- is minimal at the point M2.
MB
Figure 6 gives distinct positions of the straight line / with respect to the
points A and B:
(1) The straight line / does not pass through the points A and B and
is not perpendicular to AB: at the point M x we have the maximum
of the ratio \ and at M 2 the minimum (this case was examined above
(
in detail).
(2)
m b)
The straight line / is perpendicular to AB, does not pass through

the points A and B and is located on the side of the midperpendicular of
AB on which point B lies. At Mx we have the maximum, and there is no

minimum.
(3) The straight line / is perpendicular to AB, does not pass through
the points A, B and is located on that side of the midperpendicular of AB
on which the point A lies. At point M 2 we have the minimum, and there is
no maximum.
(4) The straight line / passes through the point B but does not come to
coincidence with AB, the minimum is at Af2, and there is no maximum.
(5) The straight line / passes through point A but does not coincide with
AB, the minimum is at point A ( the ratio in this case is zero^; the
V MB )
maximum is at point Mv
(6) The straight line / passes through point B and is perpendicular to
AB. There is neither maximum nor minimum.
(7) The straight line / passes through point A and is perpendicular to
MA AA \
(---- = — =01, and there is no maximum.
(8) The straight line / coincidesMB
with AB )
the straight line A B. The minimum
is at point A = 0 V and there is no maximum.
\M B )
14°. The locus of points M in the rc-plane is found from the condition
MA
MB
This same locus of points M in space is a sphere S of Apollonius with limit
points A and B. If 0 < k < 1, then the sphere S contains within itself
the point A, and since A lies in the rc-plane, the sphere S and thera-plane
intersect along the circle a, which is the desired locus of points M. But
if k > 1, then the sphere S contains within it the point B and here three
cases are possible: either the sphere intersects the 7r-plane or is tangent
to the 7r-plane, or does not have a single point in common with the 7r-plane.
Let us investigate these cases. On the straight line AB we assume a positive
direction from point A to point B. Then
AP = - ~ - AB,
k2 -
where P is the centre of the sphere S. The distance d from the centre P
of sphere S to the rc-plane is
k2
d — --------- AB sin <p.
k2 - 1
The sphere S intersects the 7r-plane if and only if its radius
k
r= AB
k2 - 1
is greater than d :
k 'A B kVAB .
---------- > ----- sin w
k2 - 1 k2 — 1
or
sin q> < \/k.
When this condition is complied with, the sphere S intersects the rc-plane
along the circle <x, which is the locus of points M.
If sin (p = 1jk, then the sphere S is tangent to the 7i-plane at the point M,
which is the desired locus of points (this point M coincides with the point
A if and only if \jk, then the sphere S and the rc-plane do not inter
sect and the desired locus of points M is empty.
Sec. 2. Application of analytic geometry

(problems with hints and answers)
1. Plane geometry
1. Fixed in a plane are two distinct points A and B. Given: a real number
k. Find the locus of points M for each of which
2 MA2 + 3MB2 = k.
Answer. If k < — AB2, then the desired locus of points M is empty;

6
if k = - - AB2, the locus consists of a single point, P, which divides the
directed line segment AB in the ratio 3/2:
if k > —AB2, the desired locus of points M is a circle with centre at the
point P and with radius R = — 6AB2.

2. Fixed in a plane are two distinct points A and B. Given: a real number
k. Find the locus of points M for each of which
2MA2 - 3MB2 = k.
Answer. If k > 6AB2, the desired locus of points M is empty; if k

= 6 AB2, the locus of points M consists of the single point P, which divides
the directed line segment AB in the ratio —3/2:
AP _ 3,
r— _y
if k < 6AB2, the desired locus of points M is a circle with centre P

and radius R = J/6AB2 — k.
3. Fixed in a plane are three points A , P, C. Find the locus of
points M for each of which
MA2 + MB2 = MC2.
Hint. Introduce a rectangular Cartesian system of coordinates.
Answer. If the angle C of the triangle ABC is obtuse, the desired locus
of points is empty; if C = n\ 2, the locus of points M consists of the single
point D symmetric to the point C with respect to the midpoint of AB; if C
is an acute angle, the locus is a circle with centre D and radius ]fa2 + b2 — c2
(<a, b, c are the lengths of the sides of &ABC).
4. The bisector of the interior angle A of SÂBC divides the side PC into
line segments BD = 4, DC = 2. Find the lengths of the sides AB and AC
if we know that they are given by integers and that the altitude dropped
from vertex A to side BC exceeds JflO.
Solution. It is given that the bisector AD divides the side BC into seg
ments of length 4 and 2; consequently, BD/DC = 2, that is AB/AC 2.
The locus of points A for which this relation holds is a circle of Apollo
nius of radius
2ak
\'k2 ^ T \
6 2
where 2a = CB = 6, k = 2, that is, R = 4 with centre P lying
on the straight line BC and having the coordinate
k * + l 3i•—5 = 5,<
OP = a ---------=
A:2 - 1 3
where O is the midpoint of segment BC. If a rectangular Cartesian system
of coordinates is introduced and BC is taken for the x-axis, and the mid
point O of BC as the origin, then the equation of the circle (P, 4) takes the
form
( * - 5 ) 2 + /-= 1 6 .
Let us consider the straight line y = 3. This line intersects the circle (P, 4)
at the points A^5 — |/7, 3), A2(5 + )[l, 3). From this it follows that
A,C = 1^20 - 4 /7 < 3.1, A2C = K20 + 4 /7 > 5.5. Since AH> /lO >3,
it follows that point A must lie on the lesser arc AXA2of the circle (P, 4);
AXC < AC < A2C, hence 3.1 < AC < 5.5, and since it is given that the
length of AC must be a whole number, it follows that AC = 4, then AB = 8,
or AC = 5 and AB = 10.
Supplementary questions. Suppose the bisector AD of the interior angle
A of A ABC divides the side BC into segments BD = 4, and DC = 2.
(a) What is the maximum value of the altitude AH7 (Answer. 4.)
(b) Within what range do the lengths of the sides AB and AC vary?
(Answer. 2 < < 6, 4 < AB < 12.)
(c) Prove that AC and AB increase if the point A describes a circle of
AB
Apollonius---- = 2 from the point D to its diametrically opposite
AC
point.
(d) Prove that the length AD of the bisector of angle A varies from
0 to 8.
(e) Given the length m of a median emanating from A. What is a neces
sary and sufficient condition for the existence of a triangle with given
BD = 4, DC = 2, /»? How is such a triangle constructed? (Answer.
1 < m < 9.)
5. Find the images of the circles of Apollonius (see problem 11)
/ fc2 + l \ 2 , , 4a2 k 2
l x — a ---------- } + y2= --------
V k2 - I J (k2 — i)4
under the inversion I = (A ,A B 2):
X a
x + a = 4a2 4a2
(X + a)2 + Y2 (X + a)2 + T2
Under this inversion, into what do the circles of an elliptic pencil con
jugate to the hyperbolic pencil under consideration go?
Answer. (r'):(X — a)2 + Y 2 = (2ajk)2\ this is a family of concentric
circles (C^) with common centre B(a, 0); the radius of the circle (C*) is
equal to lajk. The circle E of the elliptic pencil conjugate to the pencil r with
centre at the point 5(0, s) goes into the radical axis of the circle E and
the circle of inversion, that is, into the straight line
a(X - a) + sY - 0.
6. Prove that the equations
x2 + y2 — 2px + a2 = 0,
x 2 + y 2 — 2qy — a2 = 0,
where a is a fixed positive number, and p and q are real parameters,

are, respectively, the equations of the hyperbolic pencil of circles and
the conjugate elliptic pencil of circles. What are the fundamental points
of these pencils? Into what lines do the circles r and E go under the
inversion / = (A, 4a2), where A = (—a, 0).
Answer. The fundamental points are A(—a, 0) and B(a, 0). Under
the inversion / = (A, 4a2), the circle f goes into the circle
(X - a2) + Y2 = 4a2 - ~ °
P+ a
with centre at the point B(a, 0) and radius R' = 2a P__ a . Note that
only under the values of the parameter p, p < —a and p > a, the
equation of r is the equation of a real circle.
Under the same inversion, the circle E goes into the straight line
a(X — a) + p Y = 0 passing through point B(a, 0); this is the radical
axis of the circle E and of the circle of inversion.
If we consider the circle f as a circle of Apollonius, then k and p
,, - .. A:2 + 1 , p —a 1 -
are connected by the relation p = a --------- whence-------- = — and, con-
____ k2 - 1 p + a k2
1
/
sequently, R' = 2a 1 - ---- - =
p a k
7. Prove that if
/ = x2 + y2 — 2p1 x + a2 = 0, (p = x2 + y2 — 2p2 x + a2 = 0
are two distinct circles of a hyperbolic pencil (px ^ p2) with limit points
(± a, 0), then any circle of this pencil may be given by the equation
¥ +h<P = 0,
where A and p are numbers. Conversely, for any A and p of which at
least one is nonzero, the equation A/ + pep — 0 is the equation of a circle
of a hyperbolic pencil (for A = —p ^ 0 , the equation of the midperpen
dicular of a line segment bounded by limit points).
Consider the special case of / = x2 + y2, cp = * — a.
8. State and prove a similar location for two distinct circles
/ = x2 + y 2 — 2qx y — a2 = 0, <p == x2 + y 2 — 2q2y — a2 = 0
of the elliptic pencil E.
9. Let M be an arbitrary point in the plane. Prove that the difference
between the powers of point M with respect to the circles (O) and (O')
with centres O and O' is equal to 2(00') • (KM), where K is the orthogonal
projection of point M on the radical axis of (O) and (O').
10. AAXand BB1are medians of A ABC; CC1is the altitude. The straight
lines AA1,B B 1,C C 1 form A A 2 B2 C2. Find the ratio
(A2b 2c 2)
(ABC)
(cot B — cot A)2
Answer.
3(2 cot B + cot A) (cot 5 + 2 cot A)
11. A', B', C’ are the feet of bisectors of the interior angles of A ABC.
Knowing the lengths a ,b ,c of the sides BC, CA, AB of A ABC,
find the ratio
(A 'B 'C )
(ABC)
. 2abc
Answer.------------------------------
(b + c) (c + a) (a + b)
12. A’, B ’, C are points of tangency of a circle inscribed in A ABC
with the sides BC, CA, AB. Prove that
(A'B'C ') r
(ABC) ~ 2R
13. A', B', C' are the feet of the altitudes of A ABC. Given: the angles
A, B, C of the triangle. Find the ratio
(A'B'C ')
ABC
For what necessary and sufficient condition do we have A'B'C '\[A BC f
Answer. 2 cos A cos B cos C; A'B'C'\\ABC if and only if A ABC is
obtuse-angled.
14.. Given in the plane three circles (A), (B), (C) each of which lies
outside the other two. Let P, Q, R be the centres of internal similarity
of the pairs (B), (C); (C),(A); (A),(B). Find
(PQRl
(ABC)
knowing that the radii of (A), (B), (C) are respectively equal to Rlf R2, R2.
2Ri R2 R2
Answer.
(Rs + R») (Rj + Bi) (Ri + R >)
15. Given A ABC. Through point D lying on the straight line BC are
drawn straight lines DP and DQ, which are parallel, respectively, to AB
and AC. Prove that (DQC) + (PBD) = 0.
16. Through an arbitrary point lying inside A ABC are drawn straight
lines parallel to its sides. These lines divide A ABC into six parts, three of
which are triangles with areas Slt S2, S3. Find the area of A ABC.
Answer. (][ + V S2 + ]f S3)2.

17. Let A ", B", C" be points symmetric to the vertices A , B, C of triangle
yli?C with respect to the feet of the bisectors of its interior angles. Given:
the lengths a ,b ,c of the sides BC, CA, AB of triangle ABC. Compute
the ratio
(A" B” C")
{ABC)
8abc
Answer. 3 +
(b + c) (c + a) (a + b)
18. The straight line (and also its portion inside the triangle) that is
symmetric to the median with respect to the interior-angle bisector emanat
ing from that vertex is termed the cimedian of the triangle. Given: the
lengths c and b of the sides AB and AC of A ABC. In what ratio does
the cimedian emanating from vertex A divide the directed line segment BC1
Prove that the three cimedians of the triangle intersect in one point (the
Lemoine point).
c2
Answer. — •
b2
19. In a right triangle, construct the cimedian emanating from the
vertex of the right angle.
Answer. The perpendicular dropped from the vertex of the right angle
to the hypotenuse.
20. Express, in terms of the lengths a, b, c of the sides BC, CA, AB of
A ABC, the area of A PQR, where P, Q, R are the projections of the
centroid G of the triangle on its sides.
4 (a2 + b2 + c2) S3
Answer. where S is the area of A ABC.
9a2b2c2
21. The bisectors of the interior angles of A ABC intersect the opposite
sides BC, CA, .42?respectively at the points A', Bf, C . The points A ", B", C”
are symmetric to the points A \ B', C' with respect to the respective vertices
A ,B ,C of A ABC. Given: the lengths a, b, c of the sides of A ABC.
Find the ratio
(A"B" C,r)
(ABC)
2 abc
Answer. 6
(b + c) (c + a) (a + b)
22. Let M and N be the midpoints of the medians BD and CE of A ABC.
The area of A ABC is equal to S. Compute the area of the quadrangle
HMNC.
Answer. — S.
16
23. The straight lines AO, BO and CO intersect the sides BC, CA, AB
of triangle ABC in the points P, Q, R respectively. Prove that
m . + °Q + 0A = L
~AP ~BQ C7*
24. Given two straight lines / and m lying on an oriented plane and
intersecting in the point A. Through an arbitrary point O not lying either
on a line /, or on a line m a straight line n is drawn that intersects /
and m in the points B and C respectively. Prove that the sum
1 1
(iOAB) (OCA)
does not depend on the choice of n.
25. Through the vertex A of a parallelogram ABCD is drawn an arbi
trary straight line that intersects the diagonal j3D in a point E, and the
straight lines BC and CD intersect in points F and G respectively. Prove
that the line segment AE is the mean proportional between the lines EF
and EG.
26. Let a, p, y be points symmetric to some single point O with respect
to the midpoints of the sides BC, CA, AB of a triangle ABC. Prove that
the straight lines Act, Bp and Cy pass through the same point P. Also
prove that if point O describes some line r , then point P describes a
line r f that is homothetic to line F. Where does the homothetic centre lie?
What is the homothetic ratio?
Answer. The homothetic centre lies in the centroid M of A ABC;
MP 1
MO
27. A straight line / intersects the sides BC, CA, AB of a triangle ABC
in the points , j8, y respectively. Let a', P', y' be points symmetric to
cl
the points a, p, y respectively about the midpoints of the sides BC, CA, AB.
Prove that the points a', P', y' lie on one straight line.
28. Let A 'B 'C be a triangle that is obtained if through each vertex
of a triangle ABC we draw a straight line parallel to the opposite side;
ct, p, y are points taken respectively on the sides BC, CA, AB. Prove that
if the straight lines Act, Bp, Cy pass through a single point, then the straight
lines A' , B'P, C y also pass through one point.
cl
29. Drawn through the midpoint of each diagonal of a convex quad

rangle is a straight line parallel to the other diagonal. The point of inter
section of the lines thus drawn is joined to the midpoints of the sides o^
the given quadrangle. Prove that the quadrangle is thus partitioned into
parts of equal size.
30. The points P and Q divide the directed sides BC and CA of A ABC
in the given ratios Aand fi. Suppose the straight lines AP and BQ intersect
in a point O. Find the ratios:
(!) (2) (3).< 4 2 ® ,
(ABC) (ABC) (ABC)
Answer. (1) ------ - ------ ; (2)------- ——----------- ; (3) ---------- -

1 + A + A/r (1+A) (1 +A+A/i) (1 4~a0 (1 +A+A/z)
31. A certain point O lies in the plane of a parallelogram ABCD
(AC and BD are the diagonals). Prove that:
1°. If the point O lies inside the parallelogram ABCD, then the sum of
the areas of A OAB and A OCD is equal to the sum of the areas of the
triangles OBC and ODA.
2°. No matter where the point O lies in the plane of parallelogram ABCD>
AO AC is equal in size either to the sum or the difference of A OAB and
A OAD.
32. Points A', Br, C' are taken on the sides BC, CA and AB of A ABC.
Let Al9 Bl9 Ci and A2, B2, C2be the images of the points A, B, C respectively
under homothetic transformations with the same homothetic ratio and
with homothetic centres in the points C , A ’, B' and B', C', A Prove
that A A ^ C x and A A2B2C2 have one and the same centroid.
33. Let M1 and M 2 be two arbitrary points lying on side BC of A ABC,
and let N be an arbitrary point lying on side AB. Denote the points of
intersection of the straight lines NMX and NM 2 with side AC by Px and P2
and the points of intersection of the straight lines PMX and PM2 (where P
is an arbitrary point of the straight line AC) with the side AB by Q±and Q2.
Prove that BC, PxQ2, and Qi^ 2 belong to the same pencil.
34. Suppose P and Q are two points lying in the plane of A ABC;
AiBxCx and A2B2C2 are triangles symmetric to A ABC atjout the points P
and Q respectively. Let oe, ft, y be points of intersection of the straight
lines AxA29 BxB2, CxC2 respectively with the lines BC, CA, AB. Prove
that the points c l, p , y are collinear (Fig. 7).
Proof. Let a', fi', y' be points of intersection of the straight lines A a.
Bp, Cy with PQ. Then a', P’, y' are the midpoints of the line segments
Act, BP, Cy. These points lie on the sides of A A'B'C', where A', B', Cr
are the midpoints of the sides BC, CA, AB, respectively. Under the homo
thetic transformation (G, —2), where G is the point of intersection of
the medians of A ABC, the points a', /?', y' go into the points a", p", y",
which also lie on one straight line and on the lines BC, CA, AB, respec
tively. Here, Coe = 2B'ot', Bet" = 2B'ol' and, hence, Ca = Bctn, whence
and similarly for the other two sides. Since the points a", /?", y" are colli-
near, it follows that
and, hence,
— » — >— >
Bcc CP Ay _
— » — >— > ~
otC PA yB
Consequently, the points a, /?, y are also collinear.
35, Let the straight lines A and A' intersect the sides of A ABC in
the points Al9 Bl9 Cx (line A) and A'x, B[, C[ (line A'), respectively. Prove
that the lines A ^ , B ^ i, CXA[ intersect, respectively, the straight lines
AB, BC, CA in the points p s,p » p * lying on one line and the lines A[Bl9
B[Cl9 C[Aj intersect the lines AB, BC, CA respectively in the points Qz,
Qx, C?2> which also lie on one straight line. The straight lines PiP2Pz and
QiQzQz are called the Brocardians XRrocard lines) of the straight lines A
and A' with respect to A ABC.
36. Given: the lengths a, b, c of the sides BC,CA, AB of A ABC.
Through point C is drawn a bisector CC0 of the interior angle C. Drawn
through point A is the median AA0 to the side BC and through point B
the altitude BB0 to side CA. These lines form A AxBxCx. Find the ratio
(ABC)
J [b(a2 + b2 - c2) - a(b2 + c 2 - a2) ] 2
Answer. -------------------------------------------------------------—— •
(a2b + bz — be2 + lab2) (— a 2 + 3b2 + c2) (a + lb)
37. Let A ^ C x and A2B2C2 be triangles that are the images of A ABC
under the homothetic transformations (P, k), (Q, k). Denote by a, /?, y
the points of intersection of the straight lines AXA2, BXB2, CXC2 with the
straight lines BC, CA, ^42? respectively. Find the ratio
(aft?)
(v45C)
Answer. —&(& + 1).
38. Prove that if three diagonals of a hexagon (not necessarily convex)
have a common midpoint, then any two of the opposite sides are parallel.
39. From an arbitrary point Ax lying on side BC of a triangle ABC
draw a straight line A-fix parallel to BA up to intersection with CA at
point Bx\ then draw a straight line BxCx parallel to BC up to intersection
with line AB at point Cx and, finally, draw line CXA2 parallel to AC up
to intersection with line BC at point A2. Prove that if Ax is the midpoint
of segment BC, then points Ax and A2 coincide; otherwise, continue the
process, that is, draw line A2B2 parallel to BA, line B2C2 parallel to CB,
and line C2A3 parallel to AC. Prove that the path closes on itself, that is,
point A3 coincides with point Av
40. Join vertices A, B, C, D of a parallelogram ABCD with the midpoints
of the sides BC, CD, DA, AB so that a parallelogram is formed that lies
inside the parallelogram ABCD. Prove that the area of the thus formed
parallelogram is equal to 1/5 the area of the parallelogram ABCD. Another
such parallelogram is formed if we join the points A, B, C, D with the
midpoints of the sides CD, DA, AB, BC. Prove that the common part of
these two small parallelograms is a centrally symmetric octagon and has
an area equal to 1/6 that of the parallelogram ABCD.
41. ABCD is an arbitrary convex homogeneous lamina. Each of the
sides of the quadrangle ABCD is divided into three equal parts: AAX =
= A2A2 = A2B, BBx = BxB2 = B2C, CCx = CXC2 = C2D, DDX = D±D2 =
D2A. The straight lines A2Bly B2CX, C2Dl9 D2AX form a parallelogram.
Prove that the centroid of the lamina ABCD coincides with the point of
intersection of the diagonals of the parallelogram.
Hint. Take the diagonals of the parallelogram for the coordinate axes.
42. Given, with respect to a general Cartesian system of coordinates,
two points Mx(xl9y^, M2(x2,y 2) and the straight line ax + by + c = 0 .
It is given that the points M x and M2 do not lie on the given straight line
and that the third line M XM2 intersects the line ax + by + c = 0 at some
point M. Find the ratio X in which the point M (x,y) divides the line
----- ►
segment MXM2.
A , axx + byx + c _r. *1 + Xx2 y x + Xy2
Answer. X = ------ - - - ■- — . Hint. jc - ---------“ , y =
ax 2 by2 + c 1 -f-A 1 -f- ^
these coordinates must satisfy the equation
ax + bx + c = 0 .
43. Taking advantage of the result of the preceding problem, prove
that if the straight line / does not pass through any one of the vertices of
triangle ABC and intersects its sides BC, CA, AB respectively at the points
P, Q, R , then the product of the ratios
PC QA RB
at which the points P, Q, R divide the directed line segments BC, CA, AB,
is equal to —1 :
Xpv = — 1 .
Hint. Introduce into the plane of the triangle ABC a general Cartesian
system of coordinates; in this system, let A = (xx, y x), B = (x2,y 2),
C = (x2, ^3) and let ax + by + c = 0 be the equation of the straight
line /. Then
, ax 2 + £y2 + c . + 6*3 + c axx + byx + c
A— 1 > fi * ^ '*
ax 3+ + c axx + b y x + c ax2 + by2 + c
Remark. The converse is true as well: if Xfiv ~ —1, then the points
P, Q, R which divide directed segments BC, CA, AB in the ratios X, \i, v
are collinear. Indeed, suppose the line PQ intersects AB in the point R'.
Denote by v' the ratio in which the point R' divides the directed line
segment AB. Then X\iV = —1. From this and from the equation Xfiv =
= —1 it follows that v = v', that is,
RB R'B
Hence, the points R and R' coincide.
44. A second-order curve given by the equation
F(x, y) — ax2 + 2bxy + cy2 + 2dx + ley + f = 0
does not pass through a single vertex of the triangle ABC and intersects
the sides BC, CA, AB at the points A l9A2; Bl9B2; Cl9 C2 respectively.
Prove that the product of the ratios in which the points Al9 A2; Bx, B2;
Cl9 C2 respectively divide the directed line segments BCf CA, AB:
AC2
*i =
C2B
is equal to 1:
X1Xi n1n2 v2 v2 = 1.
Conversely: if on the sides of triangle ABC are chosen points Alf A2
(on BC), Bl9 B2 (on CA), Cl9 C2 (on AB) and if the product of the ratios
in which these points divide the directed line segments BC, CA, AB is
equal to 1, then the points Ax, A2, Bl9 B29 Cx, C2 lie on one and the same
second-order curve.
Hint. The numbers Xx and k2 are found from the equation
J *1 + Xx2 ' y x + Xy2 \ = Q
I 1+ A I +X j
where B(xlt >>,), C(x2,y 2) or
iTTFj
( * i + Xx2 \ 2 , + Xx2 y t + ).y2 , „ + Xy2 \ 2
+ 2d X2
1 +X
+ 2e * ± ± * * 1. + f ,
1 +X
or
F{x2, yi) P + • •. + F(xx, yx) = 0.
By Yieta’s theorem.
F (x x, y i)
^>1^>2 —
F(X2, y 2)
and so on.
The proof of the converse is similar to the proof given in the remark
concerning problem 43.
45. Suppose an algebraic curve / of order n does not pass through
a single vertex of A ABC and intersects each of the sides of the triangle,
BC, CA, AB in n points A if Bi9 Ct (/ = 1,2, . . . , n). Then the product of
the ratios
BA§ CBt AC{
A f i BiA C(B
is equal to (—1)" ( Carnot's theorem). Does the converse hold true?
Hint. The proof is similar to that given in the hint referring to prob
lem 44. Generally speaking, the converse for n > 2 is not true.
46. Given in the plane are m points p. (i = 1 , 2 , . . . , m). An algebraic
curve / of order n does not pass through a single one of the points Pt
and intersects the straight lines PXP^ iV V - • •, Pm-1 PmPi ln n points
respectively
A ? \ A f \ A j ”~\ AW (/ = 1, 2, . . . , m).
Find the product of the ratios
4 n- 1} Pm ^P l
Answer. (—l)mn (a generalization of Carnot’s theorem; see problem 45).
47. Al 9 A 2 9 A3; Bl 9 B2 9 B3; Cl 9 C2, C3 are arbitrary points which lie,
respectively, on the sides BC, CA, AB of A ABC and are the interior
points of the sides. Introduce a general Cartesian coordinate system into
the plane of A ABC. Since the general equation of a third-order curve
contains 10 coefficients, there exists a third-order curve that passes through
9 points A i9 Bi9Ci (/ = 1,2,3). By the Carnot theorem, the product of
the ratios in which the points A i9 B i9 Ct divide BC 9 CA9 AB, respectively,
is equal to (—l )3 = —1, and yet all these ratios are positive end their
product cannot be a negative number. Wherein lies the error?
2. Solid geometry
1. AXA 2 A 3 A 4 is an arbitrary tetrahedron; Bl 9 B 29 B39 B4 are the centroids
of its faces A 2 A3 A4, AÂ3 A4, AiA 2 A49 AÂ2 A 3 9 Ci29 Ci3, C*4, C239 C249 C3 4
are midpoints of its edges AXA 29 AyA 39 AXA 49 A 2 A3, A 2 A 49 A 3 A4. The straight
lines AxBl 9 A 2 B2, A 3 B39 A 4 B4 (and also the line segments themselves) are
called medians o f the tetrahedron A ^ A q A ^ The straight lines Q 2C34,
C1 3 C24, C2 3 C1 4 (and also the line segments themselves) are termed bimedians
o f the tetrahedron A 1 A 2 A 3 A4.
Prove that four medians and three bimedians of the tetrahedron AXA2A3A4
pass through one and the same point G, called the centroid o f the tetra
hedron A1A2A3A4; here, A f i : GBt = 3 :1 and the bimedians are bisected
by the point G. Assuming the radius vectors of the points A x are equal
to OAt = ti (i = 1,2, 3, 4), find the radius vectors r l7 of the midpoints
of the edges AxAj, the radius vectors rijk of the centroids of the faces AxAjAk,
and the radius vector r of the centroid G.
_ r, + iy + rk
Answer, r.. *,• + *j
rijk T = ri + r2 + r3 4" r4
2. Prove that the six planes passing through the edges A xAj of the tetra
hedron Ax A 2 A 3 A 4 and the midpoints of the opposite edges pass through
the centroid G of the tetrahedron.
Hint. A iA jC ki and A kA xCij intersect along the bimedian CklCxj.
3. Prove that the six planes that pass through the edges of the tetra
hedron and divide its volume into two intersect in a single point.
4. OA, OB, OC are the edges of a parallelepiped; A', B', C , O’ are the
vertices symmetric to the vertices A, B ,C ,0 about the centre of the paralle
lepiped. Prove the following statements:
1°. The diagonal OO' of the parallelepiped is divided into three equal
parts by the planes ABC and A'B'C'.
2°. The diagonal OOf intersects the planes of the triangles ABC and
A'B'C' in their centroids.
Hint. Set O = (0, 0,0), A = (1,0, 0), B = (0, 1, 0), C = (0,0, 1).
5. Let Ga, Gb, Gc, Gd be the centroids of the faces BCD, CDA, ABD,
ABC of the tetrahedron ABCD. Prove the following statements:
1°. If a straight line k intersects the faces BCD, CDA, ADB, ABC in
the points coa, cob, (oc, o d, respectively, then the midpoints of the segments
A oa, Bcob, Co)c, D od are coplanar.
2°. The tetrahedrons A 1 B1 C1 D1 and A 2 B2 C2 D2 are images of the tetra
hedron ABCD under the homothetic transformations (Pl 9 —1/2), (P2, — 1/2),
where P 1 and P2 are arbitrary points. Prove that the straight lines AXA2,
B±B2, C±C2, DxD2 intersect the faces BCD, CDA, DAB, ABC in the points
a, p, y, 3 lying in one plane.
Proof. Let G be the centroid of the tetrahedron ABCD; let a', /?', y', 3'
be the points of intersection of the straight line PXP2 with the straight
lines Aa, Bp, Cy,D3; let wa, (ob, wc, cod be points at which the following
lines intersect: Got and Gaa'; Gp and Gbp'; Gy and Gcy'; G3 and Gd3'.
— »— »— >— >
The points a', jS', / , 3' divide the directed line segments A a, BP, Cy, D3
in the ratio of 2 : 1 and therefore lie in the planes of the faces of the tetra
hedron Ga Gb Gc Gd. The point G divides segment AGa in the ratio of 3 : 1,
and therefore coa is the midpoint of segment Gaot' and Got = 3Go>a. On the
basis of item 1°, the points coa, cob, coc, a)d lie in one plane, and hence so
also do the points a, p, y, 3 obtained from the points coa, cob, coc, a>d under
the homothetic transformation (G, 3).
3°. The images co', a)b, w'c, co'd of the points coa, cob, coc, cod under the
homothetic transformations (Ga, — 1/2), (Gb, — 1/2), (Gc, —1/2), (Gd, — 1/2)
are coplanar.
AG ota' 3 G jx
Hint. See item 2°, —— =
2
A'Ga GGa a'A " \ GaA'
6 . Given: four arbitrary points A ',B ', C',D ' lying, respectively, in the
faces BCD, CDA, DAB, ABC of a tetrahedron ABCD. Let (A l 9 Bl 9 Cl 9 DJ,

(A2, B2, C2, D2), (A3, B3, C3, D3) be images of the points A, B, C, D under
a homothetic transformation with ratio k and, respectively, with centres
(D', A', B', C'), (C\ D', A', B'), (B \ C', D', A') so that
D'A1 _ A'B1 __ KCi C'D1 7 C'A

~~ ^ ^ = k and so forth.
D'A A'B B'C CD CA
Prove that the tetrahedrons Ax B1 Cx Dl9 A2B2 C2D2, A3Bz C3 Dz have a
common centroid.
7. Let Pbe an arbitrary point lying inside a tetrahedron ABCD. Denote
by A ',B ',C ',D ' the points of intersection of the straight lines PA, PB,
PC, PD with the opposite faces. Prove that
+ " + £ ? > 12,
PA' PB' PC' PD'
AP BP CP DP nt
-----------------------------> 81.
PA' PB' P C PD'
Under what condition do we have equality?
8. Let a and a', b and b', c and c' be the respective points in which
the edges BC and AD, CA and DB, AB and DC of a tetrahedron ABCD
intersect an arbitrary plane. Denote by Ga, Gb, Cc, Gd the centroids of
the triangles a'be, b'ca, c'ab, a'b'c'. Construct the points A', B', C', D'
such that
A A' = 3AGn, BB' = 3BGb, C C = 3CGr, DD' - 3DGd
Prove that the points A', B', C , D' lie in a single plane.
9. T = ABCD is an arbitrary tetrahedron; a/ty<5 is spatial quadrangle
whose sides are equal and parallel to the medians of the given tetrahedron.
Through the midpoints of the sides of the quadrangle a/fy<5 draw planes
parallel respectively to the faces of the tetrahedron (so that if a/? # A A',
where AA' is a median of the tetrahedron ABCD, then the plane passing
through the midpoint of segment a/? is parallel to the plane BCD and so
on). The four planes thus drawn form a tetrahedron Tv Prove that the
volume of Tx is eight times that of T and that the tetrahedrons T± and
cuflyd have a common centroid.
10. Draw through the vertices A ',B ', C ',D ' of a tetrahedron A'B'C'D'
parallel lines dx, d2, dz, dA. Let ABCD be a tetrahedron homothetic to
the tetrahedron A'B'B'D' under a homothetic transformation with ratio k.
Denote by Al9 Bl9 Cl9 D1 the points of intersection of the lines dl9 d2, d3, d4
respectively with the planes BCD, CDA, ABD, ABC. Prove that
Ai Bx Ci Dx = - k \ 2 k + 1) ABCD .
11. Through the vertices A, B,C, D of a tetrahedron ABCD draw
planes a, b, c, d parallel to some plane m. Through the vertices A', B', C 9 D 9
of the tetrahedron A'B'C'D' draw planes a', b ' 9 c\ d' parallel to some plane
m'. Here the planes m and m' are chosen so that the straight lines (a9 a'),
(b, b'), (c9 c' ) 9 (d, d') lie in one plane (the P-plane). Prove that when the
planes m and m' are changed (however, the change must be such that
coplanarity of the indicated four lines is preserved at all times), the P-plane
will rotate about a fixed point.
12. Straight lines joining the vertices A, B, C9D of a tetrahedron ABCD
with point P intersect its opposite faces in the points A \ B \ C ' 9 D'. On
segments A A \ B B \ CC", DDf consider points A \ 9 &19 Cl9 D1 such that
AA’
= k
A ,A 1
and the points A2, A3, A 4 in which the straight lines AxBl 9 AxCl 9 AXDX
intersect the plane BCD. Prove that the areas of the triangles A 2 CD, A 3 DB,
A4BC are equal. Consider the case k — 2.
13. Given a tetrahedron T = ABCD and four straight lines a ,b 9 c,d
that pass through one and the same point P and are parallel to the four
given lines. Let line a intersect the planes BCD, CDA 9 DAB, ABC in the
points ll 9 m4 9 n3 9 p2; line b in the points l29 mx, h4,/?3, line c in the points
/3, m29 nl 9 /?4, and line d in the points /4, w3, n2 9 pv Prove that, generally,
there exists only one position of point P under which the quadruplets
of points (ll 9 ml 9 nl 9 px) (l2 9 m2 9 n2 9 p 2 ) 9 (/3, m3, w3, /?3), (/4, w4, w4, p4) are
coplanar. Consider the case where the straight lines a9 b9 c9 d are parallel
to the medians of the tetrahedron ABCD.
CHAPTER III
THE USE OF COMPLEX NUMBERS

IN PLANE GEOMETRY
Sec. 1. Solved problems
Before examining the problems, make a brief study of the theoretical
material given in Chapter V, Sec. 4.
Since the method applied here is not readily available and is effective
and simple in the solution of many problems of plane geometry, we have
selected a large number of problems and have provided solutions; there
are also exercises in the unaided solving of problems (with hints and answers)
Problem 1. ABC is an arbitrary triangle, G is the point of intersection
of its medians (centroid); H is the point of intersection of the altitudes
(orthocentre); O is the centre of a circumscribed circle (O) = (ABC)',
A l9 Bx, Cx are the midpoints of the sides BC, CA, AB respectively; A2, B2, C2
are the feet of the altitudes; Az, B3, C3 are the corresponding midpoints
of the line segments AH, BH, CH (Euler’s points); A4, B4, C4 are the corres
ponding points symmetric to the orthocentre H about the straight lines
BC, CA, AB.
Prove that:
1°. The points O, G, H are collinear and OH = 30G.
2°. The points Al9 Bl9 Cl9 A2, B2, C2, A3, B3, C8 lie on one and the same
circle (0 9) (Euler's circle or the nine-point circle o f A ABC). The centre
of the circle 0 9 is the midpoint of the line OH, and the radius of the
circle (0 9) is equal to R/2, where R is the radius of the circle circumscribed
about the triangle ABC.
3°. The points A4, B4, C4 lie on the circumscribed circle.
Solution 1°. Take (ABC) for the unit circle. Let a, b, c be the corres
ponding affixes of the points A, B, C. Prove that a + b + c is the affix
of the orthocentre H of triangle ABC.
The slope of the straight line BC is
c —b c —b
x —-------- — ------------- = — be.
E -b J ____ 1
c b
The 6lope of the line joining vertex A(a) and point M and having the
affix a + b + c is
b+ c b+ c
x — be.
b c
Complex Numbers in Plane Geometry 83
whence
x -j- x 9 = 0 .
Hence, AM ± BC. In similar fashion it is proved that BM _L CA and
CM ± AB. Hence, M —the point with affix a + b + c—coincides
with point H (the point of intersection of the altitudes of A ABC).
Furthermore, the affix of point G is
a+ b+ c
--------------- 5
3
whence and also from the fact that the affix of point //is equal to a + b + c
it follows that
OH = 30G.
That is, the points O, G, H are collinear.
2°. The midpoints Q of the sides BC9 CA 9 AB have affixes
b+ c c+a a+ b
ax —-------- 9 b1 = --------* Ci —--------•
Let E be the midpoint of line OH. Its affix is

a+ b+ c
e -------------------
2
Since
s — ax = aj2 9 e — bx = bj2, e — c1 = c/ 2
and \a\ = |6 | = \c\ = 1, it follows that
Ifi - flil = Ifi - bx\ = |6 - cx\ = 1/2 = Rj2 (R = 1)
that is.
EAX= EBX= ECt = R/2.
Thus, the midpoints A l 9 B l 9 Cx of lines BC, CA9 AB are at equal dis
tances—R/2—from the midpoint of line OH and, hence, lie on circle
(A ^ C x ) with centre E((a + b + c)/2) and radius R/2.
Furthermore, the equations of the straight lines AH and BC are of
the form
z — a = bc(z — a),
z — b = — bc(z — ~b)
or
z — bcz = a — abc,
z + bcz = b + c.
Forming the half-sum of these equations, we find the affix z= a 2of point ^ 2:
a+ b+ c be
-------p be
a2 = -- G__
2 la la ’
whence
be 1 - JL
EA2 = |a2 — e| =
2a 2 ”2
(|a| = \b\ — \c\ = 1). Similarly, proof is given that
EB2 = R/29 EC2 = R/2.
Thus, the points Al9 Blf Cl9 A2, B2, C2 lie on a circle whose centre is the
midpoint of line OH and whose radius is equal to R/2.
Furthermore, the affixes of the midpoints A3, B3, C3 of lines AH, BH, CH
are, respectively,
ci Q b c Q
a3 = -------- = e H---- >
whence
EA3 = \a3 - e| = \a/2\ = 1/2 = R/2
and, similarly,
EB3 = ECZ = R/2.
Thus, all nine points Ak, Bk, Ck (k = 1,2, 3) lie on the circle (E, R/2)9
whose centre E is the midpoint of line OH and whose radius is R/2.
3°. Let us now go back to the equation of the straight line AH,
z — a = bc('z — a),
that passes through point A perpendicular to BC. Now find the affixes
of the points of intersection of this line with the unit circle zz = 1. Provided
\z\ = 1, we have
. \/ ----------
z — a = be 1 1
\ z a
or
z —a
z — a = — be
az
One of the roots of this equation is z = a (the affix of point A); the other is
be
the fact that point N with this affix belongs to the unit circle follows
he ‘
from the equality 1
- H
We will prove that the midpoint of line HN coincides with point A2.
Indeed, the affix of this midpoint is
, , , be
a -r b -V c— — ,
a be
---------- ~------ — = 8 - — = *2-
2 2a
Thus, point N coincides with point A4.
In similar fashion, proof is given that the altitudes emanating from
vertices B and C intersect the circle {ABC) in the points 1?4 and C4,which
are symmetric to the orthocentre H with respect to the sides CA and AB.
Problem 2. (Boutain points). Suppose A ABC is inscribed in the unit
circle (0). Let zl9 z2, z3 be the affixes of the points A ,B ,C . A Boutain
point for A ABC is a point such that when it is chosen as the unit point
(that is, a point whose affix is z 1), the equation cr3 = z1z2z3 = 1 holds
true. Prove that for a given circle (0) and A ABC inscribed in it, there
are three Boutain points on the circle (0) that form an equilateral triangle.
Proof. Take the point a (|a| = 1) for the new unit point. Then the new
affixes of the points A, B, C will be — , — 9 — • The point a will be
a a a
the Boutain point if the product of these affixes is equal to 1:
Z1 Z2 Z30 — 11 or a31 = (7,
------
This equation has three roots:
3, £21^3.
3 __
where /cr3 is any one of three values of the cubic root of <r3, and e is
3
any one of two imaginary values ^1, that is,
- I + 1/3 -1 -//3
either £ = ------ -—-— or e = --------— -—
For example, let
e = — f_ + 1 /3 _ cos ]2 0 o j sjn |2o°,
then
c2 = cos 240° + i sin 240°,
and to construct the Boutain points it is
necessary first to construct the point for the
3 __
chosen value )[<t3, then multiply it by £ (a
rotation through 120°) and then once again
by e (another rotation through 120°). We
obtain the vertices of an equilateral triangle
with affixes
3 ___ 3 __ 3 ____
f<73, 6jfo3, e2]faa.
Problem 3. Given an arbitrary triangle ABC. Take on the circle (ABC)
an arbitrary point M (Fig. 8). Let Al9 Bx and Cx be the orthogonal pro
jections of M on the straight lines BC, CA, and AB. Prove that the points
Al9 Bl9 Cx are collinear. Taking the circle (ABC) for the unit circle and
assuming that the affixes of the points A9B9C, M are equal, respectively,
to zl9 z29 z39z09 set up the equation of the straight line A ^ C x (the Simson
line for the point M with respect to the triangle ABC).
Solution. The equation of the line BC and the straight line passing
through point M perpendicularly to BC is of the form
z — z2 — z2z3(z — Z2),
Z Zq z2z3(z %o)
or
z + z2z3z = z2 + z39
Z Z2Z3Z — Zq Z2Z3Zq.
Forming the half-sum of these equations, we find the affix z = ax of the

orthogonal projection of point M on the straight line BC:
= — (z0 + Z2 + z3 - Z2Z3Z0). (*)
2
Similarly we can find the affix bt of point Bt :
bx = -(z0 + z3 + Zj — z3z1"z0).
2
Let us find the slope of the straight line A1B1; we have
— b\ — — [z2 Zj z3z 0(z2 Zj}]
\ , w, -X 1 / 1 z3 \ 1 (z2 - Zj)(z0 - Z3) _

= -2- ( I , - * , X I - V i - -2 f e — ) ( 1 - - ) = 7 ---------- •
a, - J , - - ( 1-----M ( - --------- 1 ) - — - Zi)(z. - z j —

2 Vz2 z j [ z 0 z3 / 2 zxz2z3
and so
*_ A = *l 03ZO*
ai bi zo
The equation of the straight line AXBX may be written as
Z Cl i — G3Z q{ z &i)
or, taking into account equation (*),
z — “ (z0 -\- z2 z3 z2z3z 0)
„ . A _ r * _ i ( ± + j . + ± _ i ) i .
L 2 \ Z0 z2 z3 z 2z 3 / J
This equation can be simplified:

1 / / Z2Z3\
z —
• <73 Z 0Z = -- (z0 + Z2 + Zs ---------------
2 V z0)
Z iZ 2Z3 /
— + — + —
2z0 I Z 0
> Z2 Z3
or
1( | i z2z3 <t3 ZXZ3 ZiZ2
z tr3z0z I Z 0 + z2 + z3 —
2 V zo zo zo zo
or
z — asz 0z = — (z0 + ox — <r2z0 — 0**0). ( 1)
2
where
<Tl — ~h ®2 — ^1^2 “I" ”1“ Z3Z1, O* —
The symmetry of this equation with respect to zx, z2, zs permits asserting
that line also passes through point C,. Incidentally, this becomes
evident by substituting into the left and right members of equation (1)
the affix
Cl = -- (z0 + zx + Z2 — ZjZjjZo)
2
of point Cj (the results will be the same; check this!) or by seeing that
the following equation holds:
aj ax 1
bx bi 1 = 0
cx c\ 1
(this is left to the reader to verify).
Ifz0is the unit point, then the equation of the Simson line (1) assumes
the form
z -< 7 3z = — ( l + a 1 — a2 - ( 7 3), (2)
and if we take the Boutain point for the unit point (that is, <x3 = 1), then
z - z = j ((ti - <r2). (3)
Remark. It will be proved below that if point M does not lie on the
circle (ABC), then its projections Al9 Bl9 Cx on the sides BC, CA, AB
do not lie on one straight line.
Problem 4. Prove that if for the unit point we take the Boutain point M
with respect to &ABC inscribed in the unit circle (O), then the Simson
line corresponding to point M will be collinear with the diameter of the
unit circle passing through M (see problems 2 and 3).
Solution. The equation of the Simson line will have the form
z-z=j((7 (3)
which means it will be collinear with the x-axis or the diameter of the unit
circle passing through point M because M is the unit point of the A>axis.
Note the converse: if the Simson line for point M of (ABC) with respect
to A ABC is collinear with the diameter of the circle (the diameter passing
through M), then M is the Boutain point.
True enough, for if M is the unit point, then the slope of the Simson
line is equal to <r3 and if the Simson line is parallel to the diameter passing
through M , that is, parallel to the x-axis, then o3 = 1, since the slope
of the x-axis is equal to 1.
Problem 5. Let us consider A ABC inscribed in the unit circle (0).
Prove that:
1°. The point P with affix <x2 is symmetric to the orthocentre H of A ABC
with respect to the diameter 8 of the unit circle, which diameter is parallel
to the Simson line constructed for the unit point with respect to A ABC.
2°. The point Q with affix a3 is symmetric to the unit point with respect
to the diameter 8.
Solution. 1°. The equation of the diameter 8 is of the form
Z — <73Z — 0
[see equation (2) of problem 3]. The affix of the projection of the ortho
centre H on this diameter is found from the system of equations
z — o^z = 0,
z — o1 = — cr2(z — d±)
or
z — g 3z = 0,
z + g 3z = + o 3g ±
or, since dx = — , it follows that
Z — <73Z = 0,
Z + ff3 Z = C7X + G2.
Adding these equations, we find the affix A of the projection of the ortho-
centre H on the diameter S:
X— G2
2
The affix co of the point symmetric to point H with respect to the di
ameter S is found from the equation
(O + Gx Gx + <72
2 2 ” ’
whence
co = g 2.
2°. The equation of the perpendicular dropped from the unit point to
the diameter 8 is of the form
z — 1 = — g z ( z — 1).
Solving this equation together with the equation of the unit circle zz = 1,
we obtain
z
The roots of this equation are: z = 1, z = g 3 .
Problem 6. 1°. Let A l9 Bl9 CL be the orthogonal projections of point P
on the sides BCy CAf AB of A ABC. We take the centre O of {ABC) for
the origin. Let Rzl9 Rz2, Rz3 be the affixes of the points A , B9C, where R
is the radius of the circle (ABC) = (|zx| = \z2\ = [z3| = 1). Let p be the
affix of point P. Find the affixes al9 bl9 cx of points Al9 Bl9 CL and then
express the area ( A ^ C J of the oriented A A lBlC1 in terms of the area
------>
{ABC) of the oriented A ABC9 in terms of the lengths a9b9c of the sides
BC9CAy AB and in terms of the affix p of point P.
Consider the following special cases:

2°. The point P coincides with the centroid G of A ABC.
3°. The point P coincides with the centre I of the circle inscribed in
A ABC.
4°. The point P coincides with the centre O of circle (ABC).
5°. The point P coincides with the orthocentre H of A ABC.
Solution 1°. The equation of line BC is
z — z2R = — z2z3(z — z 2R) (4)
z + z2z3z = R(z 2 + z3)

or
z + z2z3z = R(ol — zx). (5)
The equation of the perpendicular dropped from point P to line BC is
of the form
z - p = z2z3(z - p)2
or
2 Z^Z^Z = p Z2Z3/7
or
z — z2z„z = p — ZyOzp. (6)
Forming the half-sum of equations (5) and (6), we can find the affix z = ax
of point Ax:
ox = — (R<rx + P — Rzx — z xoip).
Similarly,
bx = * (Ro 1 + P ~ Rz2 — z 2o3p),
Ci = - - (Rox + p — R z3 — ZaPsP)-
From this we get
— (Rcr1 + p —Rzx—~zx<j3p) i (Rdx+ p —R z x—zxd3p) 1

2 2
Rzt -f z xa3p R z x + zfixp 1

/
Rz 2 + z 2a3p R z 2 + z2d%p 1
16
Rz 3 + z 3<T3p R J3 + z3d3p 1
Zj Z i 1 z"l Zj 1
l
/p z 2 z 2 1 + z 2 z2 1
16
Z3 Z 3 1 Z3 z3 1
W - PP R* _ Q p 2
(ABC) = -------------(ABC).
4R2 4R2
Note that OP2 — R2is the power o> of point P with respect to the circle
(ABC). Therefore we finally have
(A & C J = ---- (ABC).

4R2
2°. If point P coincides with point G (the point of intersection of the
medians of A ABC), then
OP2 = OGz = pp = — (Rzy + Rz., + R z3) (R~z\ + R z 2 + R z 3)
R2
= — (3 + zt z 2 + z 2z x + z 2z 3 + z3z 2 + z3z x + ZxZ3).
But
a 2 = (Rz 2 — Rz3) (R z 2 — RY3) — R2[2 — (z2z 3 + z 3z 2)],
whence
Z2z 3 z3z 2 — 2
R2
and, similarly.
z3Zj -)- z1z 3 — 2
R2
zxz 2 + Z2Z l = 2 -
IP
SO that
/j2 bp rp\ a2 + b2 + c2
3 + 2 - — + 2 -------- 1-2--------- 1 = R2 -
~ -t ( R2 R2 * 2/
Consequently,
a2 + b2 + c2
R2 - OG2 =
and therefore
sp -1
- hP -4- (p
But
R °bC
4\(ABC)\
hence
3°. If point P coincides with the centre / of the circle (I) inscribed
in A ABC, then pp = OP2 — Ol2 = R2 — 2/?r (Euler's formula, see
problem 33 below) and, hence.
_ (jR2 „ 2/?r)
(^ A C ,) (^ 5 C )= — (ABC)
4/P 2R
<, 1 ( ^ 0 1
a° c abc(a + b + c)
2M2fC)|
Thus,
4Q4,8C)3
(M C i)
abc(a + 6 + c)
Remark. Note the formula
(^ A A ) _ r
(,4£C) 2/? ’
where AX, B X, Cx are the projections of point / on the sides 2?C, C4, ,41?.
4°. If the point P coincides with the point O, then pp = OP2 = 0 and,
consequently,
( A ^ C ,) = — (^BC) = A (^BC)
4/?2 4
(this is also immediately clear on the basis of elementary-geometry reason-

ing).
5°. If the point P coincides with the orthocentre H of triangle ABC,
we have
R2 — OH2 R2 — (3OG)2
(AJB1C1) = (ABC) = (ABC)
4R2 4R2
a2 + b 2 + c 2)
R2 - 9 I R 2 -
a2 + b2 + c2 - 8R2
(ABC) = (ABC)
4R2 4R2
f a2 + b 2+ c2 a2 + b2 + c2
- 2 \ (ABC)
V 4R2 a2b2c2
4(ABC)2
b2 Q2
Since —° — sin2A, = sin* B, -----= sin2 C, it follows that
4R2 4R2 4R2
(A1B1C1) a2 -4- W - L
------------ — 2 — sin2A + sin2B + sin2 C — 2
(ABC) 4TP
1 — cos 2i4 1 — cos IB t 1 — cos 2C
- H-------- 7 I-------- 1 -
2 2 2
= ---- (1 + cos 2C + cos 2A + cos 2B)

2
= ---- [2 cos2 C + 2 cos (v4 + 5) cos 04 — 5)]

2
= — cos2C—cos 04 + .6) cos (A — B) = — cos2C + cos Ceos 04 — 2?)
= cos C[cos(y4 — B) — cos C] = cos C[cos 04 — B) + cos (A + B)]
= 2 cos A cos B cos C.
Thus, if A1B1C1 is a triangle that is orthocentric with respect to A ABC
(that is Alf Bl9 Cx are the feet of the altitudes of A ABC), then
(W i) 2 cos A cos B cos C,

(ABC)
whence, incidentally, it follows that if triangle ABC is an acute-angled

triangle, then A ABC and A AXBXCXhave the same orientation (Fig. 9),
and if it is an obtuse-angled triangle, then opposite orientation (Fig. 10).
Problem 7. Given A ABC and a straight line /. Let Ax, Bx, Cx be the
orthogonal projections of points A, B, C on /. Prove that the lines that
pass through the points Ax, Bx, Cx and are respectively perpendicular to
the straight lines BC,CA,AB intersect in a single point Q (called the
orthopole of the straight line / with
respect to the triangle ABC) (Fig. 11).
Taking {ABC) for the unit circle and
assuming / is given by
Z — z0 = x(~z - z0), \x\ = 1, (7)
find the affix o of the orthopole Q of /
with respect to A ABC. The affixes of
the vertices of A ABC are equal to
z l9 z 2, z3 respectively.
Solution. The equation of the straight
line passing through point A perpen
dicularly to / is of the form
z — z x = — x( "z - Z j). (8 )
Rewriting equations (7) and (8) as

Z — K Z — Z0 — X Z 0,
z + xz = zx + x z x
and forming the half-sum, we find the affix z = ax of point Ax:
ax = — (z0 + zx + x z x — x z 0).
2
The equation of the straight line passing through point Ax perpendicularly
to BC is of the form
z - fli = Z j Z * ( z - a x)
or
z — J (z0 + z, + x z \ - XZ0) = Z^ 3 ( z - — ( z 0 + J - + — - — j j
or
... If , X
Z — z *,z3z = — z„ + Zj H---------- x z 0
2 V Zi
Z2Z3 „ ^ - °3 , Z 2Z3Z0 ^ /Q \
-------------- z2z3z 0 ----------1------------- • (9)
Zx X X )
Similarly, the equation of the perpendicular dropped from point B1
on AC is of the form
2 —Z3Zx¥
= - U + z2 + - - XZ0- Z*1 - z3z1z„ - ^ + . (10)
2 v z 2 z 2 x x )
Subtracting equation ( 10) from (9) termwise, we find the number co that
is conjugate to the affix co of the point of intersection of lines (9) and (10):
z a(z i ~ z i ) = [Zl“22+X(~—
l- “j
+ zs ( — — — ) — - z*)z<> - (^1 - z2)]
\ Z2 zi J x J
or, cancelling z, — z2,
1 (. x Z3Z i -J- Z3Z2 _ Z;3Z0 \
z&> = — 1 -------- + Z3 Z 0 —
2 V ZjZ; ZiZo ■ rj'
whence
co
2 \ Z3 Z jZ 2Z3 Z jZ 2 ^ /
or
CO
2 U 3 *3 * /
whence
1 / (T2 * , - \
CO = — - — + z0 - XZo]
\ Oz oz }
1
But a 2 = — * ct3 = — and so
O3 03
CO = — (o'1 — — + zo — * z0j ( 11)
The symmetry of this equation with respect to zl5 z2, za permits stating
that the perpendicular dropped from point Cx on line AB will also pass
through the point with affix co defined by ( 11), that is, ( 11) defines the
affix of the orthopole of the straight line / with respect to A ABC. Inci
dentally, this is evident from the following: write down the equation of
the perpendicular dropped from Cx on the line AB and convince yourself
that the number co given by ( 11) satisfies the equation of that perpendicular.
In particular, if / passes through the centre O of the unit circle (O) =
= {ABC), then the affix co of its orthopole with respect to A ABC is
and if the Boutain point is taken for the unit point, then
o> = ~ ( a 1 — x). (13)
Figure 11 depicts the construction of the orthopole Q of line /with respect

to A ABC.
Remark. There is an elegant synthetic proof (which belongs to the French
geometer R. Deaux) that the straight lines lx, l2, /3 that pass through the
points Al9 Bl9 Cxand are perpendicular to the lines BC, CA9AB respectively
intersect in one point Q. Let A2, B29 C2 be the midpoints of sides BC,
CA, AB, then the radical axes of the circles (A2, A2CX), (B2, B2AX), (C2, C2Bt)
taken pairwise are the straight lines ll912, /3,
and therefore they intersect in the single
point Q (Fig. 12).
Problem 8. Given a triangle ABC and
a straight line /. The circle (O) = {ABC)
is taken to be the unit circle; zl9 z2, z3 are
the affixes of the points A, B,C, respec
tively. The line / is given by the self
conjugate equation
Hz 4 - az = b (14)
where a ^ 0 and b is a real number. Find
the affix co of the orthopole Q of line /
with respect to the triangle ABC (see
the preceding problem).
Solution. The slope of the given straight line is equal to x — — ~
a
and the slope of the line perpendicular to / is equal to x' = •
a
The equation of the straight line that passes through point A perpendi
cularly to / is of the form
z - Zl = — (z — z x)
a
or
az — az = azx — a zx. (15)
Adding the equations (14) and (15) termwise, we find the affix ax of the
projection A1 of point A on I:
az1 — a z1 + b
ai =
Id
The equation of the straight line passing through point A1 perpendicularly
to BC is of the form
z — a1 = z2z3(z — a j
or
azx — a z x + b /_ a zx — az1 + b '
z ------------------- — Z2z 3 I z
2a 0 2a
or
azt — a zt + b a z x — azx + b
z z2z3z — ^2^3 (16)
25 2a
In similar fashion we can write down the equation of the perpendicula

to AC that passes through the projection Bx of point B onto /:
az2 — a z 2 + b az2 — az2 + b
z — ZjjZiZ — Z3Zi (17)
la 2a
Subtracting equation (17) from (16) term by term, we get the number dj,
which is the conjugate of the affix co of the orthopole Q of line / with
respect to triangle ABC:
(zx - z2) a + ( - ---------] a

(zi - z 2) zzco ___________ V*2 )
2d
z2 ( — - — ) a + bzz(zx - z2)
V *2 *1 /______________
+ 2a
and, cancelling zx — z2, we obtain
a z& i + Z2) a„ +, bz3
a --------------
zzco = - *1^2 Z1Z2__________
2d 2a
whence
, a 1 1
, b
N 1I
1 1
l
1 _
i
>
2zz 2aaz 2z x 2a
N
to
ac* . b
CO —
2 ^ la 2a
and, hence,
" = + - + -)• (18)

2 V a a)
If / passes through the centre of the unit circle, then 6 = 0 and for
mula (18) becomes
CO = ~ » (19)
and if the unit point is a Boutain point (a3 = 1), then
co 1 (
t r H —
2 \ a )■ (20)
The equations (19) and (20) coincide, respectively, with (12) and (13)
of the preceding problem, since — = — x, where x = — ~ is the slope
_ a_ a
of / given by the equation az + az = b.
Problem 9. Let Alt Bu Cx be the feet of the altitudes of A ABC inscribed
in the circle {ABC) = (O), which we take to be the unit circle. Points
P, Q, R are chosen on the straight lines AAU BBU CCy so that
AAX
AP
Find the ratio
(PQR)
{ABC)
Express this ratio in terms of the interior angles A, B ,C of the given

A ABC and in terms of X. Consider the special cases: (1) X — 1,
(2) X = - 1, (3) X = 1/ 2, (4) X = -1 /2 , (5) X = 2, (6) X = - 2 .
Solution. Take the circle (O) as the unit circle. Let zl9 z2, z3 be the
affixes of the points A, B,C, respectively. The equations of the straight
lines BC and AAX are
z + z2z3z = z2 4“ Za,
Z2Z3
Z ZoZoZ — Zi
Combining these equations term by term, we find the affix z = ax of

point Ax:
ai =
and then from the relation
AAX
= A
AP
we find the affix p of point P :
cii - zx
— A, Zj — Xp Azj,
p - Zi
Ap = <7i - + Azx = y + Azr
whence
1 Z' <r3 \ , A — 1
P ~ 2x z f ) + ~~A
Similarly,
f
O’3 '\ . A — 1
- I+
4 ,1 A
(T3 )I , A - 1
0-1 - “7 + -
I A
wlicre q and r are the affixes of points Q and R. We now find
___g3 , - z 4- — 72 A- 1 1 <^2
1 + —+
2Azf ^ F " 1 2A 2kaa A 2Act3
^3 A - 1 1 I 0-2
(/'(>«)= — r, + ^ 4 ~1 -r
2;.za2 A 2A 2A<t3 ' A *2 2Ac73
<r3 . i n i r, + £ i A- 1 1 <^2
Z| 1 +
2Az§ A 2A 2A<r3 1 A *3 2Ac73
^3 i_ A - 1 2? A— 1 1
2Azf 1 A 2X<j 3 + zi
,2
A- 1 Z2 A — 1 _1
z2 — 1
2Azl 1 A 2A<r3 A z2
LA - 1 4 A- 1 1
2Az| 2A«r» A z3
z-2'z zf 1
-Z2— (73(A -1 )
“ 2‘ zl 1 zf2 z f1 1
4A2 2A2
Z3 2 z\ 1 z3“ 2 Z3-1 1
Zl zf 1 Zl Z f*
A- 1 (A - l )2
z2 zf 1 z2 z f 1 (21)
' 2AV, A2
Z3 z§ 1 Z3 Z3 - 1
Furthermore,
z r 2 z2 1 1 z2 zf
z f 2 zf 1 1 zl z| z?)(z§ - z2)(zl - zf)
Z3-2 z3 1 1 z§ z£
---- -- (z2 — Zi)(z3 — z2)(z3 — Zi)(z2 + Zi)(z3 + z2)(z3 + 2l)

<j\
1 2i 2 l
l 2 2 2 l ( 2 2 + z i ) (Z3 + Z2) (Z3 + Zl)
ol
1 2 3 2§
Z1 1 2l
Z2 1 22 (<?1 - 2 i)(ffi - Z2) ( f f i - Zz)
<*3
z3 1 23
1 4
= (ABC) (<7i a2 — aq) := —
03 I
and the first term in round brackets in (21) is
/(l — ad1) i
■(ABC). (22)
A2
Subtracting the second term, we get

Z f2 Z l1 1 *1 1 *1
1 4/
z2~ 2 Z2 1 1 *2 1 *2 = ------ {ABC),
“ a3 <*3
z f 2 Z3 1 1 *3 1 *3
so that the second term in round brackets in (21) is equal to
2/(2 - 1)
(ABC). (23)
A2
Furthermore,
Zi z \ 1 1 Zy Zy
^2 z \ 1 = ^3 1 r2 z 2 = —4ioz(ABC),
z%z| 1 1 Z3 z3
and so the third term in round brackets in (21) is

2/(2 - 1)
(ABC). (24)
22
Finally, the last term is equal to
-4 i— — ~(ABC). (25)
From the formulas (21)-(25) it follows that

i f /(I — aydi) [ 2/(2 — 1)
(PQR)
22 22
2/(2 - 1)
+ - 4 /; a - I )2 ] (ABC)
22 22
—W y 2 - 1 (2 - l)2 '
(ABC)
422 22 22
whence
( /15C) 422 ( - D ( - t)
But adx — 1 is the power of the orthocentre H of A ABC with respect

to the circle (ABC). We have
od1 — 1 = (zx + z2 + z3) (Zi + z 2 + z3) — 1
= 2 + (zxz 2 + z2z\) + ( z 2z 3 + z 3 z 2) + ( z 3 z i + Z iZ g ).
Furthermore,
AB2 = c2 = (z2 — z2) (z 2 — Zi) = 2 - (zxz 2 + z2zi),
whence
ZjZ2 + z2Z! = 2 — c2 = 2 — 4 sin2 C,
since
—— = 2R = 2.
sin C
In similar fashion we find
z2~z3 + z3z 2 = 2 — 4 sin2 A, ZgZ^ + ^1^3 = 2 — 4 sin2 P
and therefore (see problem 6)
Gj&x — 1 = 4(2 — (sinM + sin2 B + sin2 C)) = — 8 cos A cos B cos C,
and formula (26) becomes
= (PQK) _ 2 cos A cos B cos C A _ J \A _ A V 07)
** (ABC) ~ A2 aJ\ A/ ’ 1 '
(1) If A = 1, then the points P, Q, P coincide, respectively, with the
feet of the altitudes of triangle ABC.
Answer, p = 2 cos >4 cos P cos C.
(2) If A = —1, then the points P, £?, P are symmetric to the feet Alf
Bl9 Cj of the altitudes of A ABC with respect to its vertices A, B, C.
Answer, p = 6 + 2 cos ^4 cos P cos C.
(3) If A = 1/2, then the points P, Q, P are symmetric to the vertices ^4,
P, C of A ABC with respect to its sides.
Answer, p = 3 + 8 cos A cos P cos C.
(4) If A = —1/2, then the points P, g , P are obtained from the points
A 1, P l5 Cx via the homothetic transformations (A, —2), (P, —2), (C, —2).
Answer, p = 1 5 + 8 cos ^4 cos P cos C.
(5) If A = 2, then P, Q, P are the midpoints of the altitudes AAl9 PPX,
CCX of A ABC.
Answer, p = — cos A cos P cos C.
2
(6) If A = —2, then the points P, (?, P are symmetric to the midpoints
A2, P 2> C2 of the altitudes y4/tl5 BBl9 CC± of A ABC with respect to its
vertices.
Answer, p = 3 -----cos A cos P cos C.
Problem 10. Let A', B', C be points symmetric to the vertices A, B ,C

of A ABC with respect to its sides BC, CA, AB. What relationship be
tween the angles A, B ,C of A ABC is necessary and sufficient for the
straight lines AB', BC', CA' to pass through one point (Fig. 13)?
Solution. Take (ABC) for the unit circle.

We find the affix b' of point B'. The equation of the straight line AC is
z + z3zxz = zx + z3.
The equation of the straight line passing through point B perpendicularly
to AC is
Z Z 2 — ZgZ^ (Z Z 2)
or
Z
—
--- ZgZAZ
_
Z3Z1 •
Z2
*2
From the equation of AC and from this equation we find the affix of the
projection of point B on line AC:
z
H 2i+ 2,+ z’ ~ ~ t y
The affix b' of point B', which is symmetric to point B with respect to AC,
is found from the relation
b' + z2
2
whence
b' = zx + z3 *3^1
^2
From this we have

z2 zi za z2
b’ = z \ + z 3 —
Z3Z1 zxzz
We now set up an equation of the straight line AB':

z z 1
Zi z l 1 = 0
b’ bx 1
or
z ( z x — b') + (b' — z x) z + z ^ ' — 6 'Z i = 0 .
We have
— -, 1 Z3 - z1 z2 z2 zx
z x — b = ----------------------- f
Zi ^ 3 Z1 z 3z x
_1____ 1_
u, z 2 Zx Z3(Z i — Z2) z 3( z 2 — Z j)
Z1 — O = ------------------------- = -------------------------------------- f u — Z 1= ---------------------------------------9
Z3Z1
/— z3 + Zl — *2 1 [ _ Z3Zt \
ZX= Zi — Z3 + *1
Z3Z1 Zx V *2 /
UN
1
Z3 + Zx - _^3 _J_ ^3 _^2_

- 1 + * =__ *1
1
z3 Zl z2 *3 Zl z2 z3
1
- ( Z x - ■Z2) + z3 ( — ■(z1 - z 2) + z 3^ “ ^
Z3 V *2 *i / *3 ZiZ2
= (zx - z2) ( — +
V Z3 ZxZt /
and the equation of AB' takes the form
1 J3 _ = 0
*+
Z3Z1 ^3 *1*2
or
z2z + z3zxz — zi — zxz2 = 0 .
In similar fashion we can write down the equations of the straight lines BC’
and CA'. Thus,
Z .J I + z\zxz — zl — zxz2 = 0, (AB')
z3z 4 - zfz2z — z\ — z2z3 = 0, (BC)
zxz + z|z3z — z\ — z3zx = 0. (CA')
Denote by K, L, M the points of intersection of the lines B C and CA',
CA' and AB', AB' and BC. Then
z2 z 3z x z§ + z xz 2
727
z3 zxz2 .2
Zf + Z2Z3
z x Z2Z3
—2 —
z-.22 + Z3ZX
(KLM) =
4 o-3(z2z§ - - 4) (zi4 - 4 )
Furthermore,
z 2 z |z x z\ + z xz 2
A = z3 Z1Z2 Zl T z2z3 = Z?z£ + Z]Z2Z3 + z\z\ + z xz 3z 2 + z\z\ + z xz 2z 3
zx zfz3 zl + Z3ZX
-7^7 73^ Z1Z2
ZiZ2Z z ? z lz 3 — z\z\— zxz\z% — z\z%
- 4 4 ( 4 - 4 ) - 4 4 ^ 4 2 - *i) + 4 (4 - z f) + z\z3zx(z\ - z f) - z\(z\ - z\yt
- zlzxz2(z2 — zx) = (z2 — zx)(zfz2 + z\z\ — zfzfz3 + z*3z2 + z\zx + z\z\zx
-I- z\z2z\ — z\z\ — z§z|zx — z§z2zf — z\z\ — zlzxz2) = (z2 — zx)[zfz2(zx — z3)
+ zf(zf — 4) — Z2Z§(zf - 4 ) — Zxz|(zf — 4 ) + ZjZ2Zx(zx — z3)]
: - (z2 - zx)(zx - z3)(zfz2 + zfzx + zfz3 — z2zxz3 z 2z 3 zfz§ z xz 3 -|-Z3ZxZ2)
= 0*2 - z x) ( z x - z 3) [ z ! ( z | - z i) + z x( z f - zl) + z 2z 3 ( z i — z\ ) ]
(z 2 — z x) ( z x — z 3) ( z 2 — z 3) ( z f z 2 + z fz 3 + z xz l + z xz 2z 3 + z xz l + z l z 3 + z 2z § )
(Z2 - Zx)(z3 - zx)(z3 — z2)(zfz2 + z|zx + z\z%+ zfz2+zfzx+z?z3+ z xz2z3).

Consider the product
(/.. | z 3) ( z 3 + z x) ( z x + z 2) = — <t 3
= 2 Z i Z 2z 3 + z?z2 + z |Z i + z |z 3 + z 3z 2 + z |z x + z\z3.
from this it follows that the last factor — o3 — c3 = — 2<r8
so (hat
A = (z2 - Z j ) ( z 3 - z x) ( z 3 - z 2) (<rx<r2 - 2 ( t 3) .
Thus,
i (z2 — Z1) 2(ZS — Z ^ { z z — Z 2) 2 (< T1<T2 — 2(73) 2
{KLM) =
4 <r3(z2z§ — zi)(z3zi — z^Cz^l — z;j)
Note that
(z3 - z2)(z3 - z1)(z2 - zx)
1 zx z? Zl 1 zt Zl *1 1
= 1 z2 zf = z 2 1 z2 = *3 z2 z 2 1 — 4icr3(ABC),
1 z\ *3 1 *3 Z3 Z3 1
so that
4ic3(o‘1(T2 — 2<t3)2(ABC)2
(KLM) = -
(z2zi - z?)(z3z2 - z i ^ z 2 - zl)
The lines A B \ BC \ CA' pass through one point if and only if
axa2 — 2<t3 = 0
or
Gx — — 2 = 0, G l (Tl — 2 — 0,
<^3
that is,
c r ^ = 0 / / 2 = 2 = 2R* (R = 1)
or
O H = R y2.
Incidentally, it follows from this that the triangle ABC is an obtuse-angled
triangle since OH > R, that is, the orthocentre H lies outside the circle
(ABC).
A triangle for which OH = R )[2 may be constructed as follows: con
struct two concentric circles (O) and (O') with radii 1 and ^2 (see Fig. 13).
On (O') take an arbitrary point H and on (O) an arbitrary point A. Divide
the line segment OH in the ratio 1 :2 :
OG : GH= 1: 2.
Then G is the point of intersection of the medians of the triangle ABC.
Join A and G and on the extension of segment AG beyond point G lay
off segment GAX= AG/2. The point Ax is the midpoint of side BC and
therefore, by drawing a line through Ax perpendicular to the straight line
OAx we obtain points B and C as points of intersection with the circle (0).
The condition that makes the construction possible is that point Ax

lie inside circle (0). Since point is the image of point A under the homo-
thetic transformation -----j > and under that homothetic transforma
tion the circle (O) goes into the circle (O"), the radius of which is equal
i Vi Vi Vi
to -y, and the centre O" lies on segment OH, 0 0 " = — + — = ~ ,
it follows that the circles (O) and (0") intersect in points P and Q. The
point A can therefore describe an open arc PRQ of the circle (0). From
the triangle OPO" we have
PO"2 = OP2 + 0 0 "2 - 20P-00"cos(a/2),
where a = /_ POQ. And since PO" = 1/2, 0 0 " = ^2/2, OP = 1, it

follows that
= 1+ t - 2- i1/2 a i
_ cos —
whence
a _5_
cos —
2 4]fi
and, consequently,
25 9
cos a = 2 - ---------1 = — t a = / POQ = arc cos (9/16).
32 16
The condition OH 2 = 2R2 may be rewritten in a different form. Since
OH2 = 9R2 — {a2 + b2 + c2) (see problem 6), it follows that
9* 2 - (a2 + b2 + c2) = 2R2
or
a2 + b2 + c 2 = 1R2.
And, in yet another form (since a = 2R sin A and so forth),
sin2 A + sin2 B + sin2C = 7/4
or
cos A cos B cos C = —1/8
(from this it also follows that ABC is an obtuse-angled triangle).
Problem 11. Inscribed in a circle (0) is a regular 14-gon:
Let us consider the triangle T = ABC with vertices

A = A 7, B = Al9 C — A 3.
The angles of this triangle intercept arcs AXA39 A3A7, A7A± and, hence,
A = n p y B = 2n/79 C = 4n/l (Fig. 14) (that is to say, the angles of tri
angle T form a geometric progression with ratio 2). Denote by (/fl) the
circle escribed in the angle A of A ABC, and by H the orthocentre of
A ABC.
Prove that:
1°. OH = 01 a = R ]/2 (R is the radius of (0)).
2°. R = 2ra (ra is the radius of circle (Ia)).
3°. IaH = R.
4°. a2 + b 2 + c2 = 1R2 (here, a9 b9 c are the lengths of sides BCy
CA9 AB of the triangle; do not confuse with the affixes of its vertices!).
5°. OIaHAQ is a parallelogram whose centre coincides with the centre
of the Euler circle of triangle ABC.
6°. The midpoint P of segment HA6 coincides with one of the points
of intersection of (O) and (Od) ((09) is the Euler circle of A ABC).
T . The triangles AIaH, HBIa, IaHC are similar. Determine their
orientation (which pairs have the same orientation and which are
oppositely oriented).
8°. The straight lines BC, CA and AB intersect line HIa in points sym
metric to the points A, B, C with respect to the bisectors of the angles
C, A, B of triangle ABC (for angles C and B take the bisectors of the ex
terior angles).
9°. The squares of the lengths of the sides of A OIaA&and the squares
of the lengths of the sides of A A3IaH form a geometric progression with
ratio 2 .
Solution. 1°. Take (O) as the unit circle. Give point A± the affix 1 (that
s, A± is the unit point). Then the affixes ak of points A k are
(k — 1) 7T (k — 1) n
ak = cos + i sin 9
7 1
* = 1 , 2 , 3, 4, 5, 6 , 7, 8, 9, 10, 11, 12, 13, 14.
The affixes of the vertices A = A7>B = Al9 C = A3 are
6n , . . 6n
a = a7 = cos------b i sin — >
7 7
b = a1 = 1,
2n , . . 2n
c — a3 — cos------ b i sin-----
7 7
Their sum is equal to the affix h of orthocentre H of A ABC:
27T 6n ,. . 2n . 6n
whence
h= C O S ----------- b C O S ----------- b I
1 1 (
sin ------ b sin —
7 7 )•
no Problems in Geometry
Let
- , _ 4ft , „ 6n
x = 2 cos------ b 2 cos------ b 2 cos —
1 1 1
then
ft . . ft 2ft , . . ft 4ft , ^ . ft 6ft
* sin — = 2 sin — cos------ b 2 sin — cos — + 2 sm — cos —
7 7 7 7 7 7 7
. ft , . 3ft . 3 ft , . 5ft . 5ft , .. f t
— — sin ------- b sin -----------sm ----------- b sin ---------- sm ----------- b sm ft = — sm —
7 7 7 7 7 7
and, consequently, jc = —1 ; therefore
OH 2 = 3 — 1 = 2 ,
whence
= ( /? = ! ) .
Now let us determine the affix ra of the centre Ia of the circle (/,). The
bisector of the interior angle B is BA5 since point A 5 bisects arc CA. From
this it follows that the bisector of the exterior angle B is the straight line
A1Z B since the points A5 and A12 are diametrically opposite points of the
circle ( O) , and therefore BA5 _L BA12. The bisector of the interior angle
A is AA2 = A7A2. T o summarize: Ia is the point of intersection of the lines
A 2A7 and AXA12. Now set up the equations of these lines. The slope of
A2A7 is 1, since A2A7 1| A±A8, and A ^ g is the real axis. Hence, the equation
of the straight line A2A7 is of the form
ft . . . ft \ / n . . 7r \
or
(c o s ------- b i sm — I
7 7 /
= z — I c o s ----------i sm —
V 7 7 /
ft
z — z = 2i sin (28)
7
The slope of the straight line AXA12 is
#12 1 #12 1
— #12
#12 1
#12
and, consequently, the equation of AXA12 is

or
i ( 4;r . . An \
z — 1 = I cos— - + i sin — J (z — 1). (29)
Solving the system (28)-(29), we find the affix ra of point Ia From

equation (28),
— ~ • 71
z = z — 2i sin —
7
and equation (29) takes the form
« ( An . . . An \ ( . 7u \
z — 1 = 1 cos-------b i sin---- I | z — 2/ sin ------- 1 1 ,
V 7 7 M 7 ;
. . . 7r ( An , . . An \
2/ sin— I cos-------b / sin---- I
j ________ 7 V 7__________ 7 j
. . 9 2n . 27t 27t
2 sm2 -------- 2/ sm ----- co s-----
7 7 7
. . n ( An An \
/sin — • cos------- b *sin----- 1
= 1 a. 7 V 7 7 j
1 . . 2n ( 2n . 2n \
i sin---- cos--------- b i sin----- 1
7 V 7 i J
7T / 2n t . . 2n \
sm — I cos-------b i sin----- I
7 V 7 1 )
= 1+ . 27T
sm----
7
2 ;r
cos -
1+ + i sin —
n 7
2 cos —
7
From this
/ . 7T 2tT \ 2
I 2 cos — + c o s---- I
I 7 7 j
Oil = ItJ 2 = + sin2
4 cos2
a cos22 ^ i a ^ 2»n 9 2n . 9 2n
A ----- b 4 cos — co s-------b cos2 ---- + sm2 -----
7 7 7 7 7
n
4 cos2
+ 2^1 + COS j + 2 ^CO S + COS - - j

4 cos2—
7
2n An 6n\
(
3 + 2 cos--------cos-------- COS----- I
7
4 cos2—
7 7;
7
271 2 ft 671 4rc\
3 + 2 ^2 cos ------- cos----- ------- cos----- I
7 7 2 _________ 7^/
4 cos2-71-
7
^ 271 , 47T t 6tt 1
But c o s ---- + cos-------- b cos---- = -------, therefore
7 7 7 2
2 tt
3 + 4 cos + 1 1 + cos
__________ 7
on= = 2,
4 cos2— cos*
7
whence
OIa = ][2 = R f 2 = OH.
. ^4 n f . n . 2n . 3n \*
2 . ra = p tan - - = p ta n ---- I sjn------b sin------ b sin---- ) •
2 14 V 7 7 7 j
Let
7T . 2 7 1 3 tT
v = sin----- b sin-------b sin----
1 1 1
Multiplying both sides of this equation by 2 sin we obtain
^ . n ^ . n n , . . 7r . 2n , - . n . 3n
2y sin---- = 2 s in ------sin------ b 2 sin ---- sin --------b 2 s in ---- sin —-
14 14 7 14 7 14 7
n 3n 3n 5n 5n In n
= COS-------- COS------- b COS-------- COS------- b COS------ cos — = cos —
14 14 14 14 14 14 14
* If the radius of the circle is equal to 1, the chord subtending arc a is equal to
2 sin (a/2). Angle A is an inscribed angle intercepting the arc 2tt/7, hence AXA3 = BC —
= 2 sin (7T/7) and similarly for the other two sides. The lengths o f the sides may also be
found by the sine theorem: a = 2R sin A = 2 sin (n jl) and so forth.
Hence,
1 n
y = — c o t-----
2 14
and so
n 1 A n 1
ra = ta n ----------c o t------= —
14 2 14 2 2
whence
R = 2ra.
3°. Furthermore, IaM = h — xa {h and xa were computed in item 1°),
whence
2k
cos----
. 2k , 6k 7 . . 2n
h — za = c o s ------ b cos-----------------------b / sin----
7 7 _ 7T 7
7 7 2 cos — 2*7
7
- 7T 2 tt , . 6n n 2n
2 cqs — c o s -------- b 2 c o s ----- c o s -------- c o s ------
7 7 7 7 7 , . . 2tt
= --------------------------------------------------------------- h zsin-----
2^ cos —71 1
7
7T 3ft , 5tt 2ft
cos----- b COS-------b cos--------1 — cos------
7 7 7 7 .2 n
= -------------------------------------------------------- + isin-----
- ft 7
2 cos — 7
7
6ft 4ft 2ft . 2ft
— cos--------cos---------cos--------- 1 — cos------
7 7 7 7 . . . 2tt
---------------------- b isin------
- 7T /
2 cos — 7
7
1 271 1 271
------ 1 — cos------ cos-----
2 7 . 2;r 2 7 . . 2tt
--------------------------- b i sin — =>------------------------- b i sin ------
7 2~ cos —* 7
2 cos — 71 0
7 7
4tt 6 ft . 5n n
cos---- ■+ cos---- 2 cos---- cos —
7 7 . 2 k 1 7 . 2 k
-------------------------- b * sin — = ------------------------- b *s m ------
~ ft 7 7T 7
2 cos — ■ 2 cos—
7 7
5k . . 2k 5k 5n
= cos-------b i s m ---- = cos----- 4- i s m ------ =
7 7 7 7
8 810
Thus,
5n . . 5n
h — Tn cos i sin - = a,6
1
and so
HIa = |h T .I = 1 = R.
n 2n 3tt
4°. a2 + b2 + c2 = 4 sin2— + 4 sin2 + 4 sin2
~7 7 7
2n 4;t , , 66n ir \
C O S ------------ h i — COS h i — COS -
7 7 7 /
= 2 r 3, — I/ cos----
2,1 + cos — + c o s - ~ - j J = 2 [3 - ( - 1/ 2)] = 1—1R2.
L V 7
5°. The directed line segments 7fl/7 and CM6 are equivalent, that is to
say, they are parallel, equal in magnitude, and in the same direction; since
h — xa = a6 (see item 3°); hence IaHA%0 is a parallelogram. The centre of
this parallelogram is the midpoint of the line segment OH, that is, the
centre of the Euler circle of triangle ABC.
6°. The affix of the point P is
h + a Q _ 1 I"jt , 2n 6tc 5tt
1 + COS------- b COS------- b COS-----■
2 "2 L 1 1 1
2n . 6k . 57T
+ / + sin---- + sin----
—
7 7 7 /J
1 71 .( 27T . . 7T
: --- 1 — cos ----- b / 1 2 s in ---- -b sin —
2 L 7 I 7 7
■whence
I h + a* I2
^
=T [(1"cosf + sinV +““f
„n
)2 (2 )1
. . 0 2n , A . 2n . n , . _ n \
2 cos----- b cos2----- b4 sin2 — + 4 sin — sin -----bsin2— |
- i( - 7 7 7 1 1 1 )
= i r [ 2 “ 2 cosy + 2 ( 1 ~ cos 4 “ ) + 2 ( C0Sf ' “ C0S' 7 L) ]

1 n , , 4tt , 7C 3tt \ 1 f
= — I 1 —c o s b 1 — cos-------b cos cos 1= — • 2 = 1
2 V 7 7 7 1) 2
h + ^6
= 1,
That is, the midpoint P of segment HAe lies on (O). The centre 0 9 of the
Euler circle of A ABC is the midpoint of segment OH. Thus, 0 9P is
1 P
the midline of triangle OHAb and, hence, 0 9P = — OA6 = — . From this
2 2
it follows that point P also lies on the Euler circle (0 9).
7°. We will prove that A AIaH and A HBIa are similar and have the
same orientation. We have
h h - Tfl 0
|ai h 1
Tfl 1
h Ta lj
1 — k —h 1
h Ta 1
= (fl7 - A ) ( l - T j + ( A - T j 2.
Furthermore,
6k 6k 2k 6k . . 2k . . 6k
cos----- \~ i sin------ 1 cos------cos------- i sin-------- ism—
7 7 7 7
. 2k . . 2k
= —1 — cos---- i sm —
7 7
2k 2 ;r
co s---- cos
. sm
. - 71- = 7 . . K
11 — t* = 11 —
1 1 -------------------i
7 ------- i sin — >
^ K 7 K 7
2 cos — 2 cos
so
/7
(Ji — Ta)2 = ai2 = cos-------- .b ^• s•m ------=
I ®71 3K
— cos--------- i s. m. ------
3k
7 7 7 7
2k
cos
. . 71
— /i) (1 — rfl) = ^1 + cos ---- + i sin — - j ------- b 1sm—
2 cos-
2k . . 2k
co s-------b i sm -----
f 1 + cos — + i Sin — ) ------ -
y 7 77 2 cos —
2k t . . 2k
cos----- b i sin
( - 9 K , . 7T 7 7T \ 7 3k t . . 3k
= I 2 cos2----- b 2/ sm — cos— 1 ---------------------- = cos------b i s m ----
v 7 1 1 ) 2co a JL 7 7
Consequently,
A = (It - Taf + (a7 - h) (1 - T0) = 0.
------ > ------> ------ > ------ >
To summarize: AIaH[[HBIa and the triangles AIaH and HBIa are
similar and have the same orientation (precisely for the order in which
the vertices are specified).
We can similarly prove that A AIaH and A are similar and have
the same orientation.
8 °. We will prove, for example, that point B*9 which is symmetric to
point B = Ax with respect to the bisector of the interior angle A , lies on
the straight line HIa. The equation of the bisector of the interior angle A is
- . n
z — z = 2i sin----
7
The equation of the perpendicular dropped from point AL = B on that
bisector is of the form
z + z = 2,
whence we find the affix bx of the projection Bx of point B on the bisector
of angle A:
, i , . • 71
b± == 1 i sin ----
7
The affix b* of point B* can be found from the relation
i* + i
whence
71
b* = 1 + 2/ sin
1 ’
Furthermore,
, 5n . . 5n
h — ra = c o s -------h i s i n ---- >
7 7
2n
c o s ----
Tfl — 6 * = 1 H---------------b i sin n---- 1 — 2i sin
7C 7 7
2 cos — 7
1
2k 2k . . 2n 5k . . 5k
c o s ----- c o s --------- i s in ------ cos------1 sin —
7 . . n 1 1 1 1
=z---------------- / sin — —------------------------- = r --------------------------
^ 71 7 - Tt ~ 71
2 cos — 2 cos — 2 cos —
7 7 7
and, hence, the ratio

h — ra - n
------- - = — 2 cos —
ta-b * 1
is a real number, and so the points H , /fl, B* lie on one straight line. It is
left to the reader to prove the other two propositions of this item.
9°. Let us consider the triangle OIaAQy We have
OA%= 1,
2n \ 2
cos-
n
Oil T„ 6 = 1+ +sm 2—
2 cos
7
2n 2n
cos- cos2
7 . 2
= 1 +• ------- b sin2—
n 7
cos 4 cos2 —
7
. Q 7t . 7t 2n A 2k . n 7T
#w n 7T
4 cos2 ----- h 4 cos — co s------- \- cos2 ------ b 4 cos2 — sin2
7_________7 7_________ 7__________7_____ 1
4 cos2 ---
7
1 + 2 ^ 1 + c o s ^ -j + 2 ^cos - ~ + cos— j
4 cos2 ~
~ ~ f 2k 4k 6 k\
3 + 2 1 co s------- cos---------cos------|
V 7 7 7;
4 cos2—
7
f 2^ 271 1\ 2rc
3 + 2 I c o s-------b c o s ---- 4------I 4 + 4 cos -
V 7 1 2 )
= 2.
4 cos2 - - 4cos2-—
1 7
And, finally, from the parallelogram OIaHA6 we have (the sum of the
squares of the diagonals of a parallelogram is equal to the sum of the
-squares of the four sides):
Oil + HA§ + 0 A \ + IaH* = OH 2 + IaA l

But Oil = HA% = 2 (see items 1° and 5°), 0 A 6 = IaH = 1, OH = f2 ;
consequently,
2 + 2 + 1 + 1 —2 + /^ i,
whence
IaA% = 4.
The triangles and A6IaH are equal, and so the squares of the their
sides IaH2, also form a geometric progression with ratio 2.
Problem 12. Given the linear fractional function
az + ft
u = --------- J (30)
cz + d
where a, ft, c, d are fixed complex numbers and ad — be ^ 0, c ^ 0 ; z
is the argument and u the value of the function.
Prove that if |c| ^ \d\ then the image of the unit circle under this trans
formation is a circle; find its radius and the affix of the centre.
Solution. Transform the function u as follows:
be — ad
^ _ a ^ az + b a a , be — ad
c cz + d c c c (cz + d)
(1) The transformation z z H----- takes the unit circle (QJ = (0, 1)*
into the circle (£22) (Fig. 15).

1
(2) The transformation z + consists in symmetry with
z +
respect to the real axis Ox, under which symmetry the circle (Q2)=
goes into the circle ((23) = and subsequent inversion (see chapter
IV) of (&3) with respect to (&+ under which inversion the circle (Q3)
goes into the circle (&4); to construct (&4) it suffices to draw the straight
line OQ3 and to construct the images P f and Q of the endpoints P and Q
of the diameter that cuts out that line on (&3). The circle constructed on
segment P'Q' as a diameter is precisely the circle (04). Let us compute
* The symbol (z0i R) will be used to denote a circle o f radius R , the affix o f whose
centre is z 0.
Fig. 15
the radius of (fi4) and the affix of its centre. To do this, note that (&3)
passes into (G4) under the inversion with respect to the circle (fix) and
under the homothetic transformation (0 , \ / g ), where g is the power of
the point O with respect to the circle ($23).
But under the homothetic transformation (O, l/<r), the centre of (&3)
goes into the centre of (0 4); hence the affix co4 of the centre Q4 of (0 4) is
equal to
G C
and the radius is equal to the product of the radius R ^ — 1 of the circle
(Q3) by the modulus of the homothetic ratio, that is, by 1/\ g \:
1 _1_
i £4 — Rs (*3 = I)-
kl
Furthermore,
dd — cc *
a = OQl - R j= —- — - 1
c c cc
* In the general case, under inversion with respect to the point Oy we have
(O M )-(O M ') = k ,
where k ^ 0 is the power o f the inversion, and, hence,
(OM) _ k
( ON) " cr ’
where k is the second point o f intersection o f the straight line O M with circle being
inverted; # 0, since \d\ ^ \c\.
SO that
cd cc \
(Q*) = ( s r i f cc |ddj —cc\)
(3) The transformation
be — ad
1 c2
z + — d
z H-----
consists in turning the plane through an angle of arg - - ---- — and in a

c2
homothetic transformation with centre O and ratio . After
Id2
these two transformations, the circle (0 4) goes into the circle (£5), the
affix co5 of whose centre is
be — ad cd be — ad d be — ad
co5 = q 4
< dd — cc c dd — cc
and the radius is
cc be — ad be — ad
*5 =
dd — cc c2 dd — cc
Thus
d be — ad be — ad
-( c dd — cc
(4) Finally, the transformation of translation
dJ — c7\)
be — ad be — ad
c2
--------------------- ---------------
a
>—
c2
c , d
z + z H-----
c
carries the circle (Q5) into (fi6) of the same radius: R6 = R5, and the affix
of the centre Q6 of (0 6) is equal to
a d be — ad bd — ac
0)q — 1-------- I= = H'
c c dd — ac dd — cc
cc | ! |c|2 Id2 = 1.
|C2| Id 2
To summarize, then: as a result of the transformation

u = a z + b , a d - b e * 0 , c * 0 , Ic| # |d\,
cz + d
the circle = (0 , 1) goes into the circle
bd — ac I be — ad
dd — cc Idd — cci
------- >■ ■■> ~
Problem 13. Suppose BCAXA2, C A B ^ , ABC1C2 are squares with the
same orientation and constructed on the sides BC, CA, AB of A ABC,
the orthocentre of which is H. Denote by (O) = {ABC) the circle passing
through the points A, B, C. Let P, 0 , be the respective centres of the
---------------------- >. v --------------------- k.
squares A ^ C 'C " , B±C2A' A ", C ^ B 'B " having the same orientation
as the first three squares. Denote by Pl9 P2, P 3 the orthogonal projections
of point P on the sides BC, CA, AB of A ABC, by Ql9 Q2, Q3 the orthogonal
projections of point Q on the same sides and by R v R2, R3 the orthogonal
projections of point R on the same sides BC, CA, AB of A ABC. Prove
that the sum of the forces
PPi + PP* + PPz + QQi + QQl + QQs + RRi + RR* + (3 1 )
is equal to the vector OH', where H' is the orthocentre of the triangle
whose vertices are the feet of the altitudes of the given triangle ABC.
Prove that the straight line on which the resultant of these forces * lies
passes through the endpoint of the directed line segment 4OH.
Now if the squares of the second triplet have an orientation opposite
that of the first three squares, then the sum of the forces (31) belongs to
the straight line OH'.
Set up the equations of these two straight lines and take the circle (O) =
— {ABC) as the unit circle (Fig. 16).
Solution. 1°. Let a, b, c be the affixes of the points A, B, C. Find the
affixes p, q, r of the points P, Q, R. Denoting by au bl9 cl9 a2, b2, c2 the
affixes of the points Al9 Bl9 Cl9 A2, B2, C2, we have
b2 = c + CB2 = c + iCA = c + i{a — c) = ia + (1 — /) c,
(32)
ax = c + CAX = c — iCB = c — i{b — c) = {1 + /) c — ib.
Also,
p - bo + - - B.iA1 — —- B2A1 = b2 + - - (ax — b2) - (ax — b2)
2 2 2 2
_ bu2 +, •— „ --—
1 a1 t -------------------------
1 b2 tai , lb* 1 + 1 b2 1- i
— ai
2 2 2 2 2
Force is a sliding vector.
Fig. 16
and, taking into account (32), we have
p = —— [ia + (1 - /) c] + 1----- [(1 + i)c-ib)]= 2c+ — ' - - a - —--- b.

2 2 2 2
(33)
Similarly (we can carry out a circular permutation of the letters a, b, c)>
we can find
1+
q = 2a + ----- - b -
2 2
1+/
r = 2b + -— - c -
2 2
Note here that it follows from (33) and (34) that
and, hence, A ABC and A PQP have a common centroid (the point of
intersection of the medians).
Now, from the equations of the straight lines BC and PPl9
z — b — — be (z — ~b)>
z — p = be (z — p)
or
z + bc~z = b + c,
z — bc'z = p — bcp9
we find the affix px of point Pt :
Pi = — ( b + c + p — bep);
2
and, hence,
PP\ = P i — P = — (b + c + p — bep) — p = — ( b + c — p — bep).

2 2
Similarly,
PPz = p » - P = - - { c + a - p - cap),
PP$ = Pz — P = - - (a + b — p — abp),
2
whence
PP1 -\-PP2t+PP^ = 01 ---- — P — ~~ &2 P‘
Similarly,
QQi + QQz + QQ3 = - — q - - - o-2q 9
RR± + RR2 + RR3 = o1 ----— r -----—<r2r.

2 2
Adding these relations term by term, we can find the principal vector of
the resultant:
bf ~ PPx + PP2 + PP$ + QQi + 6 6 2 + QQz + RRi + RR 2 + PP$
3Gi — ^ - { p + q + r ) — -l- a2 (p + q + T) = 3crx - — P + q + r

z z 2 3
3 p + q + r 9 a+ b+ c 3 ^ a + b + c
------<x2------ -------- 3(7! 02 --------------
2 3 2 3 2 3
= 30, - { S - { - 3 (« , - { * - - - » .* ) ■
where g = (a + b + c)/3 is the affix of the point of intersection of the

medians of A ABC (or, what is the same, the point of intersection of the
medians of /\PQR).
But since a + b + c = a±, the sum h' may be transformed as follows:
Let us now prove that h' is the affix of the orthocentre of A AhBhCh
formed by the feet of the altitudes of the given triangle. The equations of
BC and of the altitude from A to BC are of the form
z + bcz
be
z — bcz — a —
a
Adding, we find the affix ah of point Ah:
and, similarly,
If ca \ If ab \
bh = ~
2 I "1 ” T ) * Ck ~ ~2 ” ~ )
The slope of BhCh is
ab ac a(b2 — c2)
bh ~~ ch be
-a\
bh — c H __C_ c2 - b 2
ab ac abc
The equation of the altitude dropped from vertex Ah to side BhCh is
or
be 1
— a2z = — a1 a 2c*i +
2 2a 2be
or
z — a2z—= — 1 2- ------------------
1 o , ------a*G, , a* ~ b*c2 (35)
2 2 2o3
In similar fashion we can write down the equation of the altitude from 5k
to ChAh:
iB- 1 1 lb- , b* — c2a2
(36)
z — b2z — — o , ------ o2o, i ------------------
2 2 2os
Subtracting (36) from (35) term by term, we obtain
/t2 ax - 1 /La 2\ - b* — a* + c \b 2 — a2)
(6s - a2) z = - - (A2 - a2) ox --------------- ----------------
2 2 ct3
or, cancelling 62 — a2,
- -j, 1 _ a 2 + b2 + c2 1 _ o\ — 2o-2
Z h C7 O' i ‘
2 2o3 2 2o3
1- trf , o2 1_ crgOj , _
= Gi i ---- 1-------= — CTi b 0i
2 2<t3 (73 2 2
3 _ ^ 2^1 1 /0 — — v
= — 0i ------r — = - - (30i - 02)
2 2 2
and, hence,
h' = (3 0 ! - OjOa).
We now set up the equation of the straight line to which the resultant
thus found belongs. Since the nine indicated forces emanate three at a
time from a single point, their resultant is equal to the sum of three forces:
PP' = PPX + PP2 + PP3y
QQf = QQi + C02 + <203,
PR' = RRi -j- RR 2 + RRz>

these forces are laid off, respectively, from the points PyQ, R. From
the relation
1 _
P’ - P = ° i -
we find the affix p ' of point P' and, similarly, the affixes of the points
Q' and R':
P' = < 7 i - — / 7 - y G2P,
1 1 _
q' - — q — 0*2q,
2 2
1 1 _
r' = *i “ — r ------ <r2r.
2 2
Remark. If on a plane (which we consider oriented by the introduction
of a rectangular coordinate system) there is given a set of forces A kBk
{k = 1, 2, . . w), then the principal vector of their resultant can be found
as the sum of the free vectors:
K=
*= 1
The straight line to which the sum of forces belongs is a locus of points
M (x,y) for which
n
£ mom,, AkBk = 0.
A:= 1
Since, in the plane, it is natural to regard as the moment of force F the

cross product
momMF = (MT, F),
where T is the point of application of the force F, it follows that the

equation of the straight line to which the resultant of the system of forces
A kBk belongs is of the form
n x y ij
I ak a'k 1; = °»
(37)
bk b'k l!
where (ak, a'k) = Ak, (bk, b'k) = Bk.
Equation (37) is, generally, an equation of the first degree; hence, (37)
is, generally speaking, an equation of a straight line [the left-hand member
of (37) vanishes if and only if the principal vector of the system of forces
is zero].
The equation of the straight line containing the vector of the sum of
forces A kBk may be written differently by introducting the affixes ak
and bk of points A k and Bk:
2 Z 1
£ ak a_k 1 = 0 .
i =l bk bk 1
(38)
Applying this equation to the given problem, we have

1 1 _
z 'z 1 z p <7,------ p ------ a2p Z p G i — ---- G o P
2 2 2
______ 1_ 1 _ _____ 1 _
p p 1 = z P oi - - - p ~ — <r2P Z P gi - — g2P
2 2 2
3
p V 1 1 1 1 1
2
z f(-{ «•») + * ( - { P +
1 1 _
+ PGi — P<Ti + - - o2p 2 ---- GiP2~
2 ~2
Adding together three similar expressions that result from this one by
substituting q for p and then r for q, and equating this sum to zero
we obtain the equation of the desired line in the form (also note th a t
d + q + r = a + b + c = <r1)
z ( T f 1_ + ~2 52<Tl) + j
+ o2(p 2 + q 2 + r2) - — a2(p2 + q2 + r2) = 0
or
( “ \ 0l + 2 ffl° 2 ) Z + ( 2 ffl “ ~2 CTl<T2) ;
+ y o2(P2 + q 2 + r 2) — — o2(p2 + q2 + r2) = 0
or
(3dj— (TjO^z — (3(7!— djffa)? + a2(p2 + q2 + r2) — a2(p2 + q2+ r 2) = 0.
We then find
/>» = 4c2 - — + — + 2ac{i - 1) - 2cb(i + 1) + ab,

2 2
•t 2 •2
q2 = 4a2 — 1- —— (- 2ba{i — 1) — 2ac(i + 1) + be,
-
2 2
•2 •2
r 2 = 4fr2 - — + — + 2cb(i - 1) - 26a(/ + 1) + ca,
2 2
and, hence,
p2 + q 2 + r2 = 4(af - 2a2) - 3<t2 = 4a? - 1la 2.
From this it follows that
P2 + q2 + = 4 a| — l l a 2
and the equation of the straight line assumes the form
' (3ai — c^Gg) z — (3(7! — câ2)z + o 2(4a? — l l a 2) — a2(46? — 11g2) = 0
or
(3 gj — a1G2) z — (3a! — a ^ j z + 4(a?o2 — a 2o?) = 0. (39)
If instead of z we put into the left-hand member the affix of the endpoint
of the directed line segment 4OH, that is, 4a1? we obtain
4(3 gx — aiG2) ax — 4(3ax — a ^ ) g1 + 4a?o2 — 4a2o?
= 120^ ! — 4a?G2 — l2c1o1 -f 4aL6? + 4o[<j 2 ~~ 4a26? = 0
That is, the carrier of the resultant passes through the endpoint of the
directed line segment 4OH.
-------------- ----------------- *.
2°. Now suppose the orientations of the squares AXB2C C ", B ^ A 'A " ,
CÂzB’B" are the same but are opposite to the orientations of each of
the squares BCA1A2, CABXB2, ABCXC2.
In order to keep Fig. 16 as simple as possible, only the centres P*, Q*, R*
------------- >. -------------- >■ --------------->■
of the squares A ^ C 'C " , B ^ A 'A " , C1A2B,Bn are constructed; the points
P*9 Q*, R* are symmetric to the corresponding points P, Q and R with
respect to AXB2, BXC2, CXA2. The figure shows only three of the nine forces;
------> ------> ------ >
namely the following forces are constructed: P*Pf, P*P$, P*P3*, where
P*>P*>P* are the orthogonal projections of the point P* on the sides
PC, CA, AB.
The affix p* of point P* is
P* = b2 + — B2Ai + — B2AX
2 2
= h + " (fli - h ) + V — b2) = ~ ~ °i + bi

2 2 2 2
= [(1 + 0 C - ib] + [ia + (1 - 0 c] = — ~ b + —^— 0 .

2 2 2 2
Similarly
1+
q* = b,
1+
r* _
c,
where <7* and r* are the affixes of points Q* and R*. Note that in this
case as well,
p* + q* + r* = a + b + c = < tv
Now, from the equations of the lines BC and P*P?9

z + fccz = b + c,
z — 6cz = p* — bcp*,
we find the affix p* of point P * :
/>* = ~ (b + c + p* — bcp*)
2
and, hence,
p*p* — p* _ p* = J_ q, _(_ c _ p* _ bcp*).
2
Similarly,
p * p * = — (c _|_ a _ p* _ ca/>*),
2
P V 3* = — (a + b - p* - abp*)
2
and, consequently,
P*P* + P*P* + P*P* = a 1 - — p * - - L o-2p *
2 2
9 -810
and
Q*Qt + 0 * 0 ! + Q*Q! = ° 1 - 4 q* - \ <r2q*>

2 2
R*R* + R*R? + R*R* = crx - — r* - — < t27*,

2 2
----- >' ------
and so the principal vector of the sum of nine forces P*P*> P*P2* ,... is
equal to
C = 3cTi — -~-(p* + q* +/•*) —— <r2 (p* + q* + 7*)

2 2
, 3 1 _ 3 1 _ 1
— 3(7x (JI ^2a l — ~ &1 °’2(71 — ~ (3c! 0 2Ox) — ll i
2 2 2 2 2
where ti is the affix of the orthocentre of A AhBhCh.
As in the case of item 1°, let us now consider the sums of nine forces
taken three at a time:
P*P'* = P*P f + P*P* + P*P*9
¥ & * = ~ Q * Q t + 0 * 0 ! +~Q*Q!,
R*R'* = R*R? + R*R? + R*R*.
Let us find the affix p'* of point P'*:
P * ? * = p ’* - p * = <T1 - ^ - p * a2p*,
2 2
whence we find p'* and, similarly, q'* and r'*:
The equation of the straight line carrying the resultant is of the form
z p * p f* z q* q ' * z r* r’*
z p* p'* -1- z q* q'* + z 7* 7 ’*
1 1 1 1 1 1 1 1 1
We have
z p* p' * z p * C l - - _ c 2p *
z p* a 1 — ■— OiP*
2j 2
1 1_ * 1
Z 1* F * = z P * i—— P ~ ~ G*P z P* Ol - — ViP*
2 2 2
_3_
1 1 1 1 1 1 1
2
= Âp** - ffl + A a2p*jz +

1 A <r2p* - —/>*) z
= ( i
+ P*o1 — 1*0 1 + A <j2p *2 — A f f 2p*».
2 2
If we write down two other similar expressions and then add them and
equate the result to zero, we obtain the equation of that line in the case
of item 2°;
Â Ox - + A o=a j Z + ^3<rx- A - A <Tj j Z

+ <7!?! — (Tjffi + A ^ (p * 2 + q* 2 + r*2) — A ct2(p*2 + ?*2 + r*2) = 0
or
(3aj — 0±02) z — (3 ^ — 0X02) z
+ ^(P*2 + tf*2 + r*2) — ^ 2(P*2 + q *2 + r *2) = 0
Furthermore,
\2 / 1_ ; 1 _i_ / \ 2
- I 1- ' b 1 1+
I 2 4+ 2
i
____j
+ Cfl
2 J 2
+ _L (fl2 + 62 + C 2) = ff2,
2
o-2(P*2 + tf*2 + r*2) = oza2.
And therefore
<*2(P*2 + q*2 + "r*2) = <72^2,
which means our equation takes the form
(3d1! — 0 ^ 2) z — (3oi — 0X02) z = 0.
Problem 14. Let P, Q, R be the orthogonal projections of point M

on the sides BC, CA, AB of triangle ABC (Fig. 17). Denote by A', B', C*
the points obtained by inversion of the midpoints A0, B0, C0 of the segments
MA, MB, MC with the circle of inversion (PQR) and by A", B", C",
the triangle formed by the polar lines of the points A0, B0, C0 with respect
to the same circle {PQR). Prove that
(A'B'C')2 = 1 (A"B"C")-(PQR).
What is the necessary and sufficient condition for A A'B'C' and A PQR
to have opposite orientations?
Solution. We take (PQR) for the unit circle. Let zl9 z2, z3 be the affixes
of the points P, Q, R and let fi be the affix of point M.
The slope of the straight line MP is
p — z1
and hence the slope of line BC
\ i - z1
and the equation of line BC is of the form
z— J* — *i (z - Zj). (40)
f l — zi
Similarly, the equation of line CA is
* -* 1 = - - ^ —
(* -* ■ )• (41)
P - *2
Subtracting equation (41) from (40) term by term, we find the complex
number z = c; it is the conjugate of the affix c of point C:
z2 - * = ( a - * . _ c + z, - z2
\H -z 2 n - z j H -Z i U— zi
or
Zi + ZA
z l) (z2 - zxy0
- 1
V ZlZ2 ZlZ2 ) zxz2
Zo z, —
(/« - * 1) - Z2) 0 < - z'l) 0 * - z 2)
Cancelling z2 — zx and multiplying both sides by (p— z x) (ji — z 2), we get
z . + z 8 \ . - , P/Z - 1
(p - ij) (p - z 2) = [ - p ----- -— b —1- ^ L ) C H--------------»
V ZlZ2 zxz2 / Z iZ 2
whence
or
( - \ i -
Z1Z2
h ± lC ] c
ZiZ2 )
(/Z - Zi) (n - z2) - Z!Z2(/l/t - 1)
( - /i - p z xz 2 + Zj + z2) c = /I2 - p (zx + z 2) + (2 - p/i) ZjZ,.

This means that
nz — n (zx + z 2) + (2 — nn) z xz 2
c=
and, hence.
p2 - p(zx + Z2) + (2 - nn) z1z2
— H ~ /iz1z2 + zx + z2
or
— n2 + n(zi + z2) + (nn - 2) zxz2
H + fi zxz2 — zx — z2
The affix c0 of the midpoint of segment MC is
1 i —^ + K zi + z2) + ( 0 ~ 2 ) z1z2\
<(i I P~r )
2 V fi + V zxz2 - (zx + z2) }
_ (pp - 1) zxz2
/i + /zzxz2 — Zx — z2
and the affix c' of point C' obtained from C0 by inversion with the circle
of inversion (PQR) is
1 H + f i z ^ z — zx — z 2 ft + pztz3 — z1 — z3
Co (mP - 1) z t z t fiji - 1
We find the affixes a' and b' of points A' and B' in similar fashion:
B + PZ2 Z» ~ Z 2 - Z3 *
a —--------------------------
HH 1
H + fizi z 1 Z3 Z1
b' =
ftp. - 1
We now find
(A'B’C )
. a’ a ' \ ft -f- ftZ3Z3 Z2 Zg ft-(- flZ3Z3 Z2 Zg 1|
b’ b' 1 ft + pz3zx — z3 — zt p + ftz3z\ — z 3 —z x 1!
4(fift—l )2
c' c' 1 |ft + pZiZ2 — Zj — z2 p + ftz{zz — z x — z 2 1
ftz3z3 — Z2 — Z3 flZzZ3 z2 z3 1
pz3zt — z3 — zx /JZ3ZX —z 3 — Zx
4 { fi ft- \ f
f . I Z^Z 2 — Z i Z2 f l Z - ^ Z ^ 2"2
jU(73Zi 1*02?! ^1 "t"

|K 7 g Z 2 ^ 1 't“ ^2 ^ 1 ■(■^ 2
4 (^ -1 f
fi(^2z z 0 1 + ^ 3 IIG&Z ^1 + z 3
/ i f f 3 Zx + Zx f ia ^ i + Zx 1
/Zff3z 2 + Z2 jUff3Z2 + z 2 1
4(ftn—l )2 pa3z 3 + z3 /iff3z3 + z 3 1
■ ■
Zx Zx 1 Zx Zx 1
f t p a 3<f3 z 2 z2 1 + z2 z 2 1
4(ftp ~ l ) s
_ 1 Z3 z3 1 Z3 Z S 1 .
/(I - fi/i) (W )
Z2 z2 I
4(1 - w ? 1—
*3 *3 1
Furthermore, the slope of the straight line OA0 is
(ftp — 1) z2z3 (ftp — 1) z 2z 3
«O^0 =
ft + pz3z3 — z2 z3 p-\- ftz3z 3 z 2 z 3
(ftp — 1) z2z3 ftp 1
- *
Z.>Z.
243
ft + ftz3z3 — Z2 — Z3 /i + /Zz2z3 — z2 — z3
From this it is easy to find the equations of the polars of points A0 and B0
with respect to the circle (PQR); these are straight lines that pass through
points A' and Bf respectively and are perpendiculars to the lines OA'
and OB' ;
2—
/I PZ2 Z2 Z2 & + Hz2z 3 z 2 z 3
- z2z3 (I -z ------
Hi* - 1 Hi* - 1
and
H + HZ3 Z1 — z 3 — z 1 H+ 1— z 3— z 1
Hi* — 1 Hi* - 1
The affix z = c" of point C", the point of intersection of these lines,
can be found from the equation
_____1 _ ( 2 _ H + i*z &3 ~ Z2 ~ z a\ H + H ^ 2*3 ~ z~a — z 3
z2z3 V Hi* — 1 ) Hi*— 1
= _ 1 / _ H + i*zszi — z3 — \ + H + HZ3 Z1 — z 3 —z x
% V Hi* — 1 / Hi* — 1
or
Z ^ 2 /j + /^ 2z3 — z2 - z 3 = ___ z _ 2 n + /iz3zx — z3 — zx
Z2Z3 Z2Z3(HH 1) Z3Z1 Z3Z1(HH — 1)
(z2 — zx) z
0’s zi ]■
whence
c" = — ^ — - (H - z3).
HH ~ 1
Similarly,
a” — z~~~r (h — Zi)
/i/t - 1
and
b" = 3—— (/i — z2)
HH ~ 1
m> that
/ 4 H ~ zi H - z i 1
( l"/i"C")
4 ( p ji- i f H ~ *2 H — Z2 1
H - z 3 ji-~ z 3 1
Zi z x 1
4 i
Z2 z 2 1 - (Pfi*).
(HH- i f 4 Zj Z3 1 (HH— I)2
Thus
(A'B'C') = ~^PQR)~, (.A"B"C") = ------ (PQR),
I - fiH (1 - n n f
whence
(A 'B 'C 'f= (p QRf = _L
(1 - w ? 4
The triangles A'B'C' and PQR have the same orientation if and only
if 1 — pp > 0, that is, the power of point M with respect to the circle
(PQR) is negative; in other words, if and only if the point M lies inside
(PQR). In Fig. 17, M lies inside (PQR) and, indeed, A'B'C'i\PQR.
Problem 15. Given A ABC. Through an arbitrary point P of circle
(ABC) circumscribed about A ABC are drawn lines parallel to the sides
BC, CA, AB. Let A', B', C' be the respective second points of intersection
of these lines with the circle (ABC). Denote by A", B", C" the points
symmetric to points A ,B ,C about the straight lines B'C', C'A', A'B'.
1°. Prove that A ABC and A A"B"C" are congruent but have opposite
orientations.
2°. Prove that 0 0 " = PH, where O and O" are centres of (ABC)
and (A"B"C") and H is the orthocentre of A ABC (Fig. 18).
Solution. Version One. 1°. Take
(ABC) for the unit circle and assign
to point P an affix of 1 (that is, let P
be the unit point). Then the affixes a ',
b', c' of points A', B', C' are
a — z2 z3, b — z2Zi, c — z±z2,
where zl9 z2, z3 are the affixes of the
vertices A, B, C of the triangle. The
slope of line B'C' is
Zj z2 - zx z3 ^ zt(z2 — z3)
~Zi Z2 Z1 .1
Zi \ Z2 z3
= — z f Z2 z 3 = — Zi (T3
and so the equation of line B'C' is of

the form
Z — Z x z3 = — Zx <72 ( — Z x Z 3)
z
or
Fig. 18 Z + Zjl C73 z = zx z2 + zx z3. (42)
The equation of a perpendicular dropped from point A to B'C' is

z — zx = zx a3(z — Zj)
or
z — z1cr3 z = z1 — <j3. (43)
Combining equations (42) and (43) term by term, we find the affix z = ax
of the orthogonal projection A1 of point A on the line B'C':
ax = — (zx + zx z2 + zx z3 — cr3) = --- (zx — z2 z3 + <j2 — c3).

2 2
The affix a" of point A ", symmetric to A with respect to B'C', is found
from the relation
*1 + a"
whence
Cl — 2flj Zx — (72 (73 Z2 Z3.
In similar fashion we find the affixes b" and c" of points B" and C":
b — (7% (73 Z3ZX, C = (72 Ct3 zx Z3.
From these relations it follows that the centre O" of circle (A"B"C")
has the affix
O = (72 (73,
and the radius is equal to 1 since
| a " - o " | - \b" - o " \ = \c” — o"\ = 1
= !)•
------- *
Let us consider 1\A qB'qC3 the affixes of the vertices of which are equal
respectively to —z2z3, —z3zl5 —zxz2. The triangle A'qBqCq is symmetric
--------> ------- >
to A A'B'C' with respect to the point 0 , and, hence, the triangle A"B"C"
is obtained by a translation of /\A'0B'0C'0 via the directed line segment OTt
where the affix t of point T is equal to <t2 — (73.
Furthermore,
(73 ... —
^2 ?3 — — &3 ^ 1? ^3 ^1 — ^3 Z2) Zj Z3 — C3 Z3.
*1
Therefore AôôCo is obtained from A A*B*C* (the affixes of the vertices

of which are z l5 z 2, z 3) by a rotation about 0 through the angle arg (— a3).
--------> ------>
The triangle A*B*C* is symmetric to AABC about line OP (the real axis
------► ------- >■
Ox), and therefore A ABC and A A*B*C* are equal and have opposite
--------> -------->
orientations. But AA*B*C* is equal to A A qBqCq and has the same orien
tation; A A qBqCo is equal to A A 'B 'C and has the same orientation, while
A A "B ”C" is equal to A A qBqCq and has the same orientation. Conse
quently, A B C l\A "B "C " and A ABC = A A”B"C" (the symbol = here
signifies congruence).
2°. The affix of point O" is equal to o" = <r2 — <r3, whence
OO” = |<72 — 0*31 = |<73| = ! — - 1 \ = \s l - I H ^ I - l\= P H .

\ff3 I
Version Two. 1°. Take {ABC) for the unit circle. Let zly z2, z3, p be
the affixes of points A, B, C, P respectively. Take the Boutain point of
A ABC for the unit point so that a3 = 1. The affixes a', b', c’ of points
A ’, B \ C' are
a , = -----
Z2Z3> »b/ = Z3 Zl ct= -----
----- Zl Z2
P P P
or (c3 = 1)
a’ = p z x, b' = p z 2, c' = p z 3.
The slope of line B’C is
b’ — c = p z 3 — p z 3 - J — = - ZL (a3 = 1)
b' — c' pz3 —pz3 z2 z3p 2 p2
and the equation of B’C is of the form
z — p z 2 = — zi p \ z — p z 2)
or
z+ Zi ~p2 z = p z 2 + p z 3. (44)
The equation of the perpendicular dropped from point A on line B C is
z - zx = Z1 p \ z - Zl)
or
Z — Zj ~p2-
p 2 z = z x (45)
Combining the equations (44) and (45) term by term, we find the affix
z = ax of projection AL of point A on the straight line B'C
The affix a" of point A", symmetric to point A about B 'C \ is found
from the relation
+ a"
= *i
2
whence
a" = lay — z, = p z 3 + p z 3 — ~p2.
Similarly,
b" = p z 3 + p z x — p 2.
We now find
a" — b" = p ( z 2 — z , ) ,
a" — b" —P(z2 - H)
and, hence,
A"B"2 = (a" - b") (d" - b") = PP(?2 - H) (z 2 - zi) = AB2.
That is. A"B" = AB. In similar manner it can be proved that B"C" = BC,
C"A" = CA, that is A ABC and A A"B"C" are congruent.
------> ------- >
In order to prove that A ABC and AA"B"C" have opposite orienta
tions, it suffices to prove that
d" 1
A = z2 b" 1 = 0 .
^3 c" 1
We have
Zy p(Z2 + Z3) - P2 1 Zl pip 1 - Zl) 1 Zi — #>Zi 1
A = Z2 P(Z3 + Zl) — P2 1 == Z2 pia 1— Z2) 1 = z2 — pz2 1
Zg p(Zy + Z3) - P2 1 z3 P in - z3) 1 H — PH 1
and so A ABC = A A"B"C" and ABC A"B"C".

2°. From the formulas
a" = p z 2 + p z 3 - p 2 = >(<?! - z ^ ) - p 2 = - p z y + p a y - p 2,
b" = — p z i + p t i - p 2,
c" = — p z 3 + ~pdy — p 2
it follows that the affix o" of the centre O" of the circle (A"B"€") is
o" = pdy - p 2
and the radius is equal to 1 (|— p z x\ = |— p z 2| = |— p z 3\ = 1), whence
()()" — \o"\ = |pay — P 21 - \p\ Wy — p\ = \&i — p\ = I<Sy — p\ — PH.
Remark, From the relation o" = p{dx — p) it follows that the directed
line segment 0 0 " is equivalent to the directed line segment obtained by
symmetry in the x-axis of the directed line segment PH and a subsequent
rotation through the angle argp (the x-axis is a straight line passing
through point O and the Boutain point).
------►
Problem 16. On the sides BC, CA, AB are constructed triangles A'BC,
------> ------>
B'CA, C A B , which are similar and have the same orientation. Let P be
an arbitrary point lying on the circle (O) = {ABC), The directed line
segments OA', OB', O C rotate about point O through angles that are
respectively equal to {OP, OA), {OP, OB), {OP, OC) (these angles are
oriented). Let OA", OB", OC" be the respectively rotated segments.
Prove that the centroid G" of A A"B"C" is symmetric to the centroid G
of A ABC about the diameter of the circle {O), which diameter is parallel
to the Simson line constructed for point P with respect to A ABC,
Solution, We take the circle (O) = {ABC) for the unit circle, and the
point P for the unit point. Let zl9 z2, z3 be the affixes of the points A, B, C.
Denote by a', b', c' the affixes of the points A', B', C , Since the triangles
------> ------> ------>
A'BC, B'CA, C'AB are similar and have the same orientation, it follows,
assuming
CA'
cp = {CB, CA'), P=
CB ’
we will have
a = z3 + p{cos <p + / sin (p) (z2 — z3) = az2 + (1 — a) z3,
where a = p(cos <p + i sin cp), and similar expressions for b' and c' with
the same value of a:
b' = otz3 + (1 — a) zu
c' = OLZi + (1 — a) z2.
The affixes a", b", c" of the points A", B", C" are
a" = zx a', b" = z2 b', c" = z3 c'.
That is,
a" = zxz2a + (I — a) zLz3,
b" = z2z3 a + (1 — a) z2 zl9
c" = z3 zxa + (1 — a) z3 z2.
From this we find the affix g" of the centroid of A A"B"C":

,, _ a" + b" + c" C2
where <r2 = zx z2 + z2z3 + z3 zx.

The equation of the diameter of (0), which diameter is parallel to the
Simson line constructed for the unit point P with respect to A ABC,
is of the form (see problem 3)
z — <73 z = 0 .
The ends of that diameter have affixes l/og (tfo3 has two values; each of
them satisfies the equation z — a2z' = 0).
In order to prove that the points G and G" are symmetric about the
diameter z — cr3 z = 0, it suffices to prove that AODG and AODG"
are similar but have opposite orientations (D is one of the ends of the
diameter z — <j3 z = 0), that is, that
0 0 1
A = g g" 1 = 0
]fc3 l/ff3 1
(for/< 73, any one of the two values may be taken; ][g3 is the conjugate
of that value). We then have
A = ][oz g - 1/V3g" = 1faz (g - c3g")
Problem 17. The altitudes of an arbitrary triangle ABC intersect the

circle (O) = {ABC) in the points Al9 Bl9 Cx; A \ B', C9are points symmetric
to point P9 which lies on (0), with respect to the straight lines OA, OB,
OC; A", jB'\ C" are points symmetric to the point P with respect to the
lines OA', OB', OC'; a, /?, y are points symmetric to the points A "9B", C"
about the line OP. Prove that the points A2, B2, C2, which are symmetric
to the points a, y with respect to the tangents to (0) at the points Al9
Bx, Ci, form a triangle A2B2C2 that is homothetic to A AxBiCi with ratio
equal to 2; the centre Q of this homothetic transformation belongs to
(he circle (0 ).
Solution. Take (0) = {ABC) for the unit circle and P for the unit
point of the complex-variable plane. Let z1,[z2,z 3 be the affixes of the
points A, B, C.
The equation of the straight line BC is of the form
z + z2z3~z = z2 + z3.
The equation of an altitude dropped from the vertex A on the line BC is

2 — Zx = Z2 Z3( z — Z j.
Solving this equation together with the equation zz — 1 of the unit circle,
we get
Z - Z i = Z2Z3 / ----------
1 1 Z Zj
or z — Zi = Z* 2a
V Z Zj Zt Z
One of the roots of this equation is z = zx (the affix of point A); the
other is
Z2 Z3
z = ax
Z1
(the affix of point AJ. Similarly,
*3 *1 zl z 2
bi = -
*2
The equation of the line OA is
z — z\ z = 0
and the equation of a perpendicular dropped from point P on the same
line is
z - 1 - - zf (z - 1).
Solve this equation together with the equation zz = 1 of the unit circle:
z — 1 = — z\ ----- 1j , that is, z — 1 == zf -----
One of the roots of this equation is z = 1 (the affix of point P), the other
is the affix a' of point A':
z = a' = zf.
Similarly,
b' = zh cf = z l
are the affixes of points B' and C".
The equation of the straight line OA' is
z — z\ ~z = 0
and the equation of the straight line passing through point P perpen
dicularly to line OA' is
z - 1 = - z\ (z - 1).
From this and from the equation zz = 1 of the unit circle we find the affix a"
of point A":
z —1
z— that is, z — 1 = z?
z
One of the roots of this equation is z = 1 (the affix of point P), the other is
z = a" = z\.
Similarly,
b" = zi, c" = z l
where b” and c" are the affixes of the points B,r and C".
The slope of the straight line OP is equal to 1. Hence, the equation
of the straight line passing through point A ” perpendicularly to line OP
is of the form
z - a" = - (z - a").
Solving this equation together with the equation zz = 1 of the unit circle,
we find the affix A of point a:
One of the roots of this equation is z = a" (the affix of point A"); the
other is
which is the affix of point a. In similar fashion we find the affixes /z and v
of the points p and y:
1 1
The slope of line OAx is

<h_ = q2 = 4 4
<*i 1 4
and so the equation of the tangent line to the unit circle (O) at the point A1
is of the form
or
-2 -2
z2 z3 —
z -------- Z
2
(46)
z1
The equation of the perpendicular dropped from the point a on this
tangent line is of the form
z ---- i = 4- ^4 ( z - z t )
4
or
-2 -2
Z2 Z3 — __
— 4- (47)
4
Combining equations (46) and (47) term by term, we find the affix a*
of the projection of point a on the tangent line:
Z? Zo
z = a? =
24
The affix a2 of point A2, which is symmetric to point a with respect to
the tangent to the unit circle at point Al9 is found from the relation
^1 + a2
= a* 9
whence
'l ♦ 1
a2 — 2a2 — — — 2O —
^ 2 Z3 2
------ ~\— i ^
----------^ Z 2 Z3
= — 2 -------- a3.2
Zi zj z* ZX
In similar fashion we find the affixes b2 and c2 of points B2 and C2:
L 9 Z3 “ 1 2 ^ Z1 Z2 2
b 2 — 2 03, c2 — — 2 -------- — 0^3.
Z2 Z3
From these relations it follows that the straight line A XA2 passes through
point Q with affix
q = o\
since the midpoint of segment A2Q has the affix
a2 + q z3 z2
— = --------=
2 zx
which is to say the midpoint of A2Q coincides with point Av Point Q lies
on the unit circle since \q\ = 1.
In similar fashion, we can prove that the points Bx and Cx are, respec
tively, the midpoints of line segments B2Q and C2 Q. Thus,
QA 2 _QB 2__QC2
= 2.
QA± QB± QC1
That is, A A 2 B2 C2 is an image of A A 1 B1 C1 under a homothetic transfor
mation with centre Q lying on {ABC) and with homothetic ratio 2.
Problem 18. 1°. Through the vertices A l 9 A2, A 3 of A A YA 2 A 3 lying
on an oriented plane, draw parallel lines intersecting the given straight
line A at the points Pl 9 P2, P3; note that the angle from the line A to the
lines A±Pl 9 A 2 P2, A 3 P3 is equal to a (Fig. 19).
Through points Bi, P 29 P3 draw straight lines ll 9 /2, 13 that intersect
the respective sides A 2 A3, A 3 A l 9 A±A2; note that angles reckoned from
the straight lines A 2 A 39 A 3 Al 9 A 1 A 2 to the straight lines ll 9 /2, l3 are all
equal to /?. Prove that the lines ll 9 /2, l3 form a triangle QiQ 2 Qz, which is
similar to the triangle AXA 2 A3; the factor of proportionality is
sin(a + P) j
sin a |
Consider the following special cases;
2°. P = 7i — a.
3°. p = 0.
Solution. 1°. Take the circle (O) = (A 1 A 2 A3) for the unit circle. Let
r j5 z2, z3 be the affixes of the points A l 9 A 29 A 3 and let
az + a ~z b
be the equation of the straight line A (a ^ 0 and b is a real number).
Put
X = cos 2a + i sin 2a, ju = cos 2p + i sin 2p.
The equation of line AXPX may be written in the form
a(z — Zj) + 2 a (z — z x) = 0 . (48)
Indeed, the slope of the straight line A is
a
x = ------
a
iuul the slope of the line (48) is
I lenee
X
In Ml()
and so
= cos a + / sin a.
That is, the angle from the straight line A to the straight line (48) is equal
to a.
If we took the other value of ][X9
^X = — (cos a + i sin a) = cos(a + n) + / sin(a + n)9
then the angle from the straight line A to the straight line (48) would turn
out equal to n + a, that is, it would be congruent to a modulo it. From
the system of equations of the straight lines A and AxPl9 that is, from
the system of equations
az + a z = b9
az + Xa z = azx + Aa z l9
we find the affix px of point Px:
whence
z =Pi = —
a a )•
1
A / 1 b a zi\_ _ A f 1 b azx\
A — 1 \ zx Xa X a ) 1 — A\ zx Xa a j
The affixes p2 and pz of points P2 and P 3 have similar expressions:

I f lb , , a~z2\
P2 ------ Z2 - - + A — 2 ,
1 —x V ci a )
P3
Furthermore, the equation of line A2A3 is of the form

z — z2 = — z2z3(~z — z 2)
and so the equation of line 7Xmay be written in the form
z - Pi = — HZ2 zz(z - p 2)
or
_L Zb + Zd = _ , ZiZ h
z _ __L _» 3 - _ A- ( _ ± +
1 —1 v a azx ) [ 1 —1 \ zx la la ) J
or
<1 - A)(z + „ w ) = z, - ^ + + (49)
a az1 zx a a
The equations of the straight lines /2 and /3 are written down in similar
fashion:
n
(1 X) (z +t p....z 3z 2 z ) — z2 tyzaZi ,
|---------------------- Paa3
j----- —----------::—* (50)
a az2 z2
^ , , -2-) = *, - + di _ _ « ? .. (so
#z3 z3 a a
Multiplying both sides of equation (49) by —z l9 both sides of (50)
by z2, and adding termwise, we obtain
(1 - 1) (z2 - Zj) z
= 4 — Z1 — ~ (Z2 — Zl) + Z3(Z2 — Zl) lp — P -G- (z2 — Zl)

a
whence we get the affix q3 of point Q3, the point of intersection of the
lines lx and /2:
lb a
Z1 + z2 -------- \- Ip Z3 — p c3
a a
O - l---------------- / / — - <X3 + (>1 /2 — l ) z 3

a a1
1 —1
The affixes and q2 of the points Q± and Q2 have similar expressions:

Ab a
a i ----------------- <r3 + (Aft — 1) zx
a a
<h r r i »
Xb a
o-i-----------p — a3 + (Xî - 1) z2
a a
<?2 = 1 —X
From the last three relations it follows that the points Ql 9 Q2, Q3 are
obtained from the points Al 9 A 2 9 A 3 by a linear transformation of the
first kind:
q = mz + n,
where
Xfi — 1
m
1-X
Xb a
—------ p — <r3
a a
and so A A tA 2 A 3 and A Q 1 Q2 Q3 are similar and have the same orientation

(see Fig. 19): n is the affix of point 0 \ into which the point O passes
under the transformation; in other words, (O') = (QiQ^Qz)- The factor
of proportionality is
_ [ ^P — 1 _ lcos + 2/0 + / sin(2a + 2/?) — 11
| 1 —X 11 — cos 2a — i sin 2a|
12 sin2(a + p) — 2/ sin(a + /?) cos(a + /?)|
|2 sin2 a — 2 i sin a cos a|
| sin(a + P)| | sin(a + /?) — 1cos (a + /?) __ sin(a + /?)
|sina| |sina — /co sa| sin a
2°. If p = 7T— a (or /? = — a), then A 6i0203 degenerates into a point.
We have the following theorem: if through the vertices Al 9 A 2 9 A 3 of
A A xA 2 A 3 lying on an oriented plane we draw parallel lines intersecting
the given straight line A at points P l 9 P 29 P3, the angle from the straight
line A to the straight lines A±Pl 9 A 2 P2, A 3 P3 being equal to a, and then
through the points P l 9 P2, P 3 we draw straight lines ll 9 /2, 4 cutting the
sides A 2 A g9 A 3 A l 9 AxA 29 the angles from the straight lines A 2 A3, A 3 A l 9 AXA 2
to the straight lines ll 9 4 , 4 respectively being equal to a, then the lines
4, 4> 4 pass through one point (Fig. 20).
In particular, if a = n/2 9 P = n/2 , we obtain the theorem on the orthopole

of a straight line with respect to a triangle: if P l 9 P2, P3 are the orthogonal
projections of the vertices A l 9 A2, A 3 of A A XA 2 A 3 on a given straight line A,
l h e n the lines that pass through the points Pl 9 P2, P3 and that are perpen
dicular to the straight lines A 2 A3, A 3 A l 9 AXA 2 respectively, intersect in
one point Q (the so-called orthopole o f the straight line A with respect
to /\A xA 2 A3\ Fig. 21).
V’. Suppose P = 0. Then we obtain the following theorem: if through
lhe vertices A l 9 A2, A 3 of A A XA 2 A 3 we draw parallel straight lines inter
s e c t i n g the given line A in the points Pl 9 P 2 9 P3, and then draw through
the points Pl 9 P2) P3 lines ll 9 /2, l3 respectively parallel to lines A 2 A3, A 3 Al 9
------- >
then the lines ll 9 /2, 13 form a triangle Q1 Q2 Q3 that is equal to the
------- >-
triiingle AYA 2 A 3 and has the same orientation (Fig. 22).
Fig. 23
Problem 19. Let A 1 A 2 A 3 A4 be a convex quadrangle inscribed in a
circle (0). Denote by A12, A23, ^ 34, ^ 4i the midpoints of its sides î/42,
^ 2^ 3? ^ 3^ 45A 4 Av Let <5 be some diameter of (0). Denote by a12, a23,
a34, a41 the orthogonal projections of the points A12, A<&,A 349 A 41 on the
diameter <5. Draw through the points a12, a23, a34, a41 straight lines p349 p l 4 9
p 129 P23 that are respectively perpendicular to the straight lines A 3 A4, A 4 A l 9
A±A2, A 2 A3. Prove that the lines p 349 P149 P12, p 23 form a quadrangle
B 1 B 2 B3 B4 similar to the given one but with orientation opposite that of
the quadrangle AXA 2 A 3 A 4 (Bx is the point of intersection of the lines p 23
and p34, B2 is the point of intersection of the lines P1 4 and P34, B3 is the
point of intersection of the lines p i 2 and P149 B 4 is the point of intersection
of the lines p 12 and p23; Fig. 23).
Solution. Regard (0 ) as the unit circle, and take the .x-axis so that the
diameter S lies on it. Let z l 9 z2, z3, z4 be the respective affixes of the points
A l 9 A 29 A3, A4. The midpoint A 1 2 of segment A 4 A 2 has the affix
and its projection aI2 on the line S (the x-axis) has the affix
zi + z2 H------ 1-----
_ a\2 ~f~ ^12 ________ Z1 Z2 ( 1|+ Zj z2) (z j+ z2)
T“ = — 2— 4 4z4 z2
The equation of the perpendicular P3 4 dropped from the point a12 to
the line A 3 A 4 (the slope of which is equal to —z3z4) is of the form
z t 12 = z3 z4(z f 12)
or
(1 + *i z2) (zi + z 2) (1 + Zi z2) (zl + z2)
z ------------------------------ = z3z4
4zx z, L 4Zj z2
or
(zx + z2) (1 + ^ z2) (1 - z3 z4)
Z Zo Z aZ (52)
4zj z2
Similarly, the equation of the perpendicular /?23 dropped from the point a 14
to the line A2A3 is of the form
_ _ - (Zl + Z4) (1 + Zj z4) (1 — z2 z3)
z — z2 z3 z = --------- (53)
4z4 z4
Multiplying both sides of equation (52) by z2, both sides of (53) by —z4,
and adding termwise, we obtain the affix z — bx of point Bx, the point
of intersection of the lines fi3i and /?23:
(Zj + z2) (1 + Zx z2) (1 — z3 z4) - ( z 4+ z4) (1 + z 4 z4) (1 - z 2 z3)
(z2 - z4) =
4zx
1
= —- (Zl + z2 + zj z2 + zl z4 — zx z3 z4 — z2z3 z4 — zf z2z3z4
4zj
-z2Zi Z3 z4 z4 z4 Zx Z4 Z4 ZJ -(- Zj Z2 Z3 + Z 2 Z3 Z4 -fZ | z4 z2 z3+zf z2 z3 z4)
= [z2 - z4 + zj(z2 - z4) + Zx z3(z2 - z4)

4zt
+ Zx(z| - z\) - zx z2 z3 z4(z2 - z4)],
whence
1 4~ z\ -j- zx z3 -f- zt z2 -j- z4 z 4 — Zx z2z3 z4 _ 1 — ff4 Ox
bi =
4z4 4zx 4 ’
where
o-x = z4 Z3 + Z4,
^4 — ^2 ^3 ^4*
Similarly we find the affixes b2, b3, bt of points B,, B3, Bx:
b3 — 1 - <?4 1 ffl-
4z2 h 4
1 - <r4
b3 =
4z3 1 4
1 - <t4
K =
4z4 4
To summarize:
bk = T + - — ( * = 1, 2,3,4).
4 4
From this relation it follows that the points Bl9 B2, B39 B4 lie on the circle
(BXB2B3B4)9the affix of whose centre is equal to aJA and the radius is 11 —
— <t4!/4. Since |<r4| = 1, it follows that the radius can vary from 0 to 1/2
(cr4 can assume the values ± 1).
Thus, if o4 — - 1, that is the Boutain point of the quadrangle is the unit
point, then the quadrangle BXB2B3B4 contracts to a point: all the lines
fi34, P23 pass through the centroid G (trJA) of the system of points
Al9 A2, A3, A4, to which are assigned equal masses.
If a1 = 0, that is, if the centroid of four points Al9 A2, A39 A 4 coincides
with the centre O of the circle (A1A2A3A4) = (0), then (AXA2A3A4) and
(BXB2B3B4) are concentric circles.
Finally, if the unit point does not coincide with any one of the four
Boutain points of the quadrangle A1A2A3A4 (these Boutain points form
the vertices of a square inscribed in the circle (0 )), then the quadrangle
--------- >
B1B2B3B4 does not degenerate. It is an image of the quadrangle AXA2A3A4
under a similarity transformation of the second kind:
We first take the points A[, A29 A 3, A4 that are symmetric respectively
to the points Al9 A2, A3, A4 with respect to the jt-axis, and then the quad
rangle A[A»A3A4 is rotated about point 0 through the angle arg -----
4
---------------->
and the rotated quadrangle A " A2A3 A4 is subjected to ahomothetic trans
formation with centre 0 and ratio |1 — a 4!/4; we get a quadrangle
---------------- >
d x 'A2 'A3 'A4 '; finally, the last quadrangle is subjected to a parallel transla
tion determined by the directed line segment OG, where the affix g of point G
is equal to <7j /4 (G is the centroid of the system of four points Al9 A2, A39 A4
--------- >
to which are assigned equal masses). We obtain a quadrangle B4B2B3B4.
Of all these transformations, only the first one (symmetry about the
----------- --------------->
x-axis) changes the orientation and, hence, and
the quadrangles A1A2A3A4 and BlB2B3B4 are similar.
Problem 20. A triangle ABC is inscribed in a circle with centre 0 ;
A 0, B0, C0 are the centres of the circles (OBC), (OCA), (OAB); Al9 Bl9 C4
are points symmetric to the points A0, B0, C0 about BC, CA, AB respec
tively. Prove that the orthocentre H of A ABC is the centre of one of the
circles (S) tangent to the straight lines BxCl9 CxAl9 AXBX and that the
circle (5) passes through the centre of the Euler circle of AABC. Prove
that the radius of (5) is equal to ~ OH.
Proof. Assume {ABC) to be the unit circle. Draw tangent lines to {ABC)
at the points A , B, C. They form a triangle A2B2C2. The circle {ABC) = {O)
is inscribed in A A 2B2C2 if ABC is an acute-angled triangle, and escribed
if ABC is an obtuse-angled triangle; if, for example, the angle C is
obtuse, then the circle {O) is escribed in the angle C2 of triangle A2B2C2.
In the quadrangle OBA2C the angles B and C are equal to 7i/2, and
so a circle can be circumscribed about it; the segment OA2 becomes a
diameter of that circle and, hence, its centre is the midpoint of segment OA2.
But the circle {OBC) coincides of course with the circle {OBA2C); hence,
the centre of {OBC) is the midpoint A0 of segment OA2.
Let zl9 z2, z3 be the affixes of the points A, B,C. Then the affix a3 of
the midpoint A3 of segment BC is
a y ??J r J s 5
2
and since point A2 is obtained from A3 by inversion with the inversion
circle {ABC), it follows that the affix a2 of point A2 is
_ 1_ 2___
#3 ~Z 2 + Z3
The affix o0 of point A0 will therefore be
1
a0 — —----- ——•
Z2 + Z3
The affix a± of point Al9 symmetric to point A0 about the line BC, is
found from the relation
ao + al
— #3>
whence
1 Zo ZQ Z2 ZZ <*2
— Z2 + Z3~" Z T — — = Z2 + Z 3 -------------
" - — 01~ —Zi=ax—
2 + Z3 Z2 + Z3 z2 + z3 Z2 + Z3
The affixes bx and cL of points BL and Cx are similar in form. Thus,

^2 1 _
Cl = C i------------ (7l y^2 a2t
Z2 + Z3
u<*2 1 —
t>i — G1 — C1 y O’2 b 2,
Z3 + Zi 2
<t2 1 _
Cl Cl — c 2 c2.
Zi + z2 2
From this it follows that A ^ A C i is obtained from A A 2B2C2 via the

similarity transformation
1
U = t T i ----------- (To Z .
2
This is a similarity transformation of the second kind: first symmetry is
executed about the jc-axis (z -* z), then a rotation about the point O through
the angle arg(—g 2) 9 and then a homothetic transformation with centre O
and ratio |tr2|/2 = laJ/ 2 ; finally, a translation defined by the directed line
segment OH (since o1 is the affix of point H).
As a result of these transformations, A A 2B2C2 goes into A îîC l5
------- N
which is similar to A A 2B2C2 but has opposite orientation.
Furthermore, since under the transformation
1
U = 17, ----------- Go Z
2 “
the centre O of the circle {ABC) that is tangent to the sides of A A 2B2C2
goes into the orthocentre H of A ABC, it follows that the circle (5), into
which the circle (O) passes, will touch the sides of A A ^ C x (into which
A A2B2C2 passes), and point H will be the centre of (S).
Under the succession of transformations resulting in the transformation
1
U = (J1 ----------- Go Z,
the radius of {ABC) changes only under the homothetic transformation

{09 /ffil/2); consequently, the radius of (5) is equal to
2 2
since the radius R of {ABC) is equal to 1.
Problem 21. Let AXA2AZA4 be an arbitrary quadrangle inscribed in
a circle {O) and let P be an arbitrary point. Denote by P12, P1Z, Pu , P23,
P 24, P 34 the points symmetric to point P about the straight lines AXA2,
A xAZ9 A xAa, A2A39 A2A49 A zA4. Let R 9 S9T be the respective midpoints
of segments P12, P ^ 9P139 P24, P14, P2Z9 and let O' be a point symmetric
to the centre O of (O) about the centroid of the system of four points
Al9 A29 AZ9 A4 (to which are assigned equal masses; Fig. 24).
Prove that the points R 9 S 9T, O' lie on one straight line.
Solution. Assume {O) to be the unit circle. Let zl9 z2, z3, z4 be the
respective affixes of the points Al9 A29 Az, A4. Denote by g19 g 29 g Z9 g 4 the
basic symmetric polynomials of the (complex) numbers z i9 i = 1,2, 3,4:
— Z1 + z2 + Z3 + Z4>
C 2 = Z X Z 2 ~f- Z j Z z + zl + ^2 Z Z + Z 2 Z4 + Z3 Z4>
— Z2 ZZ Z4 + Z \ Z3 Z4 + Z 1 Z 2 Z 4 4 ~ Z 1 Z 2 Z 3>
<74 = Zi Z2 Z4.
The affix g of the centroid G of the system of points Al9 A2, A3, A4 is
equal to a j 4, hence, the affix o' of point O', which is symmetric to point O
about point G, is equal to
o’ = ffJ2.
The equation of the straight line ALA2 is

z + zl z2 z = z1 + z2. (54)
The equation of the perpendicular dropped from point P to the line AXA2
is of the form
z - p = z1 z2(z - p)
or
z - z1z2z = p - zxz2p , (55)
where p is the affix of point P. Adding termwise equations (54) and (55),
we find the affix z = of the projection P& of point P on the line AXA2:
P lZ = (^ 1 + ^ 2 + P ~ Z l ^2 P)-
The affix p12 of point P12, which is symmetric to point P about line AXA29
is found from the relation
P + Pi 2
= Pl2,
whence
Pi 2 = 2P% — p = z1 + z 2 — z1z2p . (56)
Of similar form is the affix p34 of point ^345 which is symmetric to point P
about line A3A4:
P m = *3 + z4 - z3 z4p . (57)
From (56) and (57) we find the affix r of the midpoint R of segment ^12^34:
r = — k i - (z 4z2 + z3 z4)p]. (58)
The slope of the line O'R is
—r -----^ f a - (z1 z2 + z3 z4) p ]

2 2 p
x = = — <r4>
P
2 2 2 L *3*4/ J
and therefore the equation of O'R is
z7
2
This equation may be rewritten thus:
P rr
— •——
p
<J4
(*-a
2pz — 2p o4~z + <r3p — pa1 = 0 . (59)
Since this equation involves only symmetric polynomials of the affixes
zl9 z2, z3, z4 of points Al9 A2, A3, A4, the equations of the straight lines
O'S and O'T will be the same as equation (59), that is, the points O', S, T, R
lie on one straight line. Incidentally, we can see directly that the affixes s
and t of points S and T,
s = ~ [ ° i — (* 1 *3 + *2 *4) P ]
2
1= — “ (zi z4 + Z2za) P]
satisfy the equation (59).

Remark. Since equation (59) passes into an equivalent equation if p
is replaced by Xp, where X is an arbitrary real number, it follows that
the straight line (59) does not change if point P describes a straight
line passing through the centre O of circle (O) (provided, of course, the
point O itself is excluded from the line).
Problem 22. Let A1,B 1,C 1 be projections of the vertices A ,B ,C of
A ABC on the diameter <5 of the circle {ABC), Denote by A2,B 2, C2 the
points symmetric to points Al9 Bx, C1 respectively with respect to the mid
perpendiculars of the sides BC ,C A ,A B of A ABC, Let A39 B3, C3 be
the midpoints of segments BC, CA, AB, and let A 4, I k C\ be the midpoints
of segments A2A3, B2B3, C2C3, Prove that A ABC and A ^ A C 4 are similar
and have opposite orientations. Prove that if the diameter <5of (ABC) = (O)
rotates about the point 0 , then the centre S of the circle (S) = (AaB^Ca}
describes a circle (0) that is concentric with (ABC), and the rotations
are in opposite directions. The radius of (S) is equal to — OH, where H
4
is the orthocentre of A ABC. The radius of (0) is equal to R/4, where R
is the radius of (ABC) (Fig. 25).
Solution. Take (O) = (ABC) to be the unit circle, and let the diameter &
be the x-axis. Let zl9 z2, z3 be the affixes of the points A, B, C. The affix aL
of the projection Ax of point A on the straight line <5 is
Since line BC has a slope of —z2z3, it follows that the equation of the
straight line OA3 is of the form
z z2z3 z
and the equation of the perpendicular dropped from point Ax to line OA3 is.
From the system of equations

z z2z3 z — 0 ,
(60)
we find the affix z = a* of the projection A f of point Ax on the midperpen

dicular OA3 of segment BC:
The affix a2of point A2is found from

cii + a2
whence
a2 — 2a* — ax
2 2
_ 5 L ± Ii z r
The affix aAof midpoint A 4 of segment

is
#2 + #3
# 4 = ----------------- 9
2
where a3is the affix of the midpoint A3
Fig. 25 of segment BC, that is,
a3 ± i *.
2
To summarize:
“I ('•+ £z7 + Z! + z‘)= H"* +S)'

O f similar form are the affixes bt and c4 of points Bt and C4. We have
a\
l + 4 i ,’
/ a3 , (72 (61)
04 = ---- 1----- 9
4 4za
c4 — 0----
*3 . ^2
r •
4 4 z3
From this it follows that the points At, Bt, C4 are obtained from the
points A, B, C via the similarity transformation
(7o (To —
U= — + — Z
4 4
of the second kind, and, hence, A ABC and A A A Q have opposite orien
tations (and are similar).
From the relations (61) it also follows that the points A4, B4, C4 lie on
the circle (5) with centre S whose affix s is
s= < tJ4
and the radius of the circle (S) is equal to

1*2 1_ l*ll _ 1
P
4 4 4
The distance d from the centre S of (/f4#4C4) to the centre O of
(O) = (ABC) is
I - *
d = ksl (R = 1).
4 4 “ 4
This means that when the diameter 8 is rotated about the point O, the
centre S of (/f4Z?4C4) describes a circle (f2) whose radius is equal to
— = Rj4, where R is the radius of (ABC). If the diameter S is turned
4
through an angle a, then the affix of its end, which had an affix equal
to 1, will have an affix P = cos a + / sin a; and if this point with affix p
is taken as the new unit point, then the new affixes of the points A, B ,C
2 Z Z
will be — » — » — j and the new affix of the new centre S* of the new
P P P
triangle A fB fC f will be <r3//?3. This means the affix of point S* in the
initial system will be <r3//J2. From this it follows that the radius OS of
the circle (Q) will rotate in a direction opposite that of the diameter 8,
and the angular velocity of rotation of OS will be twice the angular
velocity of rotation of 8.
Remark. Let us also find the ratio
(a 4b 4c 4)
(ABC)
We have
*3 , *2 *3
•°
—2 z-, i1 °± 7l i
4~+ 4z1 7 4 z4 4
*3 , *2 *3 I *2 *2
(a 4b 4c 4) = - -- z 2
4 4 *~4z2 4 + T4 Za 1 7 4 z2 4
*3 , *2 *3 ^2
1
4 T 4z3 4
J i *1 1
i O2 O2 k 2|2 (ABC) = kil - (ABC) = OH2
Z2 22 1 (ABC)
4 16 16 16 ~r<r
z 3 Z3 1
whence (6 2 )
(A4B4C4) OH2
(ABC) ~ 16
------------—
From this it also follows that A^B^C* if ABC.

Problem 23. Given a triangle T such that there exists a straight line t
intersecting its sides at angles equal to the angles of the triangle, that is *,
(AB, t) - (AC, AB) = A,
(BC, t) = (BA, BC) = B, (63)
(CA, t) = (CB, CA) = C.
1°. Find the angles of the triangle T.
2°. Through the points A, B ,C draw straight lines parallel to the line r.
Denote by Ax, Cx, Bx, respectively, the points of intersection of these lines
with the straight lines BC, CA, AB. Prove that A ACB and A B xCiAl9
are similar where
A <--> Bx, C <-> Ci, B <—
►Ai,
and that they have opposite orientations.
3°. A transversal xx is constructed for A B XC1A1 in the same way as
the transversal t was constructed for A ABC. Let the straight lines passing
through the points Al9 Bx, Cx respectively and parallel to the straight line xx
intersect the lines BXCX, CXAX, AXBX respectively, in the points A2, B2, C2.
Prove that A ABC and A C2B2A2 are homothetic and have the same
orientation, that is, prove that the straight lines AC2, BB2, CA2 pass through
the same point S and AB\\ C2B2, BC\ \ A2B2, CA 11 A2C2.
4°. Prove that the diameter of the circle (ABC) passing through the
point S is perpendicular to the straight line t (Fig. 26).
Solution. 1°. Suppose that C < B < A. By the Chasles theorem we find
B = (BA, BC) = (BA, r) + (t, BC) = A - B,
whence
A = 2B.
Furthermore,
C = (CB, CA) = (CB, t) + (t, CA) = B - C
* The symbol (p , q) is used to denote the oriented angle from straight line p to
straight line q {p and q lie on an oriented plane). If a is one o f the values of the angle (p , q),
then all values of this angle are given by the formula
(A q) = « + kn,
where k assumes all integral values; stated differently,
(p} q) = a (mod n).
The Chasles theorem for three straight lines p ,q ,r lying on an oriented plane holds:
(A q) + (A r) = ( a r) (mod n).
The relation (63) is to be understood as follows: one of the values (CA, r) is equal
to C, one of the values (AB, BC) is equal to B, and so on.
whence
B= 2C .
Thus
A = 2B = 4C
and so
n 4tt
C = —> A=
7 7
2°. Assume {ABC) to be the unit circle, and A the unit point. The ver
tices A, By C of A ABC will be the vertices of a regular heptagon
AC'CB'A'QB inscribed in the circle (ABC). Setting a = cos (2n/l) +
+ i sin (27t/7) we find the affixes a, c \ c, b'> a', co, b of the vertices of this
heptagon:
a — 1, c = a (= a1), c = a2, 6' = a3, a' = a4, cu == a5, b = a®.
The quadrangle is a rhombus because we can take the straight
line for the straight line t (then all of the relations (63) of the statement
of the problem will be fulfilled) and, hence, by construction we have
t - AAX||CC', CB\\AC', C'C = C'A. The quadrangle CA,BB1 is also
a rhombus since its opposite sides are parallel: A'B | | t || CB1,A B B 1 | | A'C
and A'C = A'B. Similarly, we see that the quadrangle B'BCXA is a rhombus
as well.
From this we can find the affixes of the points A ^ C c , since
A^C + A ^ A = A^C',
it follows that
c — ax + a — ax = c' — al9
1 1 -8 1 0
whence
al — c + a —- c' = a2 + I or.
Furthermore, from the relation
A'C \ A B =- A'BX
we find
c — a’ + b — a’ = bx — a
and, consequently.
bx — c + b — a! = a2 + a6 or.
Finally, from the relation
¥ a +]F b = B T ,
we find
a —b' + b — bf — c, ~ b \
c1 —- a + b — // — 1 + ar>— a3.
To summarize:
= 1 — a + a2,
bx = a2 — a4 + ot6,
— 1 — a3 -f- a6.
->
The triangles ACB and are similar and have opposite orientations
if and only if the determinant
a bx 1
c cx 1
b 1
is equal to zero. We have
i a2 —a4 -|- a6 1 1 a2 —a4 + a8 1
A = a2 1 —a3 + a6 1 — a5 1 —a3 + a* 1
a8 1 — a + a2 1 a 1- a + a 2 1
---• 1 — a3 + a* + a6 — a® + a7 + a3 — a* + a7 — a -f a4 a7 - 1
+ a — nr® — nr7a7 + ot9 - a11 = 0.
Now we find the centre of similarity* of A ACB and AîCVî. Let us

consider the similarity transformation of the second kind under which
* The centre o f similarity of two mirror-similar triangles (that is similar triangles with
opposite orientations) is understood to be the fixed point o f a similarity transformation
of the second kind that carries one o f these triangles into the other.
the points Ax and B5 go into points B and A (then point C, goes into C).
Let z be the affi x of an arbitrary point M of the plane, and let M \u) be
its image under the indicated similarity transformation. Then
z u 1
bx 1 1 = 0
ax cl 1
or
u(ax — bx) + z( I — a) + abx — ax = 0
or
n(l — a -f- a4 — aB) + z(l — a) + a3 — a5 + a7 — 1 + a — a2 = 0,
whence, cancelling 1 — a, we have
u( a5 + a4 + I) + z + a(a3 + a2 + 1) = 0.
The fixed point of the similarity transformation satisfies the condition
z(a5 + a4 + I) + z + a4 -f- a3 + a = 0, (64)
whence (passing over to conjugate numbers) we have
z(a5 + a4 + 1) + z + a4 -1- a3 + a = 0
or, multiplying the left-hand side by a7 = 1,
z(a2 + a3 + 1) + z + a3 + a4 + a6 --- 0.
From the last relation we find
z = —z(a3 f a3 + 1) — a3 — a4 — a6
—z(a2 r a3 + 1) (a5 + a4 -[- 1) — (a5 + a4 + 1) (a3 + a4 + a®)
+ z + a3 + a4 + a = 0
or
z( I — a7 — a6 — a2 — a8 — a7 — a3 — a5 — a4 — 1)
= a8 + a9 + a11 -f a7 + a8 -f- a10 + a3 + a4 + a6 — a3 — a4 — a
or
z(—a® -- a5 — a4 — a3 — a2 — a — 2) = a6 + a4 + a3 + a2 +■ a + 1.
But a7 == 1, and since a — 1 ^ 0, it follows that
a6 + a5 -f- a4 -f a3 + a2 + a + 1 = 0
and the last relation takes the form
whence
z = a5.
That is, the fixed point of the~similarity transformation of the second kind
--------> ------ >
that carries A BXCXA^ into A ABC is the point Q.
Setting
a5 + a4 + 1 =- A, a3 + a2 + 1 r~ p,
we find the relation
Xu + z + olh = 0 (65)
that relates the affix z of the arbitrary point M of the plane with the affix
u of its image M ' under the similarity transformation at hand. The points
Bx and Ax go into points A and B under this similarity transformation, so
that
bx + <xfx +afi,
a= ,----- 9 b ... ------------------
X
whence
bx
a ~ b = <*i
X
and, thus,
A\B\
■= W-
AB
However,
a5 + a4 + l = a 2 + a 3 + l.
a5 + a4 + 1 -- a2 + a3 + 1;
consequently,
|A|2 = (a5 + a4 + 1) (a2 + a3 + 1)
~ a7 + a8 + a5 + a6 + a7 + a4 + a2 + a3 + 1
= 2 + a6 + a6 + a4 + a3 + a2 + a + 1 = 2,
whence
A\BX_Y2 9 __
AB ~ ’ A1B1 ~~ ](2
---------------------------- *
is the proportionality factor that carries A B ^ A i into A ACB; it is,
equal to 1//2.
3°. Let Ox be the centre of the circle (AXBXCX). The point 0 1 is the image
of point O under the indicated similarity transformation (item 2°). There
fore, the affix ox of point Ox is found from equation (65) by setting u = 0
in the equation:
ox ~ — a(a3 -f a2 + 1).
Let us consider a regular heptagon inscribed in a circle ( A ^ C j) with
the vertices arranged in the following order:
A qCqQqA qBqBqCo,
where A0 = Bx, B0 Cx, C0 = Ax. Such a regular heptagon exists since
the angles of A BXAXCX are
and, hence, AXBXis one of the sides of the heptagon and Cx is its third ver
tex. The affixes of the vertices of the heptagon are
Oi + (a0 — °i) 1 (k = 1, 2, 3, 4, 5, 6, 7).
Expanded, they are:
a0 = bx = a2 — a4 + a8,
c0 = = a2 — a + L
ct)0 = — a4 — a3 — a + (a2 + a6 + a3 + a) a2
— —a4 — a3 — a + a4 + a8 + a6 + a* = oc\
a'0 = — a4 — a3 — a + (a2 + a8 + a3 + a) a3
= — a4 — a3 — a + a5 + a9 + a* + <*4 — a8+ a5 —a3 + a*— a,
bo = — a4 — a3 — a + (a2 -f a6 + a3 + a) a4
= — a4 — a3 — a + a6 + a10 -f a7 + a6= a® + a5 — a4 — a -f- 1,
*o = = a6 — a3 + 1,
ci =-- — a4 — a3 — a + (a2 + a6 + a3 + a) a6
= — a4 — a3 —■a + a8 + a12 + a9 + a7 = a6 — a4 — a3 + a2 + 1.
From this it follows that Q0— Q and that this point is the centre of the simi
larity transformation under which A C 2A2B2 goes into That
these triangles are similar and have opposite orientations follows from the
-------------------> . ------------------- > .
fact that the construction of A C2A2B2 on the basis of ABiCxAx is exactly

the same as the construction of A B ^ A x on the basis of AACB; the
figures under consideration with two circles circumscribed about A ABC

and l\A 1B1Cl and all the corresponding triangles are samilar.
Furthermore, since CqA zA qBqis a rhombus, it follows that
Cq “ Qq b0,
whence
— Qq 4~ Co feo
- oc2 - a 4 + a6 + a 2 — a + 1 - a* - a5 + a4 + a - 1 = 2<x* - a5.
Again, BzA0CqB0 is a rhombus and so
a0 bz ~ Cq
whence
b2 = a$ -(- b0 Cq
= a 2 — a4 + a6 + a8 — a3 + 1 — a5 + a4 + a* — a 2 — 1 = 2a6 — a5.
Finally, since CzC0AqB0 is a rhombus, it follows that
c0 cz = Gq b09
when ce
CZ — Cq -f* 6 q Qq
= a2 — a + 1 + a8 — a3 + 1 — a6 — a5 + a* — a2 + a = 2 — a5.
Thus,
a2 = 2a2 — a5,
b2 = 2a6 - a5,
c2 = 2 — a5
and so A C2B2A2 is obtained from the triangle ABC (the affixes of whose
vertices are 1, a6, a2) via a homothetic transformation with centre Q and
ratio 2 since
c2 — 0) 2 - a5 - a5
2
z,,
a — 0) 1 - a5
b2 - - 0) 2a® — a5 - a5
b - 0) ~ a®— a5
a2 ~- (0 2a2 - 2a5 - 2
--
c — 0) a2 — a5
4°. The slope of the straight line OQ is equal to
(X5
— = a'° = a*.
a5
The straight line t passes through the points A and B' and so its slope is
1 - a3 1 - a3
a3.
1 - a3 ~ i __ 1
a3
The sum of the slopes of the straight lines OQ and x is 0 and so OQ _L x.
Problem 24. 1°. On the sides of the hexagon AyA2A3A4AbA6are construct
ed equilateral triangles
---------->.---- :----- > ----------> —--------> ----------> ----------►
AyA2Al9 A2A3A29 A3A4A39 A4AbA49 AbAeAb, AÂyAB (66)
which have the same orientation. Prove that the ends P[9 P3, Pb of the
directed line segments OP[9 OP3, OPb (O is an arbitrary point), which are
------ -------- >.------
respectively equipollent to the directed line segments A[A49 A3Ab, AbA2y
form an equilateral triangle T = P[P3Pb9 which has an orientation oppo
site that of any one of the triangles (66). In particular, A T can degenerate
into a point.
2°. Prove the converse: namely, that if an arbitrary equilateral triangle
T — P 1P3P 5 and a point O are chosen in a plane and directed line segments
AyA4f A3A3, AbA2 are constructed thatjare respectively equipollent to the

directed line segments OP[9 OP3', OPb (the position of the directed line
------- > .------- > -------->
segments A[A49 A3A£, AbA2 are otherwise arbitrary), then, by choosing
another arbitrary point Al9 it is possible to construct the points A2, A3,
A49 Ab, A6 such that the hexagon A[A2A3A4AbAb is obtained from the
hexagon A1A2A3A4AbA6 by the construction indicated in item 1°. Prove
that if the point Ay is changed in the plane, then the principal diagonals
AyA4, A3A69 AbA 2 of the hexagon A lA2A3A4A bA6 (which will change to
gether with any change in point Ay) will rotate about three points Ol9 0 2, 0 3.
How are these points constructed?
3°. The equilateral triangles
A ;A 'A ;',A 'A ;A '', A'3A ’4A'3'9 A 'A 'A 4'9 A'bA'A'b\ A ,bA[A,bf (67)
are constructed so that they all have the same orientation and the orien
tation of any one of them is opposite that of any one of the triangles (66).
Prove that
------ > ------ > ------ > ------> ------> ------ >
A\Ay = A2A39 A2A2 = A3A4, A3A3 = A4Ab9
----- > ------> ------> ------ > ------> ------>
A4A4 = A bA69 AbAb = A6Al9 A€A6' = AyA2
where == is the sign of equipollency of (directed) line segments.
4°. Prove that
A['A'a' = A W = A W = 3OG,
where G is the centroid of the triangle T* = M14A/36M52 and M liy Af36, A/52
are the ends of the directed line segments OMUy OM36, OM52 laid off
from an arbitrary point O and respectively equipollent to the directed line
------> ------> ----- >
segments AÂA9 AÂ^ AÂ^.
5°. Prove that the midpoint of the principal diagonal AXA4 of the hexagon
A ^ A ^ A ^ A q coincides with the point/C14, in which segments A['A'2' and
AZ'A$' intersect and are bisected; the midpoint of A3A6 coincides with the
point K3S, in which segments A z’ 'A a and A's A[' intersect and are bisected;
finally, the midpoint of A 5A2 coincides with the point A^52, in which seg
ments A Â ^ and A2A'^ intersect and are bisected (Fig. 27).
Solution. We introduce in the plane a rectangular Cartesian system of
coordinates Oxy. Denote by a and a the imaginary roots of the equation
jc3 + 1 = 0 :
l+ijfa - 1-/K3
a
2 ’ “ 2
Let alt a2, a3)a4, a5, a6, p[, p'3, p3, a[, a2, a3, a[, a3, a«, be the respective
affixes of the points A u As, A3, At, A5, A„, P[, P3, P3, A[, A3, A3, A4, A3, A3.
Then
«2 — fli = <x(a[ - at),
o3 a3 = oc(n2 @z)y
o3 — oi(a3 a3).
as - a 4 = ct(a4 — at),
a6 - a 3 = a(as — a5),
— a6 = <x(a3 - a6).
From this we find (note that since a2 — a + 1 = 0 it follows that a — 1 +
+ a = 0, whence 1 — a = a)
a. = aa[ + aa4,
a3 — aa'3 + a a2,
a4 = aa3 + a a3,
(68)
o5 = aa4 + aa4,
= afls + a^s.
a4 = cta'i -f aa8.
Fig. 27
Multiplying these relations by a5, a4, a3, a2, a, 1 respectively and adding,
we obtain (note that aa = 1, a3 = a3 = —1)
a'2 — ab = a (a'6 — aJ) + a (a'A— flO-
Since
a’4 — a[ = p[, ai — a3 = Pz9 of2 ~~ afb = p b,
it follows that
pi = 05P* +
or
pi = <*/>3 + (1 - a) = />J
or
/>5 ~ P'l = a (pZ - p[). (69)
From this it follows that AP[PZPZ (provided it does not degenerate into
a point) is equilateral because from relation (69) it follows that the directed
line segment P[P$ is obtained by a rotation of the directed line segment
P[PZ through the angle tt/3 (a = (1 + i J/3)/2 cos (tt/3 + / sin (it/3)).
------>
The orientation of A P1P3P5 is opposite that of any one of the triangles
of (66): this follows from relation (69) and, for example, from the relation
a-i — at = a (a[ — ax);
since aa = 1, it follows that
a [ - a x = a (a2 — a t),
that is, the directed line segment AXA[ is obtained by a rotation of the
directed line segment AXA2 through the angle —tt/3 (a = cos (—n/3) +
+ 1 sin (-tt/3)).
The case p[ -- pZ is also possible; then also p[ = pZ, that is, A P[PZPZ
contracts to a point. This occurs if and only if the directed principal di
agonals A[Ai AZAZ, AZAZ of the hexagon A[AZAZAZAZAZ are equipollent:
a [ a '4 - azaz = azaz.

2°. Let P[PZP* be an arbitrary equilateral triangle and O an arbitrary
--------^ > ------- >
point in the plane; let A[A4 AZAZ, AZAZ be arbitrary directed line segments
,
respectively equipollent to the line segments OP[, OPZ, OPZ:
OP[ - A[Ai OPZ = A fzA l OPZ = AZAZ (70)

Choose an arbitrary point Ax in the plane. Construct the following equi
lateral triangles with the same orientation:
------y ► » y > >
AXA2AX, A2A3A2, A3A4A3, A4Ar>A4f AÂÂ^, AqAxA$
but with orientation opposite that of A P{PZPZ- On the basis of item 1°,
I with the aid of the construction given in item 1°, the hexagon AxA2A3A4AhA^
leads to the chosen points A[, AZ, AZ, A\, AZ, AZ and then to the chosen
triangle P[P’zP i
3°. The affixes a[, aZ,aZ,aZ, aZ,aZof the points A[, AZ, AZ, AZ, AZ, AZ
are connected with the affixes a", a2 , a2 , a4", aZf, aZ' of the points A x , A2 ,
A 3', A'a A'b\ Ab by relations of the form (68) in which the positions of
only a and a need be interchanged:
a2 — aa[f + cta[9
a'z = aa2 + aa2,
a\ = aaz + ctaZ9
ab == 0W4' + aaA,
= otab -f- 010r5,
a[ = aê' -|- a^g*
From these relations and from the relations (68) we find
ata" = a2 — owj = a3a — 02& + <*i&= afl3 — ota2 + &Q\
and so
o-y ~ Q\ q 2 -|-
Similarly
a2 ~ a2 “F ^ 4 ?
# 3 ~ # 3 # 4 “ t“ ^ 5 >
(71)
# 4 * = £ 4 + ^ 6 ’
tffi* “ a% “t~al>
a* = a< — + a2.
From these relations it follows that
Oy Q\ ~ qz a2,
Cl2 Q2 = Q\ (2
Gz Qz ~ Qb 0 A,
aA aA= a6 ^5,
£6* #5 = ^1 ^65
#6 = @2 ^1»
and therefore
A-yA" = A2A39 ^ 2^2 = ^3^49 AZA3 = A4A5,
A4A4 == AbAe, AbAb = AgAj, A qA q — AyA2.

4°. From the relations (71) it also follows that
ai' — a" = ai' — 03' = ^2' — ^ 5' = ^4 — — <*1+ — #3- (72)
Consequently,
A Â 4 = A Â i' = a7 a ;\
If we construct the directed line segments

------- > y y y y ------- >.
OMu = AjAf, OAfgg = AgAg, OAf$.> = A$A2,
then the affixes m14, m36, m52 of the points M u , Mw, Afsa will be
and so the affix g of the centroid of A will be

g = 0*4 — Oi + </6 — + a, - «s)/3.
From this and from the relations (72) it follows that
A['Al' s A'3'a \' = A'Â” = 3OcT

5°. The midpoints of segments AtA4 have the same affix
Qi + a«
2
[see formulas (71)] and, hence, these midpoints coincide with the point Kl4.
The remaining propositions in item 5° can be proved in similar fashion.
Problem 25. Let Av Blt Ct be the orthogonal projections of the point P
------►
on the sides BC, CA, AB of triangle ABC. Let us construct A CXBXQX
------>
similar to A BCP but with opposite orientation. Let A', B \ C be the
second points of intersection of the straight lines PA, PB, PC with the
circle (ABC) = (O).
1°. Prove that the triangles AXCXQX and CAP are similar and have op
posite orientations.
2°. Prove that the triangles BXAXQX and ABP are similar and have op
posite orientations.
3°. Prove that the triangles AXBXCX and A'B'Cf are similar and have
the same orientation. Find the proportionality factor.
4°. Prove that the point Qx is a point obtained by inversion with the
circle of inversion (.AXBXCx) (or by symmetry with respect to the straight
line AXBXCXfor the case when the points Ax, Bx, Cx lie on one straight line)
of the midpoint T, of line segment PP*9 where P* is the image of point P
under inversion with the circle of inversion (ABC).
5°. Find the affix of the fixed point of the similarity transformation
------- > ------- >
that carries one of the triangles AXBXCXand A'B'C' into the other.
6°. For what position of point P does it coincide with point Qx (Fig. 28)?
Fig. 28
Solution. 1°. Take (O) — (ABC) for the unit circle. Let z„ zt, z3 be the
respective affixes of the points A, B, C. The equation of the straight line
BC is of the form
Z — z3 = —z., z3 (z — zs)
or
z + z2z3z = z2 + z3. (73)
The equation of the perpendicular dropped from point P to line BC is
z - p = z.,z3(z - p )
or
z- Z2Z3 Z = p — z2z3p, (74)
where p is the affix of point P.
Adding termwise the equations (73) and (74), we find the affix z = ax
of point A1:
ax = —- (z2 + z3 + p — z2z3p).
Similarly, we can find the affixes and c, of points Bx and C ,:

t 1 ,
b i y (Z3 + Zi + P - Z3Z ]/7),
Cl = - J - (Z i + z2 + p - z , z 2p ) .
2
Since it is given that A BCP and A C1B1Q1 are similar but have opposite
orientations, it follows that the affix of point Q1 can be found from the
condition
b cx 1
c b2 1 = 0
P <h 1
or
------ - - (z1 + z 2 + p - zxz2p) 1
2
— ~ (Z3 + zl + P - Z3Zxp) 1
z3 2
~P <h 1
or
1
—
Z1 + Z2 P — z& p 1
^2
1
Z3 + Z1 + P ~~ z 3zlP 1
*3
P 1
Subtract the second row from the first row of the determinant to get
- — — z2 — z3 + (z3 — z2) pzl 0

Z2Z3
— z3 + Z i + p — z 3Z j p 1
Z3
p 2 1
Cancelling z3 — z2 # 0, we get
Multiplying the left-hand side by z2z3, we obtain

-s + Zl + p — z3z,p - 2q1 — (z, - ztzzp) (z,p - I) - 0
or
Z3 + Z 1 + P — z%Z\P — 2<7i — z2ztp -f r, — a3p %— z.tz3p -= 0
whence
9i y ~ P + P + * i)-
2°. The symmetry of the right-hand side with respect to z l 9 z3, z3 permits
asserting that (item 1°) A A y C 1Q l and A CAP are similar and have oppo
site orientations and that (item 2°) A P y Ay Qy and A A B P are similar and
have opposite orientations. At the same time we found the affix of
point Qy.
3°. Associated with the directed line segment Aybx is the complex
number
Let us find the affixes a \ b \ c' of the points A \ B \ C . The equation of

the straight line PA is of the form
P — z\
zZi — —----- ~ ( z - Zj).
P - *i
Solving this equation together with the equation zz = 1 of the unit circle
(ABC), we obtain
P - Z 1 Z — Zy
P - Zi ZZy
One of the roots of this equation is naturally z --- Zy (the affix of point A);
the other root (the affix a' of point A') and be found from the equation
P — Zy 1
P — Zy ZZy
whence
1 P - Zy
z --= a
Zi P ~~ Zi
and b' and c' have similar expressions. Thus,

l p - z 1
a = -------- : ---- 9
Zi ~ p - Zl
b, = _ _1. P - z2 ^
Z2 p — ~Z2
C' = - i m .
Z3 P — Z3
To the directed line segment ^f'5' there corresponds the complex number
b, _ , = J _ P - Z i _ _ L P - Z2
zi P — z i z2p z2
= z*(pP ~ ~ z iP + z i* a ) ~ zi(PP ~ ZiP ~ Z*P + z i z 2)
ZjZ2(p - z ,) (p — z 2)
_ z2pp — ZiZ2p — p + Zi — Z|PP + p + ztz2p — Zjj
ZiZ2(p - Zj) (p - z 2)
_ pp(z2 — zt) — (z2 — zt) __ (z2 - Zt)(pp - 1)
ZiZ2(p — Zl) (P — Z2) ZlZ2(P - Zl) (P - z 2)
b’ - a' = __________ 2 ( p p - 1)__________
*1 - flj ZjZ2(p - Zj) (p - z 2) (z3p - 1)
2(pp — 1) _ 2(pp-l) .
<r3(P - Zi) (p — z 2) (p - z 3) a3(p 3 — ^ p 2 + d2p - d3) ’
J _ O’2 - Gl
and since a, = — t a2 = — 5 it follows that
CO
eo
b’ - a ' 2 (pp - 1)
(76)
— fli <r3P3 - <r2P2 + —1
Denoting the right-hand side by A, we have
t 2 (p p -l)
<r3P3 — <r2p 2 + atp — 1
and then obtain
b’ - a' = A(b! - a!>
and, similarly (since A is a symmetric function of zl9 z2, z3),

c' - b' - A(cx - bx)9
a' — c' = Afa — cx).
--------> -------->.
Consequently, A A'B'C' and A A1B1C1 are similar and have the same
orientation.
On the basis of (75), the proportionality factor may be written as
2\OP2 - R 2\ 2\OP2 — /?2|
~ Ip - z x\ |p - z t \ |P - z t | “ PA-PB-PC
4°. Let us consider the similarity transformation that carries A A,B,C,

into A A’B'C’. Under this transformation, the centre Ox of the circle
(A& C ,) goes into the centre O’ = O of the circle (A'B'C'), and the point
H„ the point of intersection of the altitudes of A A,B,C„ goes into H',
the point of intersection of the altitudes of A A'B'C'. Since the indicated
similarity transformation is of the form
z' = Xz, + p
and since under this transformation the directed line segment OxA, goes
into the directed line segment O'A' = OA', it follows that
a’ = X(a, — ox). (77)
where a', au o1 are the respective affixes of the points A', A„ Ox. Simi
larly,
b' = Mb, -
(78)
C' = MCl — O,),}
where b', c', b„ c, are the respective affixes of the points B', C', B„ C,
From (77) and (78) we find
o' + b' + c' — Mai + b, + cx — SOi).
Now, using the earlier obtained expressions for a', b', c', we have
o’ + b ' + c ' = JL P ~ J ± _ J _ P ~
Z\ P Z1 Z2 P Z2
___1_ P - Z3 _L x (79)
Z3 P Z3 a3 Y
where
X = z 2z3(p - z,) (p - z 2) (p - z 3)
+ Z-MP - z2) (P ~ z 3) (P + Zi) + zxz2(p - z3) (p - z,) (p - z 2),
Y — (p — z x) (p — z 2) (p - z 3).
12-810
'178 Problems in Geometry
Furthermore,
z 2z 3( p - z 2) ( p ~ Z 3) ( p - rO = z , z 3[ p a — (z2 + z 3) p + Z , z 3] ( p - z ,)
= \ z 3z 3 p l - ( z 2 + z ;.) > + 1] (/> - Z i)
--== Z j Z 3p - p - (z , + z 3) p p + p - ffj/;2 + (z tz2 + Z iZ j) p - zt.
The other two terms in the numerator of (79) are of the form
Z3 Z1 ~P%
P - (z3 + Z2) pp + p - o3pa + ( z 2z 3 + ZjZj) p - z 2,
zLz2p ap — (Zj + z2)pp + p — o3p a + (ZjZj + ZjZ2) p - Zj.
Adding the last three expressions, we obtain
X = pp2<r9 — 2(7!/?/? + 3/? — 3(7.,/?2 — 2(T2p —
To summarize:
, . ,, , , -pp-O i + 2ffipp - 3p -f 3<r3pa - 2<s2p -f Ox
a + b + c —------------------- -3--------------------- -------------
o*(P - * i) (/> - - 2) iP - ^ 3)
Using the expressions for au bu cx obtained above, we get
1
«i + ^ + c2 = ' (2(71 + 3/? - cr2/?);
2 {p p - 1)
and since A — it follows that the relation
°*(P - z'i) (P - z%) (P - ^ 3)
a' + bf + c' — A (^! + + cx — 36?x)

or
— 3*! — --- (a! + bf + c ) — (î + bx + Ci)
A
may be rewritten thus:
- to , = + 2°,PP ~ » + 3» ^ * ~ + ». —
PP - 1
3/>V + 3o3p3 — 3<r2p 4- 3*7,
pp - 1
hence,
P*P — o3 p a - a ? p4 — Oy
O, —
2(pp - 1)
From this we obtain
o 3p 2 — <r2P + P + < ri p %p — o 3p 2 4- o ,p — <Tj
9i - Ox
2(PP - 1)
O a P P * — O i P P %Jr P i P + Q i P P — Q ^ + Q i P — p — O x - I ^ P + o ^ - o -iP + O x
2(PP - 1)
a3pp3 — a2pp2 -rOxPP —p _ o3p(p3 — <bP2 + aiP — a3)

2(PP - I) 2(pp ~ 1)
<r3p(p — Zj) (p - z 2) (p - z 3)
2(pp —1)
The affix t of the midpoint T of segment PP* is
PP + 1
We have
iK ) 2P
pp + 1 P2P — <
t3p 2 + <r2p — o',
t - o 1=
2p 2(pp - 1)
p2p 2 — 1 —p2p 2 + a3p3 — <r2p 2 + ffjp _ a3p 3 — a2p 2 + o,p — 1
2P (PP - 1 ) Tpipp - J)
<r3(P ~ Z j) (P ~ z't) 12 !p — z3\2 _
(?i - <h) (tj - o,) =
4(pp - I)2
But this means that the points T and Qx are obtained one from the other
by inversion with the circle of inversion with centre Ox and radius
|p — Z ,| Ip — Z 2 1Ip — z 3| PA-PB-PC
(R = 1).
2 \pp — 11 2 | OP2 - R2 1
The radius of the circle (AXBXCX) — (Ox) can be found from the simi
larity relation z' = Az1 = p that carries the circle (AXBXC}) into the circle
(A' B' C'). Since the radius of (A'B'C') is R — 1, it follows that the radius
Rx of (AXBXCX) is
R R ! o3(P — Zi) (p — z >) (P — z3) _ PA-PB-PC _
1 |A| | 2(Pp - 1) ~ 2 \OP2 - R2\ P'
5°. The affix co of the fixed point Q of the similarity transformation that
carries one of the triangles, , into the other, A' B' C', is found from
the reasoning that under this transformation A OxAx Q goes into

A OA'Q so that
0 1
a f ax 1 = 0,
CO C0 1
whence
o,a
co —
Oi + a - ax
Furthermore,
Zl ~ P Z2 — Z ^ + p — Z2ZzP
a - ax = ------- --------------------------- ------
ZiP - 1 2
2zj — 2p — zvztp — zxz%p — zxpp + a3p 2 + z3 + z3 + p — ztz3p

2{z{p - 1)
Z\ —P I- ffi — (Tjp — ZiPP + <r»p*
2(zlP - 1)
p*p - g8p a + a2p - ffi zt — p + - <T2p - ztpp + <raP2
<>i+a'— ax =
2(pp — 1) 2fop - 1)
: ---------— — ------------------ — (P2P % - a3p % + a & p * - axzxp - pip

2(p p - 1) (zlP - 1)
+ <r3p s - atp -f a l + p p z l — p 2p + PP<J\ — p 2p<r2 — Zippp*
+ <r»PP* — z1 + p — a1 + pa 2 + Zxpp — a3p2)
2pp(zx - p) - (zL - p) + gaP2(î ~ P) ~ OiP(zx - p) -<x3p3(zt - p )
2(PP - 1) (ZiP - 1)
(zi ~ p)(2pp - 1 + g2P2 - ffiP - JaP8)
2(PP - l)(ziP - 1 )
and, consequently,
P*P — o'sP2 + <r2p — ffi Zi — P
_________2(pp - 1)______ z,p — 1___
(zi —p) (2pp — 1 + ff2P2 — ffiP —ff3P3)
2(PP - 1)(zjP - 1)
_______P2P — g3P2 + ff2p ~ gi______
2(pp — 1) - 0 3P3 - <t2P2 + crtp - 1)
But
P2P — G-iP1 + <*lP — <*1 = 2ox(pp — 1),
<r3P* — <*iP' 4 - <?iP — 1 = 2p(pp - !)(/ — o,)
and so
20i(/>P - 1) 0i
0) — ----- =------------- ———---------------- “ ------------------ .
2(pp - 1) — 2p(pp — 1)(/ - o,) 1 - p(t - 0t)
6°. The points P and Qt coincide if and only if p = qlt that is,
P = - (0-jP* — OtP + P + ffi)

2
or
0 , p 2 — ff2Jp —P + 0i — 0. (80)
From this.
a*p%— — p — 0! •= 0
or
— P*- =0.
03 0g 0S
that is,
p* — 0tp + 08
0,
and (80) takes the form
— (p* + alp* + 0 ! — 20, p* + 2/>*02 — 20j02p)

^3
— 1 (0 8p2 — 0 10 a/> -h 0 !) —p + 0 i = 0
<*3
or
p* — 2 0 jp* + ( 0 ? -I- 0 2 )p 2 — ( 0 X0 2 + 0 3) p + 0 1 0 3 = 0 . (81)
It is clear that p = is a root of this equation.

Dividing the polynomial in the left-hand side of (81) by p — <rl9 we
obtain p 2 — Oi/?2 + <x2p — <73 = 0 or (p — z j (/? — z2) (/> — z3) = 0.
Thus, the remaining roots of equaton (81) are
P — z 1, p = z2, p = z3.
Assuming that P does not coincide with aoy of the vertices of the given
triangle, we conclude that the only possibility is p = ov Thus, the points P
J 82 Problems in Geometry
and Q 1 coincide if and only if P is the orthocentre of the given triangle

ABC.
Problem 26. Take an arbitrary point P on a circle (ABC) = (O). Let A*
be the point symmetric to point A with respect to the straight line OP, and
let A' be the point symmetric to P with respect to the straight line OA*.
Construct points B' and C' in similar fashion. Denote by A", B",C,r
points respectively symmetric to the points A', B', C’ with respect to
lines jBC, CA, AB.
1°. Prove that the triangles ABC and A"B"C" are similar and have
the same orientation.
2°. Prove that A ABC is inscribed in A A " B " C namely, that the
triplets of points A , B”, C"; B, C", A"; C, A", B" lie on one straight line.
3°. Prove that the orthocentre H of A ABC is the centre of the circle
(A"B"C").
4°. Find the radius R" of (j4"2P'C") and indicate a method for construct
ing it if we know the position of the unit point on the unit circle (ABC)
and the position of point P on that circle.
5°. What is the maximum value of R"1 What position of point P is
associated with this maximum value of P"?
6°. For what positions of the point P do all the points A", B ", C"
coincide with the point //?
7°. For how many and what positions of point Pare the triangles A ”B"C”
and ABC congruent?
8°. Find the affix of the centre of the similarity transformation of
A ABC and A A nB " C \ assuming (ABC) to be the unit circle and know
ing the affixes a, b, c, p of the points A, B ,C , P.
9°. Prove that if the point P describes the unit circle (ABC), then the
centre Q' of the similarity transformation of A ABC and A A ,,B ,,Cn de
scribes the orthocentroidal circle of A ABC, that is, the circle (GH) con
structed on the line segment GH as a diameter (H is the orthocentre of
A ABC, G is its centroid, hence the term orthocentroidal).
10°. For how many and what positions of the point P do the points G
and Q' coincide?
11°. For how many and what positions of the point P do the points H
and Q' coincide?
12°. What line does the centroid G" of A A"B"C'f describe if point P
describes the circle (ABC)1.
13°. Prove that the affix of the centre Q' of the similarity transformation
of A ABC and A A"B"Cff is connected with the affix of the centroid G"
of A A"B"G” by a linear fractional relation, which, consequently, deter
mines a certain circular transformation of the plane. Prove that this trans
formation leaves in place the orthocentroidal circle (GH) of the triangle
ABC. Find the fixed points of this transformation.
14°. Prove that the centroid G" of A A nB nCn, which centroid cor
responds to the centre Q[ of similarity of A ABC and A A"B/fC", is the
second point of intersection with the circle (GH) of the straight line 0 }Q*,
where Ol is a point symmetric to the point O about the midpoint K of
segment GH, and Q* is a point symmetric to the point Q' about the mid
perpendicular of segment GH.
Solution. 1°. We take {ABC) for the unit circle, and Boutain point
of A ABC for the unit point. Let a, b, c, p, a', b', c , a", b", c" be the re
spective affixes of the points A ,B ,C ,P , A f,B f, C , A ",B ",C ". Then
~ abc — 1. The equation of the straight line passing through point A
and perpendicular to the straight line OP is of the form
z — a — —pl (z — a).
Solving this equation together with the equation of the unit circle zz 1,
we get
2 z —a
z — a = p * --------
az
One of the roots of this equation, z = a, is the affix of point A, the other
root is
z = a* ----- p2ja.
which is the affix of point A*.
The slope of the straight line perpendicular to the straight line OA* is
z - p = x(z - p ), P‘
a a
The equation of the straight line passing through point P perpendicularly
to OA* is of the form
P4
* ~ P = ------ (* - P l
a2
we find the affix a' of point A':
One of the roots of this equation is naturally z = p (the affix of point P),
the other is the^ affix of the point A \ that is, a! — p2/a2. In similar fashion
we find the affixes of the points B' and C . We thus have
a’ PL
c2 '
The equation of BC is
z + bcz = b + c.
The equation of the perpendicular dropped from point A ' to the straight
line BC is
or
Adding these last equations and the equation of BC termwise, we find

the affix z = a'l of the projection of point A' on line BC:
We find the affix a" of point A " from the relation
That is,
whence (since a3 = 1)
a" = b + c — a = a1 — a( 1 + p 3).
n3
P6
In similar fashion we find b" and c". Thus,
a" = ffj — (1 +p~*)a,
b" = <rx - (1 +p~3)b, (82)
c" = 0 i — (1 +P~3)c-
And so A A"B"C" is obtained from A ABC by a similarity transformation
of the first kind:
2 " = (Tx - (1 + p - * ) Z . (83)
Hence the oriented triangles ABC and A"B"C" are similar and have the
same orientation.
2°. We now prove that A ABC is inscribed in A A"B"C ". We will
prove that the straight line B"C " passes through point A. To do this it
suffices, for example, to prove that
a — b‘
a — c"
is a real number. Taking advantage of formulas (82), we find

a — b" _ a — + b{\ + p~9) _ bp 9 — c _
a — c" a — + c(l + p 9) cp~3 — b
We have
1 3 1
p3 ------
b c cp9 — b bp 3 — c
u = —-------- — = ------------ — — u.
1 3 __ 1 bp9 — c cp 3 — b
c b
In a similar manner, proof is given that line C 'A " passes through point B
and line A"B" passes through point C.
3°. The image of point O under the similarity transformation (83) is
a point with affix aly that is, the orthocentre H of A ABC. Under the simi
larity transformation (83), A ABC goes into A A"B''C” and so the centre
0 of {ABC) goes into the centre of (A''B''C"), that is, the point H.
4°. The radius R" of (A"B"C,f) is equal to
| 1 + p~3| = '| 1 + p3| and, hence, may be con
structed as follows (Fig.29): draw a tangent line to
the circle (O) at the point P and through the unit
point Q (a Boutain point); draw the straight line
1 parallel to that tangent line; the second point of
intersection, point Q, of the line / with the unit
circle will have an affix p2. Through point Q draw
a straight linew parallel to line PQ, the second
point of intersection, point T, of line m and the
Fig. 29
unit circle will have the affix p3. Construct a rhom
bus with sides OQ and OT. Its diagonal OS — |1 + p3| = R ", where R" is
the radius of (A"B"C").
5°. Let p = cos (p + / sin <p, then 1 + p3 = 1 + cos 3and,
hence,
R" = | 1 + p 31= /( I + cos 3(p)2 + sin2<p = 2 | cos (3<p/2)!.
From this it follows that R" is a maximum [twice the radius R of {ABC)]
if and only if jcos (3<p/2)| = 1, whence 3<p/2 == ± kn and, consequently,
:
<Pi = 0, <Ps = 2tt/3, <^5 = 4 n/3.
For <?! = 0 we obtain the unit point, that is, the Boutain point 7\ of A ABC;
Ts and T5 are the other two Boutain points [A TxT^Tb is an equilateral
triangle inscribed in the circle {ABC)]. Thus, R" assumes the maximum
value, equal to R" = 2 = 2R if and only if the point P coincides with one
of the three Boutain points of A ABC.
6°. The points A ", B '\ C" coincide with the point H if and only if /?3 +
3
I 1 ~ 0, that is, p — 1. The affixes of these points T2, T4, have the
arguments
(p2 = */3, </>4 == n, <p6 = 5n/3.
These points T2, r 4, Tb together with the Boutain points of A ABC form
a regular hexagon TxT2T3TATbT%inscribed in the circle {ABC). Thus, al!
the points A ”, B", C" coincide if and only if the point P coincides with
one of the three vertices T2, TA, T6 of a regular hexagon inscribed in the
circle (ABC), whose three other vertices are the Boutain points Tl9 T3, Tb
of A ABC.
7°. The triangles ABC and A"B"C” are congruent under six positions
of the point P. Indeed, A ABC and A A"B"C'f are congruent if and only
if | p3 f 11= 1, that is, 2 | cos (3<p/2) | ~ 1, whence
cos (3<p/2) — ±1/2, 3<p/2 == 2kn ± tt/3, 3<p/3 — 2kn ±2n/3;
consequently,
4 , 4
<P - kn ± — n. cp - kn ± ...
3 9 3 9
On the circle (ABC) there are altogether six points with such arguments
and they are obtained by rotations of the radii OTl9 OT3, OTb (Tl9 T3, Th
are the Boutain points of A ABC) through the angles ±40° (rotations of
the radii OTu OT29 OT3 through the angles ±80° lead to the same six
points).
8°. From the relation
z" = <7j — (1 + p " 3)r, (84)
which define a similarity transformation that carries A ABC into
A A "B nC,r9it follows that the affix to' of the fixed point Q' of the similari
ty transformation (84) is
co = - --------
2 + /T 3
[in relation (84), put z” = z \
9°. From the relation obtained in item 8° it follows that if point P
describes a unit circle, then the point Q' describes the ciicle (£') obtained
from the unit circle via the linear fractional transformation
z = — —.
2 + z
We will now prove that (Q') is an orthocentroidal circle of the triangle
ABC, that is, a circle with diameter GH. The affix k of the midpoint K of
segment GH is
Furthermore,
\z ' — k\
- | - " 1 - —(Ti | - | --------i------L-

j 2 Jr z 3| i 3(2 -(- z)
=- i 1j .... .w . izi i2 + z \ = ]?_i[
3 ; 2 -I z j 3 -I2-M 3’
hence,
KQ' r= 0///3 = GHI2.
Consequently, the point describes a circle with diameter GH.
10°. The points Q' and G coincide if and only if
0\
2+P3 ~ 3 ’
whence p3 = 1, that is, the point P coincides with one of the points Tu
Th.
IT. The points Q' and H coincide if an only if
whence p3 = —1, that is, the point P coincides with one of the points
t 4, r 6.
12°. The centroid G" of A A"B/fC'f corresponds to the centroid G of
A ABC under the similarity transformation
* " - * 1 - 0 +/>■*)*•
That is, the affix g" of point G" is
*" = * 1 - 0 +/> *) *3— V3 (2-/»-*)•
From this we have

2cr,
i* " - * i ■= i— (2-/»-•)
-3 \ _
3 3
That is, if the point P describes the unit circle, then the point G" describes
the orthocentroidal circle (GH) of A ABC.
13°. Eliminating p from the relations
(O 8" = y (2 - P 3)>
2 + />-*”’
we obtain
V'
2+ p 3 —y ,
co
whence
3g" co' — 4co'<71 + a? = 0.
In this relation, ax is fixed and so it expresses g" as a linear fractional func

tion of co' (both points G" and Q with affixes g” and co' lie on a circle with
diameter GH) and therefore this function transforms the orthocentroidal
circle (GH) of A ABC into itself. We find the fixed points of this transfor
mation from the equation 3z2 — 4axz + o\ — 0, whence
2 — CTj, Z — Oi/3
are the affixes of the points H and G.

14°. The affix ox of point GAis
4
- 3-
The slope of line GH is equal to Hence, the equation of the straight

line passing through point Q' parallel to GH is
z — to' — -a_l- (z — co')
or
<*1 - ^1
2 2 = CO — (85)
^1
The equation of the midperpendicular of line segment GH is
2— *1
<*1
or
z + - l- z = — ffj. (86)
trA 3
Adding termwise the equations (85) and (86), we find the affix z = 7r
of the projection of point G' on the midperpendicular of segment GH:
The affix co* of the point Q* can be found from the relation
co' + CO*
----------- —n— <h CO
2
whence
CO S '
01
but
a/ — 01
2 + p 3’
Hence,
to* 01
■3
+ P
We have
4 4 0i
<7i o1 — aU 2 - p 3)
0i to* g" 3 3 2 + p~ 3 3
0i S* g" 4 _ 4 _ 0i <r, «.
(7l o1 — L (2 -/7~3)
1 1 1 3 3 2+/T3 3
1 1 1
4 1
3 2+ p 3
= 0]01 4 1_ -0.
T 2-}V* 3<2 + ' ‘)
1 0 0
Problem 27. Let P, 0 , R be the orthogonal projections of point M
on the sides BC, CA, of AABC; let A0,B 0, C0 be the midpoints of
segments M A, AfP, MC, and let A', B', Cf be the points obtained by inver
sion of the points A0, B0, C0 with the circle of inversion (PQR). Prove
that the ratio of the area of A A'B'C' to the area of A PQR is equal to
the ratio of the square of the radius of (PQR) to the power of the point M
with respect to that circle with sign reversed (Fig. 30).
Solution. Take (PQR) as the unit circle. Let zl9 z2, z3i z0 be the respective
affixes of the points P, Q, R9M . Since the radius of (PQR) is regarded
as equal to 1, it follows that the power of the point M with respect to the
circle (PQR) is equal to a — z0 z 0 — 1. We thus have to prove that

(A'B'C) _ 1
(PQR) ~ a
The equation of the straight line MQ is of the form
*0 ^2 / -v
z — z 0 == - - (z — Z0)
Z« -
or
( z n — z-i) z — ( z 0 z — z 0z2 — z 0 Z j .
— z 2)
Z0 "
The slope of the straight line AQ is equal to — —— — - and therefore the
z0 —z%
equation of this line is of the form
(z 0 — z 2) z + (z0 — z2) z — z2 z 0 + z 2z0 — 2. (87)

Similarly, we can obtain the equation of the straight line AR:
(z 0 — z 3) z -f- (z0 — z3) z = z3 z 0 + z 3 z0 — 2 . (8 8 )
Solving the system of equations (87) and (8 8 ), we find the affix z = a of
point A :
j z 2 z 0 + 7 2z0 — 2 z0 — z, |
a A i z3 z 0 z :iz0 2 z0 z3 1
A I Zq — z2 z0 z2 j
1^0 ^3 -o I
We have
A' = (z2z 0 + z 2z0 — 2) (z0 — z3) — (z3"z0 + z3z0 — 2) (z0 — z2)
” z 2Z qZ q -j~ z 2Z q 2z0 z 2z 3Z q z 3z 2Z q -f- 2z3
z3ZoZ0 z 3Zq -f- 2z0 -f- z2z3z 0 ~h z3z2Zq 2z2
=-- (z2 - z3)z0z 0 + 4 ( ' --------1 ) + z« ( Z‘“------------ — 2(z2 — r3)

\ z* z3 / V Z3 z2 /
(■
Z? . Z2 + z 3
= (z2 — z3) Zo'z'o --------------h - ... — Z„
Z2Z3 Z2Z3 - 2)
zqZqZ2z3 — Zq 4" (z2 ~h ^3 ) 2z2z3
= (z2 - z3)
^2^3
A = ( z 0 — z 2) (z0 - z3) - (z0 - z2) (z 0 - z 3)
= ZqZq ZqZ3 z 2z0 + ZjiZa z3z 0 -|- z0z 3 -)- z2z 3 z2^ 3
_|_ -z.3 ____ Z 2

= z 0(z2 - z3) + z„ ( 1 ------- M
V z3 z2 / Z2 *3
Z0 + Z2Z3Z0 Z2
= (z2 - z3)
Z2ZS
Thus
— z | + (z2z3z 0 + z2 + z3) z0 — 2z2z3
Zq + Z 2Z 3 Z 0 z 2 z3
The affix a0 of the midpoint A0 of segment AM is

Z0 __ Z0 ~\~Z 2Z 3Z 0 Z 0 Z 2Z 0 Z 3Z 0 ~ f'Z 2Z3Z0 Z 0 ~ h ^ 0 ^ 2 ~f~^0^3 2 z 2Z3
_
2(z0 + z2z3z 0 - *2 - 2s)
_ ___ Z2Z3( z 0Zq 1)
* 0 "t" Z2Z?,Zt) Z2 Z3
The affix a' of the point A' is equal to l/a 0, that is,
1_
z 0 H-----— zo ---- -- -
a = Z2Z3 Z2 Z3 Z q - f - Z 2Z3Z q Z2 Z3
1 Z0 Z — 1
(ZflZo - 1)
Z9Z0
--- (z0 + Z0Zj<X3 + Zx — <Ti).

a
We now have
o' = - 1- (z 0 + Z^Zydz + z \ — IT,).
a
Similarly,
b' = — (z0 + z0z2ff3 + z2 - ^l),

(7
C' = - (z0 + Z o Z 30>3 + Z3 ffl)»

(7
b' = — ( z 0 + ZoZg^g + z 2 —
<7
C' — ---- (Z0 + Z0Z3(73 + Z3 — 0^)

G
and, hence,
a' I z o “f“ z o z i ° ,3 ' i ~ z i — ° i Zo+ZqZ^j + z x &1 1
(A 'B C ) = b’ b' 1 Zo+ ZoZ2<734-z2—ax z 0+ z 0z8ff3+ z'a 1
4a2
c' c' 1 Z0~KZ0Z3°’3"l'Z3—^1 z 0+ZoZ3ff3+ z 3—a t 1
. zo*>3+zi ZqZ ^ + Zj I ; IZ| z l li Zi z \ 1 ■
/
^ 2 ^0Z2°3 i~Z2 Z0Z2^3 "hz 2 ^ ; Z0 Z» j Z 2 Z2 1, z 2 Z2 1
4(J2
! Z0Z3^3H~Z3 Z0Z3^3~KZ3 ^ I ! z 3 Z3 lj z3 z 3 1 .
= - 1 (PQR)
G
whence
(A'B'C') _ 1
(PQR) “ (7 ’
Problem 28. AXA2A3 is an arbitrary triangle; P is an arbitrary point.
Through P draw two mutually perpendicular straight lines S and 8'.
Let these lines intersect side A2A3 at the points B2 and B3, respectively,
and the altitude (dropped from point Ar on side A2A3) at points B2 and B3.
In similar fashion construct the points C3 and Cl9 C3 and C[ for side A3AX
and the altitude dropped from vertex A2 on that side; also construct the
points Dx and D2y D[ and D2 for side AXA2 and the altitude dropped
from point A3 on side AXA2. Prove that the centroids of the three groups
of points B2y B39 B2, B3\ C3, Cl9 C3', C[; Dl9 D2y D[yD2 lie on a single straight
line (to all indicated twelve points are assigned identical masses, see Fig. 31).
Solution. Take (AXA 2 A3) for the unit circle. Let zl5 z2, z3,/? be the re
spective affixes of the points A l 9 A2, A3, P. Let us write down the equations
of the straight lines 6 and S' in the form
z —p = t(z — p) or z — tz = p — rp, (89)
z — p = — t(z — p) or z + t ’z = p + xp, (90)
where | t | = 1. The equation of the straight line A 2 A3 is of the form
z + z 2z3z = z2 + z 3. (91)
Subtracting (89) from (91) term by term, we obtain the number z = ft2,
which is the conjugate of the affix b2 of point B2:
Z2 + Z 3 — P+'c'P (92)
b2
Z 2 Z3 + T
whence
1 + 1 - p + 1 p
Z2 ^3 T t (~ -2 + Zg z2z3p ) -{- z, z3p
h (93)
Z 2Z 3 + T
— +
Z2 Z3 T
Substituting —t for t in the formulas just obtained, we find the affix b3

of point B 3 and J>3: _
1 — 't(z2 + z3 z2z3p) + z 2 z 3 p
b3 — - 9
Z2ZZ T
_ _ z2 + z 3 — p — zp
bz —
Z2ZZ ^
1 3 -8 1 0
From these formulas we find the number b23, which is the conjugate of
the affix b23 of the midpoint B2Z of segment B2B3:
^ = b 2 + bz = J _ / z 2 + z 3 — p + T p + z 2 + z3 - p - x p \
2 2 V z 2 z3 + t z2z3 — t )
= z 3 z3(z2+ z3) — p z 2 z3 — t2 p
z \z \ - T2
whence
1 / 1 1 \ _ 1 1
-----I ------- 1---- I — p --------- — P _
_ Z2z3 \ Z2 Z3 ) Z2Z3 T2 _ z\ z l p + T2 p z 2 z3—(z2+ z 3) T2
" 1 1 z2 z§ - t2
2 | Z§ T2
Furthermore, the equation of the straight line passing through point Ax

perpendicularly to line A2A3 is of the form
* - *1 = Z2Z3(Z - Z\)
or
z 2 Z3
z — Z2 Z3 Z *1 (94)
Z1
Subtracting (94) from (89) term by term, we find the affix z = bz-> which
is the conjugate of affix b2 of point B2:
Z2 Z3
p - zp - z 1
Z1
b* =
z 2 Z3 T
and, similarly,
p + x p — Zt + —
______________ £i_
b3 = Z2 Z3 + T
The number £ 23*which is the conjugate of the affix b23 of the midpoint £23
of segment B2B3y is
p - z p - z± Z2 Z3
p + tp — Zi + *2*3
Z1
Z2 Z3 T *2 *3 + T
72 72
z2 z3p — a3 .-----------
* 2 *3 o —
x2p
Z% Z% — T2
whence
l _ _1_ Z2Z ,
2 ^3 o -- «
— p J i— i , pz\ Z%— Zx T2 -f T2 — Z 2Z 3 p t 3
z 2 Z3 °3 z|z§ T2
1 z| z| — T2
-2 -2
Z2 Z3
The affix A of the centroid of the set of points B2, 2?3, 2?3 is equal to
the affix of the midpoint of segment B2ZB23, the ends B23 and B23 of which
are the midpoints of segments B2B3 and B2B3:
^ _ ^23 + ^23
2z| z\ p
■ (-¥ )
2(z| Z§ — T2)
(95)
Let us now prove that the point with affix A lies on the straight line A
given by the equation*
2z^2p + ^ - dx j + 2z^2p + — <rx j
" ( 2' , + ^ ) ( 2 ? + ^ ) _ ',‘ ” ‘ <96)

We have
z \z \ t2 V z 2 z3/ _ (Ti z \z \ — a3 — 2p t 2
A = (97)
2(z| z| — T2)
if-1 -- 1 )
Ul 4 t2y
* The equation of this straight line could have been set up by similarly determining
the affix /i of the centroid o f the second set o f points C3, Clf C3, C\ :
_ 2 2
2z3 z ip
2(zi zf - t2)
also, by determining the affix v o f the set o f points D l t D2tD i9 Z>2*itis possible to verify
the validity of the necessary and sufficient condition o f collinearity o f the three points
A A 1
P P 1 = 0.
V v 1
196 Prooiems in Geometry
Let us compute the left-hand side of (96), setting z = A, z = A (the de

nominator z\ z\ — x2is dropped for the time being):
^ 2p z\ z| — t 2ax + x2
*3 )
+ ^ 4 z l ^ - o s - 2 p x2j ^2p + z— - <7! j
= 4p p z \ z\ + 2p z\ z\ — — 2p — zf zf — 2p x2o’! — Oi <r3 + 2t 2
t2 cr3 ff3
+ 2p x%-2— + ^ - °j. x2^ + 2/> z2 z§ ** + t 2 - I z2z§

Zi -^1 ^3 Z2 (J3 <r3
— 2| zi — 2p a3 — t 2 + (T3 o’! — 4p px2 — 2p - — f- 2p x2 ax

0*3 ®*3
= 4/> p(z! Z§ — T2) + zf z\ T2 (zi Z§ — T2)

^3
+
^ (z| zi - T2) + (Z2 Z2 - T2)*
t2 o3
Thus, the left-hand side is, for z = A, equal to
<,p- + l - ^ + 2 / |+ ^ .
<73 T2 <73
The right-hand side of equation (96) is
x2 a3 cr3
which is the same as the left-hand side. The symmetry of equation (96)
proves the theorem. At the same time, (96) is the equation of straight
line on which lie three centroids of three sets of points:
B2, B3, B2, B3\ C3, Ci, C3, Cii Di, D2, Di, D>.
From this theorem there follows a corollary (the Droz-Farny theorem)
(see Fig. 32): if through the orthocentre H of triangle AXA2A3 we draw
two mutually perpendicular straight lines, which cut off on sides A2A3,
A3Ax, A xA2 segments B2B3, C3CX, DXD2, then the midpoints of segments
B2B3, C3Ci , DxD2 lie on one straight line. Of course, this theorem can be
proved in a straightforward fashion, and its proof is technically not so
elegant as that given in the problem generalizing the Droz-Farny theorem.
In a direct proof of the Droz-Farny theorem, one should bear in mind
that Ci is the affix of the orthocentre H of the triangle A i A2A3.
It is obvious that the Droz-Farny line is obtained from the straight

line A given in the generalization (for the same case P = H) under the
horn othe tic tran sfor mat ion
('■ i) : the points B2, B3 merge with the
point 77, and the centroid of the set of points B2, Bz, B2, B3 is the midpoint
of segment HB23, where B23 is the midpoint of segment B2B3. And similarly
for the points C3', C[ and D[, D2.
Problem 29. Let 7 be the centre of a circle (/) inscribed in A ABC ;
let 7), F, F be points of tangency of (7) with the sides BC, C/4, AB, respec
tively; let H be the orthocentre of ADEF; let <5 be some diameter of (7);
let D \ E \ F ' be points symmetric to the corresponding points 7), E, F
with respect to the straight line <5. Prove that if Pa9Pb9Pc are points sym
metric to an arbitrary point P with respect to the sides F 'F ', F'D \ D'E'
of A 7)'F'F' and a, /?, y are points symmetric to the points Pa9 Pb9 Pc respec
tively with respect to the sides BC, CA, AB of A ABC, then the centre Q
of the circle (a/fy) has the property that the Simson line of point Q with
respect to AD EF is parallel to the straight line <5. Prove that if we construct
a directed line segment QN equivalent to a directed line segment HP*j2,
where P* is a point symmetric to point P with respect to the straight line 5,
then the radius of (a/ty) will be equal to 2IN.
Solution. Take (7) = (DEF) for the unit circle and take the diameter 5
for the real axis. Let zv z2, z3, p be the affixes of the points Dy E, F, P.
Then the affixes b and c of points B and C are
2 z 3 Zj 2 z1 z2
> c—
Z3 4 ~ Z1 Z1 Z2
and their conjugates are

2 _ 2
~b > c .
Z3 + Z1 Z1 + Z -1
Let us find the affix pa of point Ptt. The affixes of the points D', E’, F'
are ~zlt z 2, z 3. The slope of the straight line E'F' is
_L_ J.
Z2 Z 3 __ z2 z3 _ 1
Z2 z3 z2 z3 z2 z3
The equation of the straight line is
Z - Z2 = — (Z - Z2)
”2 Z3
or
Z + Z 2 Z 3 Z — Z j - |- Z 3. (98)
The equation of the perpendicular dropped from point P to this line is
z-p — < J ~ P )
%2^3
or
z z 2z3 z — p z2 z 3p. (99)
From equations (98) and (99) we find the affix p* of the projection of
point P on the straight line E 'F ' :
P* = ~r (*a + z 3 + P - *2 *3 P)-
2
The affix pa of point Pa, which is symmetric to point P about the straight
line E'F', can be found from the relation
P +„Po- = P* = — (z2 + z 3 - z 2 z3p + p),

z z
whence
Pa — Zs Z2 Z3 p.
The slope of the straight line ID is z1f z 1 = zf, and the slope of BC
is equal to —zf. The equation of BC is
Z — ZX = - zf(z — z 2)
or
z + zf z = 2zx. (100
The equation of the perpendicular dropped from point Pa on the side BC is
Z - P a = zl& -P a )
or
2 — z\ Z = pa — z\ p a. (101)
From equations (100) and (101) we find the affix p , of the projection of
point Pa on line BC:
P- = \ (p‘ ~ z' p ‘ + 2z^

and from the relation
X \ Pa ~ Pa ~ “ (Pa A p a + 2ZX)
2 2
we find the affix X of point a :
X = 2z1 — z?(z2 + z3 — z2z3p) = 2zj — zx(zx z2 + z1zs — a3p)
= 2zj — Zy{a2 — z2z3 — <r3p) = <r3 + zx(2 — <r2 + p<r3).
The affixes p and v of points P and y have similar expressions:
p = <xs + z2(2 — cr2 + pff3),
V = ff3 + z3(2 — <r2 + p<r3).
Thus we see that the points a, /?, y lie on a circle, the affix of whose centre
is equal to a3, and since |er3| = 1, it follows that this centre lies on the
circle (DEF). The radius of the circle (a/ty) is
P — |2 — o3 po31— |2 — c2 + p<r3| 2 —— + —
^3 O’3
|2ct3 — + p|
= |2<r3 - ox + p | + -d -O i)
k 3l
The fact that the Simson line corresponding to the point with affix a3
is parallel to the real axis follows from problem 3.
The affix p* of point P*, which is symmetric to point P about the straight
line <5, is equal to p* = p. To the directed line segment HP* there corre
sponds a complex number p* — ax — p — av *
The directed line segment HP* 12 = QN is associated with the complex
number (p — <x1)/2 = n — a3y where n is the affix of point N. From this
we have n — < j 3 + (p — Ci)/2. Since n is the affix of point N and (/) is
taken as the unit circle, it follows that
IN = |n| + — (P - Oi)
2
* Let the points A and B have affixes a and b, respectively. We will say that the
directed line segment AB is associated with the complex number b — a.
Fig. 33
Problem 30. On the circle (O) take four arbitrary points Al9 A2, A39 A4.
Take any one of six segments joining these points in pairs, for example,
segment AXA2. Construct an isosceles triangle A1A2A12 with vertex A 12
and base AXA2 so that its centroid coincides with the point O. Let A[2
be a point symmetric to point A12 about the straight line A3A4. Prove
that the six points thus obtained, A[29 A[39 A[49 A23) A24, A349 lie on one
circle, (S), that is concentric with (O); note that the radius of (S) is
four times the distance from point O to the centroid G of the system of
four points Al9 A2, A3, A4 (Fig. 33).
Solution. Take (O) for the unit circle and let zl9 z2, z3, z4 be the respective
affixes of points Al9 A29 A39 A4. The affix a12 of the midpoint of segment
AXA2 is a12 = (z4 + z2)/2; hence, the affix a12 of vertex A12 of the isosceles
triangle AXA2A12 (AxAl2 = A2A12)9the centroid of which coincides with O, is
ai2 — (zi ^2)-

The equation of the straight line A3A4 is
z + z3z4~z = z3 -[- z4. (102)
The equation of the perpendicular dropped from the point A12 to the
straight line A3A4 is
z + z x + z 2 = z3 z4(z + z x + Zo)
or
z — z3z4 z = — z1 — z2 + z 3z4 z 1 + z 3 z4 z 2. (103)
Adding equations (102) and (103) termwise, we find the affix z = a*2 of
the projection of point A12 on A3A4:
a*2 = — (z3 -j- z4 z4 z2 + z3 z4 z x -f- z3z4 Zg).
The affix a[2 of point A[2 is found from the relation

012 + 012 *
or*012
~ Z1 ~~ Z2 + 012
Y (Z3 + Z4 “ Zl ~ Z2 + Z3 Z4 Z4 + Z3 Z4 Z2),
2
whence
012 ~ Z3 + Z4 + Z 3 Z4 Z 1 + Z 3 Z4 2 2 == Z 3 Z a { 2 1 + Z 2 + Z3 + 2 4) = Z 3 Z A &1*)
where ax = zx + z2 + z3 + z4. From this relation it follows that

OA'12 = \a’12\ = Io-jI (|z3| = |z4| = 1, |ctx| = |<tx|).
Similarly,
OA[3 = OA[4 = 0 ^ 3 = Oa ; 4 = OA34 = laj.
Thus, the points ^ 2, d[3, A[4, A23, A24, A34 lie on one circle, whose radius
is equal to p = |o-J = 4|o’1/4|. But gJ4 is the affix of the centroid of the
system of four points Al9 A2, A3, A4; hence, \gJ4\ = 0(7. Thus,
p = 40G.
Problem 31. Let D, E, F be points of tangency of the sides FC, CA, ^42?
of AABC with an inscribed ciicle. Consider an arbitrary diameter S of
the circle (I) = (DEE). Draw through points 2)*, F*, F*, in which the
altitudes of A DEF intersect the circle (DEE), straight lines parallel to
the straight line <5, and let A \ B \ C be points at which these lines intersect
the circle (/) a second time. Denote by A *, 2?*, C* points symmetric to
the points A \ B \ C", respectively, with respect to the sides BC, CA, AB
of A ABC. Prove that the centroids G and G* of A D F F and AA*B*C*
are symmetric about the orthopole a> of the straight line 8 with respect
to A DEF (Fig. 34).
Solution. Take {DEF) for the unit circle, and let the diameter 8 be the
real axis Ox (point O coincides with point I). Denote by zl9 z2, z3 the
respective affixes of the points D, F, F. The equation of the straight line
passing through point D perpendicularly to line F F is
Z — Zi = z2z3(z — zx).
Fig. 34
we find the affix z — d* of the point D*. Indeed,
Z~ Z,. = Z2Z3 ^ ~ ~ y j
or
Z* Z*
Z Zi — ----------- ( Z - Z i ) .
zx z
Naturally, one of the roots of this equation is z == zx (the affix of point Z)),
the other root is
Zi
which is the affix of point D*. Similarly we find the affixes
e* _ _ El E1 } f * = —
Z2
of points E* and F*.
The equation of the straight line passing, for example, through the
point D* parallel to line <5 is of the form
Zj_ Z2 Zg
Solving this equation jointly with the equation of the unit circle zz = 1,
we find the affix of point A':
z_ H,------
Z 2 Z3
= — 1+ , -----
Z1
,
Zl 2 Z2 Z3
z±z + z 2z3 = zxz + z 2z3
Z\ Z2 Z3 Z
One of the roots of this equation, z = — z2z j z l9 is the affix d* of point Z>*,
the other root is equal to the affix a' of point A':
a' =
z 2 Z3
Similarly,
Z3 Z1 Z1 Z2
where b’ and c' are the affixes of points B' and C" respectively.
The slope of ID is equal to z1j z 1 = z\; hence, the equation of BC (the
slope of which is — z\) is
2 — Zi = - z\{~Z — Zj)
or
z + zI z = 2z2. (104)
The equation of the perpendicular dropped from point A ' to line BC is
_ Z1 2 ( ; Z 2 Z3
Z --------- = Zl \ Z ---------
z2 z3 V z1
or
z — z\ z = — ---- c3. (105)'

Z 2 Z3
From (104) and (105) we find the affix a of the projection of point A'
on line BC:
“ = - ( 2Z1 + - - ^ 3 )•
2 V z2z3 /
The affix a* of point v4* is found from the relation
a' + a*
= a
2
or
whence
0 * = 2z1 — cr3
and, similarly,
b* = 2z2 — <r3, c* = 2z3— <r3,
where fe* and c* are the affixes of the points B* and C* respectively.
From the relations obtained it follows that
d* —cr3 __ 2zx — 2<t3
Z1 03 Z1
and, similarly, that
b* — <t3 = 2 c* — <r3 = 2
Z2 <X3 Z3 <J3
That is, the point P with affix = cr3 is the centre of the homothetic trans
formation under which AD EF goes into A A*B*C* (see Fig. 34).
Furthermore, the affixes g and g* of points G and G* respectively are
The midpoint Q of segment GG* has the affix co = (a1 — cr3)/2 and,
hence, coincides with the orthopole of the straight line 6 (the real axis)
[formula (11), problem 7, z0 = 0, x = 1].
Remark. Let us find the affix of the point symmetric to point A' about/)/:
the equation of the perpendicular dropped from point A' to D l is of
the form
or
z + Z1 Z — — -— (- <t3.
The equation of D l is
z — z \ z = 0.
From this we find the affix z = X of the projection of point A' on line DI:
The affix p of the point symmetric to point A' about DI is found from
the relation
^ r ~ S- = a
2
or
whence
V= =P-
Thus, all points symmetric to points A', B \ C about the straight lines
DI, El, FI coincide with the same point P, which is the centre of the homo-
thetic transformation r = (P, 2) that carries A DEF into AA*B*C*.
Problem 32. The circle {DEF) = (I) inscribed in A ABC is taken as
the unit circle; D9E, F are the respective points of tangency of the straight
lines BC, CA, AB with the circle; zl5 z2, z3 are the respective affixes of
the points D, F, F. It is required to find:
1°. The affix o of centre O of the circle (O) = (ABC).
2°. The radius R of the circle {ABC).
3°. The radius p of the Euler circle of A ABC.
4°. The affix h of the orthocentre 77 of A ABC.
5°. The affix e of the centre 0 9 of the Euler circle (0 9) of A ABC.
6°. Prove that the circle inscribed in A ABC is tangent to the Euler
circle constructed for that triangle at some point <P0 (called the Feuerbach
point). Find the affix cp0 of the Feuerbach point <P0.
7°. Prove that the three circles {Ia), {Ib), (7C) — they are escribed circles
of A ABC at the angles A, B ,C — also touch the Euler circles of A ABC
at the points <Pl9 <P2, # 3 (also called Feuerbach points). Find the affixes
xa, t ^ Tc °f the centres 7fl, h ,Ic of the circles (7fl), (7ft), (7C). Find the affixes
tl912, t3 of the points Tl9 T2, T3 of tangency of the circles (7a), {Ib)9(7C)
to BC, CA, AB. Find the radii of these circles (Fig. 35). Find the affixes
<Pi><P2 ><?3 of the Feuerbach points &l9 <P2, # 3.
Solution. 1°. The affix of the midpoint of segment EF is equal to
(z2 + z3)/2 and since point A is obtained from this midpoint by inversion
with respect to the unit circle (7), it follows that the affix a of point A is
2
a = -3----- — .
*2 + ^3
Similarly, we can find the affixes b and c of points B and C :

t 2 2
b = ----------- , c = —----- 3—.
z3 + z\ ZL + z 2
The affix o of the centre O of the circle (O) = {ABC) is found from the
system of equations
(o — a) (o — a) = (o — b)(b — b),
(0 — b) (d — b) = (o — c) (b — c)
or
— o(a — b) — o(a — b) = bb — aa, (106)
— o(b — ~c)— o(b — c) = cc — bb. (107)
We have
_ v 2 2 2(z1 - z2)
a — b = --------- = -----------—--------
[z2 -\r z3 z3 + zx (z2 + z3) (z3 -f- zx)
a —b — — 2zaZl = 2z« (za ~ zi)

Zj 4" Z3 Zj Zx (Zj + Z3) (z3 + Zj)
bb — aa
z3 4z.z. zx(z| + 2 z2 z3 + z|) — z2(zf + 2 z3 zx + zf)
= 4z«
(“3 + ^ l ) 2 (Z2 + Z 3) 2 " (Z 3 + ^ l)2 ( z 2 + Z3) 2
Zi zf + Zxzf — Z2zf — Z2 zf 4 z 3 [Zj z2(z2 — Zj) — zf (z2 — Zj)]

= 4za
(Z3 + Zj)'“ ( Z 2 + Z3 ) 2 (Z 3 + (z2 + z3)2
Z j)2
= 4z3(z2 — z2) (zt z2 — zf) .

(z3 + Z j ) 2 (z2 + z3)2
The equation (106) takes the form
2(z2 — zt) 2 z 3 (z 2 — Zj) 4z3(z2 - z i ) ( z 1z2-z § )
—o
(Z2 + Z3) (z3 + Zl) (Z2 + Z3>(Z3 + Zi) (z2 + Z3)2 (z3 + Zj)2
or
2 z3(zj z2 — zf)
O — Z§ 0 = (108)
(z2 + z3) (z3 + Zj)
Similarly, equation (107) is transformed as follows:
2zi(z8 z3 - zf)
0 — zf 0 = (109)
(z3 + Z l) (zx + z2)
Subtracting (108) from (109) term by term, we find the complex number 0,
which is the conjugate of the affix 0 of point O:
(zl - zf) 0 = 2Zl(z2 Za ~ zt) _ 2 z3(z! z2 - zf)
3 (Zl + z3) (Zi + z2) (z2 + z3) (z3 + zx)
= (z2 + z3) (2<r3 — 2zf) — (zt + z2) (2<r3 — 2zf)
(Z2 + Z3) (Z8 + Zi) (Zj + z2)
_ 2<t3(z3 — Zj) + 2z2(zf — zf) + 2 z3 zx(zf I— zf)
5
0102 03
whence
/ , x — 2(T3 + 2z2(z3 + z 3 z 1 + zf) + 2 z3 z 1( z 3 + zx)
( z 3 + Z j) o = -----------------------------------------------------------------------------------
01 0*2 (T3
4<t3 + 2z2z\ + 2z2 zf + 2zt zf + 2z3 zf
01 (T2 <r3
Zi z2 Z3 + Zi z2 z3 + z2 zf + z2zf + z x zf + z3 zf
= 2
01 0*2 03
Z2 Z3(z3 + Zj) + z3 zt(z3 + zx) + Z y Z 2( Z y + z3)
= 2
01 02 03
Consequently,
2<7o
o=
G 1 G2 (73
and from this we have
2 5.
2d2 2(Ti (73
O = _________ __
o-! (73 (72 1 (7i (72 (73
r2
°3 U’a
2°. jR2 = \o — a\2 = (o — a) ( 0 — a) = do + ad — do — oa
4(7! (72 (73 4(7! (73 1 4(7o 1

+
(CTj (72— (73) 2 ( z 2 + Z 3) ( Z 2 + Z 3) (7i (72 (73 Z2 + Z 3 (7j (72— (73 Z 2 + Z 3
i*2 Zo2 Z
^3
< (7i (7o Z2 Zo (7o
+ •
4 (cra (J2- <73) 2 (z 2 + Z 3)2 (<7X <72— ff 3) ( z 2 + z 3) (Z is + Z s X ffiffg — ff3)
_ 01 02 0 .3 _j_ Z 2 Z 3( f f 1 0 2 0 3) ( z 2 "j- Z3) <7j 03 Z2 Z 3( z 2 - f - Z3) <72
(ff! <72 - 03)2 (z2 + 23)2 Ol ^2 - 03)
CTi<72 ff3 z 2z 3( z 2 + z 3) ( z 3 + z 1) ( z 1 + z 2) — ( Z a + Z a ) ^ ; , — z 2z 3( z 2 + z 3)(T2
(01 <r2 - 03)2 (z2 + z3)2 (<7! <t2 — 03)
01 02 03 z2 z3(zt + Z3) (zx -(- Z2) 0i <73 z2 Z3 02

K <72 - <73) 2 0?2 + Zs) (O’! 0-2 “ ff3)
G\ @2 G3 Z1 (73 (7! (73 0 1 C72 <73 <7:j ( z 2 Z3)
(0 1 <72 — <73) 2 (Z2 + Z 3) (<7!<72— (73) (<7j 0 2 — 0 3) 2 ( z 2 + Z 3) (<7j <72 0 3)
<7i (72 (73 (73 (75
((7X<72— 03)2 <7i <72 — <73 (<7X(72 —C73)2

And so
Ri = 4ff3 = / 2<73 \2
(t7x <72 - 0 3) 2 V <73 — <7X <72 /
We will now prove that the number

Complex Numbers in Plane Geometry 20£
is real and positive. We have

1
^ 03___________ ^
^3 ^2 1 0 2 &3 O3 01 02
<*3 <*3
Hence, A is a real number. Done differently, we have
A= - _____ ?____ = _____ L _ = — i _ .

c3 - <Ti G2 l - a i— 1 - °1°1 1 - kil2
0 3
We now prove that \c±\ < 1. Indeed, since all angles of AD EF are always
acute, it follows that the orthocentre of ADEF, the affix of which is equal
to cl9 lies inside ADEF and, hence, also inside the circle (DEF). But we
assumed the radius of (DEF) to be 1, and so \cx\ < 1, and, hence, A > 0;
now since R2 = 4A2, it follows that R = 2A, that is,
_ 2o-3 2
R = - - - - - - - -— = — .
C3 Ci C2 1 CiCi
3°. The radius p of the Euler circle of A ABC is

_R 0*3 = ___ 1_____ _ 1
2 <r3 — c2 1 — \ci\2 c’
where c is the power of the orthocentre H' of AD EF with respect to
the circle (DEF).
4°. Since the sum of the directed line segments
OA + OB + O C = OH,
where O is the centre of the circle (O) = (ABC), and H is the orthocentre
of A ABC, it follows that
a — o + b — o + c — o = h — o.
whence
h = a + b + c — 2o,
where h is the affix of H. We have
h z 2 Z3 _|_ zz zi zi z 2 2 c1 c2
2 Z2 + z3 Z 3 + Z1 z l + ^2 02 — 03
= Z 2 Z 3 (Z 3 + Zl) (Z1 + Z 2) + Z3 Zl(Zl + Z 2) ( Z 2 + Z 3 ) + Z l Z 2 (Z 2 + Z 3> ( Z 3 + Zl )
(Z2 + Z») (Z3 + Zl ) ( Zl + Z 2)
1 4 -8 1 0
2<Ti 03 z2z3(4 + <r2) + z3zx(zl + <r2) + z x z2(z| +ff2) 2<rx <r3

01 02 03 (z2 + Za) (z3 + *i) (Zi + z2) 01 02 03
ai ff3 ~r «t| 2ffj ff3 02 0i 03
0! 02 03 01 02 03 01 02 02
Hence,
h = 2
02 01 0*3
01 02 03
5°. The affix e of centre 0 9 of the Euler circle of A ABC is equal to
£ = ( ^ 7 -°)^ sjnce q 9 js the midpoint of segment OH. Thus,
e= 0I 0i 0*3 4~ °*3
0102 03 01 02 03
6°. The equation of the radical axis of the circle (DEF) (zz — 1 = 0)
and of the Euler circle, the equation of which is
(z - e) (z - e) 0I
= 0,
(^3 — CTi GT2)2
is of the form
o
zz — 1 — (z — e) (z — 1) + ^3 = 0
(<r3 -o-jff,)2
or
— 1 + ez + ez — e~e = 0.
(<73 - ffiffjj)2
Solving this equation together with the equation zz = 1 of the circle (DEF),
we find
- 1 + — + «z - es + ------^ ------ = 0, ( 110)
Z ((T3 -ffi(T2)2
and since
*5 *1
<r2 ___1_ 01 02 03
*3
it follows that ee~ = a\ 0 |/( 0 3 — ax 0 2)2 and, hence,
_2 _2 —2
03 03 0102 03 + 01 02
—ee +
(ff3 — (Tj <r2) 2 (<r3 — <7X<t2)2 <r3 — cr1 <j2
Equation (110) takes the form
+ + -1 = 0
Z G3 Gi G2
or
■i. + l2 + -..??■ »■. = o
z a3 0i <t2
az2 H------”— z + e = 0. ( I ll)

0’s 0"l 0"2
The discriminant of this equation is equal to zero:
^ o \ a \ - o \o \ g \ g \
A = -------------------------s s = ---------------------------------------------- = 0 .
— Gx <72) 2 (<73 — G x G 2) 2 (g 3 — G xG 2) 2
Consequently, equation (111) has equal roots. This means that the radical
axis of the circle (I) and of the Euler circle (0 9) of A ABC has a single
point <*>o in common both with the circle (/) and the circle (0 9), that is,
the circles (0 9) and (/) are tangent at the point <P0- The affix <p0 of point <P0
(the point of tangency) is found from equation (111):
^ Gi G2 G i G2 Gi G2 G3 G2
(g 3 G 1 G 2) £ G3 (Ji (^2 Oi 0*1
And so we have
G2
<Po = — •
<*1
Remark. If for the unit point on the circle (DEF) we take a Boutain
point of A DEF, then g 3 = 1 and the affix cp0 of the Feuerbach point <P0
may be written thus
0i
7°. We find the affixes xa, xb, zc of the centres of the circles (Ia), (Ib), (Ic)
escribed in the angles A, B, C of A ABC. The slope of the straight line IB
is equal to b/b and, hence, the slope of the straight line perpendicular to
IB is equal to —b/b, that is,
_ _b Zl +*3 Z\ z3.
X~ b z 1 Jr~z3
and the equation of the bisector of the exterior angle B is of the form
_2_
^1 + Z3 V Z1 ++ Z3
*3 ./
or
- 4zj z3
z -f Zj z2 z = - ( 112)
Z1 + z3
In similar fashion we can write down the equation of the bisector of the
exterior angle C:
, - 4zx z2
z + zx z2 z = (113)
Zi + Z2
From the system of equations (112) and (113) we find the affix z = za
of the point /a:
(Z3 - Z2) Ta = 4(T3 ( — ---------- — 1— ),

V ^+ Z g z1 + z 3J
^3 ^2
(z3 — z2) = 4u3
(zi + z2) + z3)
whence
4<73
=
Oi + z2) (zx + z3)
Similarly,
4<t3 _ 4<r3
n = :> Tc
(Z2 + Zi) (z2 + z3) (z3 + Zj) (z3 + z2)
Since
(z 2 + z 3) 0 3 + Zj) ( z t + z2) = al a3 — a 3,
these formulas may be rewritten thus:
4<73 t, +I Z
(z2 _3)%— • 4a3
(o-i - *l),
°2 °3 <7l <r2 <r3
4g3 4cr,
** = (z3 + Zi) = (ffl - Z2),
<Ti er2 <r3 0*1 <72 (73
4<r3 4o-3
(Zl + ^2) (tfl - z3).
&1 &2 ^3 &i cr2 (73
or thus ^divide the numerator and the denominator of the fraction by <rs
*2
and substitute dx for
)■
4
r„ = • ( Z2 + Z3) kx - *i)>
- 1 k i I2 — i
4
- ks + z l) = - k i - z2),
k il2- i k il2- 1
4 4
Zr = ■k i + z2) = • k i - -3).
k il2 l k il2- i
or, finally, thus:
ra = — 2fl(oi — zj) =— 2/? o’!+ 2R zu
zb = — 2/?(o'1 — z2) =— 2/? o'!+ 2i?z2,
t c = — 2-R(oi — z3) =— 2/? 01+ 2R z3.
From this it follows that the centre of the circle (IJblc) has the affix
4oj o3 4ff!
— 2R o 1 =
Gi <J2 C3 (Ti (Ti 1
and the radius of the circle (IaIbIc) is equal to 2R.
We now find the affixes tl912, f3 of the points 7\, T2, ^3 of contact of
the circles (7fl), (Ib), (7C) with the sides BC, C4, ,42? respectively. Since
the line segments TXD and BC have a common midpoint, it follows that
t x -{- zx = b + c, whence
* a _L_ _ 2zj z3 2z2 zx _ 3z2z3+ z 1z3+ z 1z2—z?
/i — 0 + c — z1 — — r z,1 = z.
Z i + ^3 *2+ ^ 1 " Oi + Z 3 ) (zx + z2)
2 z 2 z3 + (T2 — zf 2(73 +
z 1 < 2 —
t z \
= *1
k i + z2) (zx + z3) (zx + z3)(zx + z2)
Similarly,
2 o -3 + z2 o-2 — zi , _ 2 o-3 + z3 o2 — z3
^2 — t3 =
(z2 + Zi) (z2 + z3) (z3 + Zi) (z3 + z2)
We now find the radii ra, rb>rc of the circles (70), (Ib), (Ic). We first find
4(7, 2o 3 + zx o 2
za — t, =
k i + z2) (zx + z3) k i + Z 2 ) ( z 1 + z3)
= 2zxz2 z, — zx(z2Z3 + Z3 zx + Z i z2) + zx
k i + ^2) k i + z3)
ZaZa-ZjZg-ZjZa+zf _ z ^ -z ^ -z ^ -z ,) k j - z 2) k i-^ s )
zi — —— :— —---------------- —zi = z i
k i + z2) k i + z3) k i + z ^ k i + z3) (zi+ z2) k i + z3>
- h (Z l - Z 2) ( Z 1 - Zo)
ra
ID Zl (Z l + Z 2) ( z x + Z q)
and, similarly,
= (z2 - *i) (z2 - Za) r = (23 - zx) (z3 - z2)
(*2 + Zl) (z2 + *3)’ C (z3 + Zl) (z3 + z2)
If for the unit circle we take the circle (Ia) = (DaEaFa) escribed in the
angle A of A ABC [Da9 Ea9 Fa are the points of tangency of this circle (7a)
with the straight lines BC, CA, AB], then the proof of the fact that this
circle touches the Euler circle constructed for A ABC will be precisely
the same as the proof that the circles (/) and (0 9) are tangent; however,
now it is necessary to assign the affixes zl9 z2, z3 to the points Z>fl, Ea>Fa.
Then the affixes a, b, c of the points A, B ,C will remain
_ _ ? u _ _? c __ _
z z + Z 2 Z 3+ Z 1 Z x + Z2
and even the affix <p± of the Feuerbach point &l9 in which the circles (7a)
and (0 9) are tangent, will be <Pi = — , where a1 and <r2 are expressed
n terms of the affixes zl9 z2, z3 of the points Da, Ea, Fa. We have to express
(pl9 <p2i <p3 in terms of the affixes zl9 z2, z3 of the points 7), E9F. The fore
going simplifies this problem because the fact (established above) that
the circles (7a), (7^), (7C) are tangent to the circle (0 9) will simplify the
computations (see below).
The equations of the circle (7a) and the circle (0 9) are
(z - ra) (z - T a) - r l = 0
(z — c) (z — e) — p2 = 0
By what has been proved, these circles are tangent at the Feuerbach point <Pl9
the affix of which is, consequently, found from the equation
'a
+ Ta ~ — — + «,
Z -* a z —e
r2
'a
+ Ta — e = 0,
Z — Ta
rl(z — t) — p \z — Ta) + [z2 — (Ta + e) z + xa e] (za — e) = 0,
(ia — e) z 2 + [rl — p2 — (xfl + e) (fa — b)] z — £r2+xa p2+ t a £(fa—e)= 0.
Since the circles (Ia) and (0 9) have only one point in common, it follows
that this quadratic equation has equal roots, which are the affixes (px of
the point 4>x of tangency of the circles (/fl) and (0 9):
r? — n2 (ta + e ) (fa - e) rl-p* Ta + £
<Pl =
2(e - ta) 2 (6 - t a) 2
zfz?z|
(zi - z2)2 (zx - z3)2 ____________________________
( Z i + Z2) 2 (zx + Z3) 2 (zx + Z2) 2 (z8 + z3)2 (z3 4- Z i ) 2
(Z i + Z2 + z 3) 2 4zx(z2 + z3)
f e + Z t ) (z% + Z3) (z3 + zx) (zx + z2) (z2 + z3) (z3 + Z i) _
, i r 4zx z2 z3 , ,_______________________
(z2 z3 + z3 zx + zxz2)2 1
2 L(zx + Z 2) (zx + Z j) (zx + Z2) (z2 + z3) (z3 + zx) J
_ _1_ (zt — z2)2 (zx — z3)2 (z2 + z3)2 — zf zf zf
2 (zx + z2) (z2 + z3) (z3 + zx) (z2 + z3 — zx)2
| 1 4zxz2z3(z2 + z3) + (z2Z 3 + Z3 zx + zxz2)2
2 ( Z X + Z2) (Z g + Z3 ) (z3 + zx)
Furthermore,
Ol — z2) (zx — z3) (z2 + z3) — zx z2z3 = (zx — z2 — z3) (zx z2+ z x z3—z2 z3)^
' Z2) (zx — z3) (z2+ z 3) + z x z2 z3= (z x—z2—z3) (zx z2+ z xz3—z2 z3) + 2<x3.
Hence,
(zx - z2)2 (zx - z3)2 (z2 + z3)2 - z? zf zf = (z2 + z3 - zx)2
X (zxz2 4- zxz3 — z3 z3)2 — 2zxz2 z3(z2 4- z3 —zx) (zx z, + zx z3 — z2 z3),
and so
1 (z24-z3—zx)2 (zxz24 - z xz 3 —z2z3)2—2zx z2z3(z24-z3- z x) (zxz24 z xz3 - zz3
<Pi=
2 (z2 4- z3) (z3 4- zx) (zx 4- z2) (z2 4- z3 — zx)2
1 4 z xz 2z 3( z 2 4- z3) + (zxz24- z2z34- z3zx)2 = J_ (zxz2 4- zxz3 — z2z3)2
+
2 (z2 4- z3) (z3 4- zx) (zx 4- z2) 2 (z2 4- z3) (z3 + zx) (zx + z2)
i
2 Z j Z 2 Z’3 ( z 2 + ^ 3) ^1 ^2 ^ 3 (^ 1 ^2 4“ ^1 ^3 ^ 3 ) ________
4-
(z2 + Z3) 0 3 + *l) (* 1 + ^2 ) O2 + *3> (*3 + * l) O l + z 2> O 2 + *3“ * l)
1 <t| <TZ Z j Z2 Zx Z3 Z2 Z3
+
2(<7j<72— (73) 2 <7j 02 (Tz Gi O’2 G3 ^2 "1" ^3^1
1 a\ g \ gz Zi z 2 + Zj z3 z 2 z3
+
2 (Tjj — ^3 &1 & 2 ^3 0*1 ^2 ^3 ^2 ^3
In similar fashion we find

Os Z2 Z3 + Z2 Z1 ZS Z1
<p2 9
&2 ~~ O3 0 \ g2 g3 Z S 4“ Z 1 Z2
°2 Os Z3 Z1 + ZS Z2 Z1 Z 2
<P3
0\ o2 — °3 Oi G2 <73 Z 1 4“ Z 2 — Z S
The formula for the affixes of the Feuerbach points are better
as follows:
<NCl
°3 + Z1 Z2 Z2 Z3
to
Z1 Z3
(Pi + 9
&i a 2 03 03 Gi G2 Z 2 + Z 3 —Z 1
a\ Os Z2 Z1 + *2*3 Z1 Z3
<p2 + 9
®2 ~~ ° 3 G3 Gi O’2 Z 3 4~ Z 1 Z2
o\ Os Z3 Z1 + *3*2 Zi Z2
<Pz + 9
0\ o2 °3 03 Gi O’2 Z1 + Z2 — z3
or as:
Z\ ZS l Z1 Z2 Z2 Z3
<Pi = £ + P
Z 2 4~~ Z 3 Z1
Z2 Z1 Z2 Z3 Z3 Z1
<p2 = £ + p
Z3 Z! Z2
Z3 Z1 + Z3 Z2 Z1 Z2
(p3 = £ + P
Z1 + Z2 — Z3
or as:
<Pl = « + PUU <P2 = « + PW2, <P3 = £ + pW3,
where
_ zxz2 + Zj z 3 — z 2 z 3
111 — >
Z2 + Z3 — Z1
Z 2 Z 1 4 “ Z 2Z 3 Z3 Z1
U2 =
Z3 i Z1 — Z2
Z3 Z1 4 “ Z3 Z2 Z1 Z2
Mo =
Z1 + Z2 — z 3
[Note that the coefficients of p are complex numbers equal to unity in

absolute value, for example,
_ z\ Z2 Z1Z3 Z2Z3
Wi — ; >
z 2 ~r z 3 zl
— Z1 Z2 Z 1 Z3 Z2 Z3 Z 2Z 3 Z1
--------------------- —--
J _ + J _______ L Z l Z2 + Z1 Z3 — Z2 H
Z2 Z3 Z1
and, hence, u1u1 = 1. This of course immediately follows from the fact
that \<pk — e\ = p {k = 1, 2, 3).]
Problem 33. ABC is an arbitrary triangle; {ABC) is the circle circum
scribed about it with centre O; the radius of circle (O) is equal to R;
(/) is the circle inscribed in the triangle ABC; I is its centre and r the
radius. Let d be the distance between the centres O and I of the circles
circumscribed about and inscribed in A ABC. Prove that
d 2 = = R 2 _ 2 R r.
Solution. Let D, E, F be the points of contact of the sides BC, CA, AB

and the circle (/). For the unit circle, take the circle with centre /. Then
the affixes of the points D, E, F, will be rzl9 rz2, rz3. In the preceding
problem we obtained the following expression for the affix o:
. _ &3
o >
<J1 <T2 03
where al9 o2, (t3 are the basic symmetric polynomials of the affixes of the
vertices of the triangle. However, since these affixes are now taken in
the form rzl9 rz2, rz3, we obtain the following expressions for the affix o
of point O and for the radius R of the circle {ABC):
20~i 0~3 2(7,
r, R =
(7i C2 0"3 &3 0"! (T2
Thus,
o = — Rol9 o = — Rdx
and so
d 2 - OP = o-o = R?c1d1
= R? = R%( 1 - ~ °2 \ = r 4 l _ = R2—2Rr.
°3 V ^3 / V -R /
Problem 34. Construct a triangle ABC if we know the points A0, 2?0, C0
of intersection of the bisectors of the interior angles A, B ,C with the
circle {O) = {ABC) (Fig. 36).
Solution. Take the circle (ABC) for the unit circle (O) = (ABC) =
= (A0 B0 C0) circumscribed about the given triangle ABC. Let zu z2, z3
be the respective affixes of the points A, B, C. The bisector of the interior
angle A of A ABC inscribed in the circle
{ABC) bisects by a point A0 the arc BC
subtended by the chord BC; note that the
points A 0 and A are located on different
sides of line BC. The situation of the points
B0 and C0 is similar (see the statement
of the problem). Therefore, for the affixes
a0 = z3, *o = K* 3 Zi, c0 = i *2 014)
of the points A0, B0, C0 values must be
chosen (each of the radicals has two values)
Fig. 36 so that the points A and A0 lie on different
sides of line BC, the points B and B0 lie on
different sides of line CA, and the points C and C0 lie on different sides
of line AB. It is under these conditions that we have to solve the system
(114), which we rewrite as follows:
z3z3 == =L bo, Zi z2 Cq. (115)
From the system (115) it follows that
z\ z\z\ = alblc%
so that
Z\ ^2 Z3 — i Qq bo Cq.
In the case
Z\ Z2 Zq = Qq Cq (116)
we have
___________ Cqa0 a0 b0
Z\ — » z2 - 5 z3 . (117)
Qq bQ Cq
In the case
Zi Zo Z3 = QqbQCq (118)
we have
bQCQ CqQq ^0^0
Zi - z2 — , z3 , (119)
Q0 ^0 ^0
Now we have to establish which of the two solutions (117) or (119) (or
neither or both) constitutes a solution to the problem. Let us investigate
the values of zu z2, z3 given by formulas (117). Since the foot of the perpen-
Zo j £
dicular dropped from point O on line BC yields the affix — ---- —, it
2
follows that the equation of line BC may be written in the form
—L_ + = 2
Z 2 ~f~ Z3 Z 2 “f” Z Z
2 2
or
z
+ — ■-z — - 1 - 0 . ( 120)
Z2 + Z3 z 2 + z 3
We have
c0a0 a0b0
*2 + z3 = ■— i ------- ~ ( * i + cS),
O0 Co D0 C0
(i+ 6p + eg
_boCo
z 2 T~ Z 3 — --------
*0 c ') 0q b0Cq
and the equation (120) of line BC takes the form
f(z) = -----bo£o----z + a_oboC_o j _ l = o.
a0(bl + Co) b% + c§
We find /(a 0) and /(z,) = / ( ^ £2j :
^ 0 Cp , b p Cq _ (b p Cq)2
/(a.) =
bo + Cq b o ~ { - Cq bo + Cq
fljeg 0q _ j ^ (0q - fro) (0q ~ c%)

/(*i) =
alibi + eg) b\ + eg 0g(Ag + eg)
From the last two relations we have
f ( zi) == (al—bl) (al—eg) _ _ (fl0—60) (a0—e0) (bp—Cp) (a0+fr0) (0o+eQ)
/(0b) “ 02o(6o - Co)2 al(bo - e0)3
but
0o 0o 1 al 1 a0
i
( A qB qC q) — ^ 60 *o 1 bp 1 bp
4 Qq bp Cq
Cq Cq 1 eg 1 e0
- — :----------(c0 “ 0o) (c0 - *o) (bo ~ 0o)

4a0 b0 Cp
and, consequently,
4
(cq — a0) (c<) — bo) (b0 — a0) = a0boCo(AoBoCo) = 4/Vz0ioô(ôôô)-
i
Thus,
f(z l) (a0 -}~ b0) (dp 4~ Co) (b0 c0)
4icio bQCq(AqB qCq)
f(ao) a&b0 - c0)4
We consider the point A * with affix a* = — a0, which point is diametri
cally opposite the point A0 on the circle {ABC) = (A0 B0 C0). We have
(*o + b0) (a0 + c0) (b0 - c0) = (b0 - fl?) (c0 - a%) (b0 - c0)
4/tf0b0c0(A* B0 Co).
To summarize:
f&i) = - 16(^0 B0 Co) (A* B0 Co)
/K ) (b0~ c 0)4
= - 16(ô B0 Co) {At B0 C0) [ - &o Co --12. (121)

L{bo ~ Co)2]
The number u = b0Co/(bo — c0)2 is a real number. Indeed,
1
— ___ ^ 0^*0 ___ b o Cq ___
/_L _ M 2 ~~ (*• - co)2

'^ 0 Co)
Therefore, u2 > 0 (u # 0). Now note that A A qB0C0, whose vertices are
the points of intersection of the bisectors of the interior angles A, B, C
of AABC with the circle (ABC), is always an acute-angled triangle. Indeed,
the interior angles A0, B0, C0of A A 0B0C0are respectively equal to
B+ C 71 A
2 ~ 2 2'
C+A 7t B
~~2 ~2 2’
A+B n C
2 ~~ 2 2
irrespective of whether A ABC is acute, obtuse or right-angled.
From this it follows that the diameter A0A$ cuts the chord B0C0 and,
------> ------>
hence, A A 0B0Co and A A ^B 0C0 have opposite orientations. Thus, the num-
bers (AqBqCq) and (A*BqC^) are of opposite sign, and from formula (121)
it follows that f(zi)lf(a0) > 0, that is, the points A 0 and A lie on one side
of the chord BC. Thus, the values (117) do not afford any solution to
the problem. The required solution of system (115) is given by formulas
(119), provided that A AqBqCq is acute-angled. If it is obtuse-angled or
right-angled, the problem does not have any solution.
The points A, B9C are points of intersection of the altitudes of A A 0 B0 C0
and the circle (ABC). Indeed, the slope of line B 0 C0 is equal to —b0 c0
and so the equation of the perpendicular dropped from point A 0 to the
line B0 Cq is of the form
z a0 = b0 Cq(z ab
solving this equation together with the equation zz = 1 of the unit circle,
we obtain
z a0 — b0 Cq j I,
V z <*o)
_ b0 Co(z a0)
z — a0 = ------------------ .
a0z
One of the roots of this equation is naturally z = a0 (the affix of point A0)y
the other is z = — bQColaQ, which is the affix of point A :
Z\ — bo Cq/(2 q.
Similarly, proof can be given that
[z2 — coa o/bo> ~3== a0 bo/co,
which are the affixes of points B and C respectively.
Remark. From what has been proved it follows that formulas (114)
should be written as follows:
a0 = — ]fz2 ][z39 b0 = — ]fz3 J(z l 9 c0 = — Vzx ^ z2, (122)
where we take the same value for ][zx, / z 2, ]fz3 (in that case, for example,
Vzi ]fzi = Kz2 ]fZ 2 — z2» Yzs Yzs — Z 3 9 but if for ]f zx we take different
values in the last two formulas of (122), then ^zx ][z1 = — zx).
Problem 35. Construct A ABC if we are given the points A 1 ,B 1 ,C 1;
these are the points of intersection of its altitudes with the circle (O) =
= (ABC) circumscribed about A ABC [(O) = (ABC) = (A^C ^)].
Solution. Take (A ^ C ^ ) = (ABC) as the unit circle. Let al 9 bl 9 cx be
the affixes of the points A l 9 B l 9 Cl 9 and let z l9 z2, z3 be the affixes of the
vertices A , B, C of the desired triangle ABC. Since the slope of the line AB
is equal to —z 1 z 29 it follows that the equation of the altitude dropped
from vertex C on the side AB is of the form
Solving this equation together with the equation zz = 1 of the unit circle,
we obtain
/1 1\
z3 — zx z2[ I’
Vz z3)
*3= (Z - z 3) .
Z3 Z
One of the roots of this equation is naturally equal to z = z3 (the affix

of point C). The other is
z = ct = — zx z2/z3, (123)
which is the affix of point Cx. In similar fashion we find the affixes
0i = — z2 z3/zx, bt = — z3 zx/z2 (124)
o f the points 2?x and Ax.

Let us solve the resulting system for zx, z2, z3. Multiplying these equations
together pairwise, we obtain
4 = bi ci> 4 = ci ai> 4 = 0i bl9

whence
Zi = yî z2 = ]/cx |/"tfx, Z3 — V"01 K^l> (125)
and since each of the products of two radicals in the right-hand member
of each equation has two values, the system (125) has 8 solutions. Assuming
that in each of the equations (125), any value is taken for each radical
K , J/î, Yci (but one and the same value in each equation), we can
write all the solutions as follows:
S 1
,Fi
Zl = f ^ ]fct.
II
Z2 = Yc1 f a 1,
Zi = - V bi Vci> *2 = - / c x fa x. ?3 = f« l f* l.
Z\ = Y h V clt *2 = - Y a i> z3 = - Y<*i Y ^ i,
zt = - V bi Y c i, *2 = fc i /fli. z3 = - Y a x f £ ; .
zi ~ Y b i Y ci> *2 = - f c x ffli. z3 = fa f^ i.
1
Zl = - Y b i Y c 1. *2 = fc i ffli, z3 = f ax f />i,
(127)
Zi = Y b i Y c i> *2 = - f<a f a , 1 z3 = fa i f*i,
Zi = Y'bi Y c i» *2 = fc i fai. 2 3 = — f« i f*l- .
From the system (123), (124) it follows that dxbxCx = — zxz2zz and so
all number triples (126) fail to serve as solutions to the system (124). Now
any row of relations (127) is a solution of the system (123), (124). Indeed*
*1 Z 2 _ _ ][b i 1 /c x f c x Va! _
Z2 - ] f a i I fo U
Z 2 Z 3 __ __ )fc i )[<*! Y a t ] f b i
- V bi ] fc .
Z3 Z1 = _ ] fa i y b i y Cl _ b
Z2 - yCl y a i 1
and also for the other three relations (127).
To summarize, then, there exist four triangles that satisfy the condition
of the problem. Let us construct, for example, a triangle corresponding
to the first row of solutions (127). To do this, draw to the circle (A ^ C x )
any one of two tangent lines parallel to the straight line QAX(& is the unit
point); the point Pi, which is the point of contact of the tangent drawn
to the circle (^j^C j), has as its affix one of the values ] (Fig. 37).
In similar fashion, construct the points Qx and Rl9 whose affixes are the
values 1fbx and ]fc±. To construct a point with the affix Yb± ]/ci, draw
through the point Q a straight line parallel to the line QxBxl the second
point P9 the point of intersection of the drawn line and the circle (AxBxCx),
has the affix Yî Yci• Finally, point A , which is symmetric to point P
with respect to the centre O of (AxBxC^) has the affix ~~Ybi Yci-The
points B and C with affixes —Yci an(* —Yai Y&i are constructed
in similar fashion.
The other three triangles that satisfy the conditions of the problem are:
AQR, PBR, PQC.
Problem 36. Inscribed in the unit circle is A ABC, the affixes of whose
vertices are zl9 z2, z3. Find the affixes t 0, rl9 t 2, t 3 of the centre of the circle (/)
inscribed in that triangle (affix t 0) and of the centres of the circles (Ia)9
(/*), (/c) escribed in that triangle (in the angles A9B9C9 respectively).
Solution. The centre / of the circle (/) inscribed in A ABC is the point
of intersection of the bisectors of the interior angles. These bisectors inter
sect the circle (ABC) in the points A0, B0, C0; note that /\A QBQC0 is always
an obtuse-angled triangle (see problem 34). The affixes a09 b09 c09 of the
points A0, B0, C0 are expressed by the formulas (122) of problem 34:
Qq = YZ2 YZ3’ *0 = YZ3 YZ1* Co YZ1 Y^2> (128)
where the same values are taken in all formulas for the square roots J/z^
Yz 2 >YzZ’ note that these values for Yzi> / z2, Yzz must always be taken
so that formulas (128) define just the points of intersection of the bisectors
of the interior angles A, B, C of the given A ABC with (ABC). Namely,

for |f z l9 ][z2, J/z3 we have to take the values so that the inequality \][z1 +
+ V Z 2 + Yzz\ < 1 holds, that is, so that the triangle whose vertices
have the affixes |f zl9 Y z2, ]f z3 is acute-angled. Indeed, then
lô + b 0 + c0)| = | - Y z 2 f z 3 — V z3 Y z i — Y z i V Z 2 \
= |V*2 Y Z3 + V z3 Y Z1 + Y Z1 Y Z2\
= \ Y z i Y zz + Y zz Y z i + Y zi Y z 2
iKzi + Y z 2 + Y z I
3
jY z 2 Y zs + Y z 3 + Kzil/*a lfzl YZ2 VzZ1

— i / zi + y"z2 + v £3! < i*
That is, A A 0BqC0 is acute-angled' [there are only two such choices for
the square roots of zl9 z2, z3; if one of them is denoted by Yzi> Yz 2> Yzs>
then the other will be —|/zl5 —J/z2, —]/z3, and any one of these choices
yields the formulas (128)].
To summarize, then, AA0, BB0, CC0 are the bisectors of the interior
angles A, B, C of A The slope of the bisector AA0 is equal to
Y" = zi YzzY zs,

a a — Z,
YzzVz3 zx
and, hence, the equation of the straight line AA0 is
z — Zi = Z jI/z ss | / z 3 (z — Z!>
or
z — Z! Vz2 /z ,z = z1 — / z 2 / z 3. (129)
The equation of BB0 is
z — z2 ]/z3 J/zx z = z2 — ]/z3 / z x. (130)
From the system (129), (130) we find the affix t 0of the centre I of the circle
(/) inscribed in A ABC:
t 0(z2 V*i ~ zi If**) = - zM Y zz - / zi) ~ Yzi f z2 Vza (z» ~ zi)>
or
To K z i / a ( / z2 - |/ z i) = - Z iZ 2 ( K z g — /z j)
- fz j Jfz2 |/z3 [(l^z2)2 - ( /z ,) 2]

That is,
To = - YZ1 VZi - f zS( / z2 + /A ).
and, finally.
To = — f z 2 / z 3 — ][z3 ^Zj — / z x l/z2 = a0 + *0 + c0. ( 131)
If we now take the point B$, which is symmetric to the point B0 with res
pect to the centre O of the circle {ABC), then the affix of the point B* will
be equal to 6* == —b02LndBB* will be the bisector of the exterior angle of
the triangle at the vertex B. The slope of BB* is
?2 + ^0 - YzaYz
= — Z* / Zs / Zl.
^ 2 ~T bo
Yzs f z
and the equation of line BB* is
z - z2 = z2 / z 3 f z x (z - z2)
or
z + z2 YzzYzi z = Z2 + YzaYzv '(132)
1 5 -8 1 0
From the equations (129) and (132) we find the affix t1 of the centre Ia of
the circle (Ia) escribed in angle A of triangle ABC:
Ti (zx } f z 2 + Z2 Yzx) = ZZZX (fz 2 + Yzl) ~ Y zl f Z 2 Yz 3 ( z 2 - Zl)
or
Tj j/zj / z a (/z x + / z 2) = Z jZ g (]/zx + ]/z2) - f z , 1fz3 [(/z2)2 - fl/z^2],
that is,
Z1 = / zi KZ2 — YZ3 ( / Z2 — / z l) = f Zl f Z2 — Y Zs Y z z + fz s/z x ,
and, finally,
Tj = — / z 2 f z 3 + f z 3 Yzi + f zi f za = a0 — b 0 — c0.
Similarly we find t 2 and r3. Thus,
T0 = — Y Z1 Y Z3 — Y Z3 Y Z1 — Y Z1 Y Z2 = «0 + b0 + C o,
Zx — | / z 2 V z 3 "I" K z 3 V Z1 "i" K Z1 K z2 — Qq ho C o,
x2 = fz-t / z 3 —Yz3 Yzx + Yzx f z 2 = —a0 + ho — c0,
Ta = Y z 3 ] / z 3 + / z 3 Y Z1 — Y z 1 f z 2 — — — h0 + C0 .
From formulas (133) it follows that

*0 + Zl _ „ T« + T2 _ A Zo + z3
— Oo, ---- ~------— c0.
2
That is, the midpoints of the segments IIa, IJb, 1IC coincide respectively
with the points A0, B0, C0 in which the bisectors of the interior angles of
A ABC intersect the circle (ABC) --- (O) circumscribed about Is ABC
(Fig. 38).
From formulas (133) it also follows that
*2 + Z3 _ _ Z3 + Ti _ u Zx + Z 2 _
------------ — a0, - — oo, — Cq,
that is, the midpoints of the segments I J a, IaIb coincide respectively

with the points A$,B$,C* in which the bisectors of the exterior angles A, B, C
of A ABC cuts the circle {ABC) = (O) circumscribed about A ABC. The
points A*,B$9C$ are symmetric respectively to the points A0, B0, C0 about
the centre O of (0)=(ABC) so that A0A$9 B0B£, C0C£ are diameters of
that circle. Incidentally, from that it follows that the circle (0)=(ABC)=
= (A0B0C0) = (A*B*C*) is the Euler circle of A IJbh and therefore the
radius of the circle (/fl4 /c) is twice the radius of the circle {ABC) [this has
already been proved analytically in problem 32, item 7° (see solution)].
Finally, since IA _L IbIc, IB _L IC _L IaIb, it follows that I is the

orthocentre of A IJbh> furthermore, since (tx + t 2 + t 3)/3 = — (a0 +
(- b0 -f c0)/3, it follows that the centroid G* of A IJbh is symmetric to
the centroid G of A A qBqC0 about the centre O of the circle (ABC), or
the point G* is the centroid of A ^*B$C*, which is symmetric to A A 0B0C0
with respect to the point O.
Problem 37. ABC is an arbitrary triangle and P is an arbitrary point.
Prove that the straight lines a!, b', c', which are symmetric to the straight
lines a = AP, b = BP, c = CP about the bisectors AI, BIy C l of the inte
rior angles A yByC of A ABC also pass through one and the same point Q.
The points P and Q are said to be isogonally conjugate with respect to the
triangle ABC (Fig. 39). Prove that if the circle (ABC) is taken as the unit
circle and if the affixes of the points A, B, C, P, Q are equal respectively
to zl9 z2, z3, p , q, then they are connected by the relation
p + q + o»P q = <?! (134)
(this relation was obtained by the English mathematician Morley). Express
q in terms of zl9 z2, z3, p.
Solution. A0, B0, C0, the points of intersection of the bisectors of the
interior angles A , B, C of A ABC with the circle (ABC), always form an
acute-angled triangle A0B0C0. If for ]fz l9 ]/z2, ]/z3 we choose values such
that !Yz± + J/z2 + |/z31< 1, then the affixes a0, b0, c0 of points A0, B0, C0
arc
ai>=— Vz2 )[z99 b0 = — Yz9 Yzl9 c0 = — Yzi Y z 2
[problem 36, formulas (128)]. The slope of the bisector of the interior angle
A of A ABC is
.*■ + = *, V* yf,
Zl + ]f^2
and the equation of this bisector is
z — zx I!zn ]fz3 z = z1— ][z2 ]fz3 (135)
[see problem 36, equation (129)].
The equation of the perpendicular dropped from point P on this line
is of the form
Z -P = — Zi ]fz 2 ]fz 3 ( z - p)
or
z + Zj, ]fz2 ]fz3 z =~-p + zx f z 2 1fz3 p. (136)
Adding the equations (135) and (136) term by term, we find the affix
z —p' of projection P' of point P on the bisector of the interior angle A
of A ABC:
p' = ~~ (Z i + P - ]fza ][z3 + zx ][z3 ]/z3p).
The affix p f of the point P *, which is symmetric to point P about the bi
sector A l of the interior angle A of A ABC is found from the relation
p + p*
= p' = ( Z i + p —Vzz ]fz3 + Zj ][z2 ][z3p).
2
whence
P t = Z1 - V*3 + 1^2 1^3 P-
The equation of the line >42^* may be written thus:
2 Z />f
Z ZX JP f 0
1 1 1
or
Z *1 *1 — f * 2 V *3 + Zx If z 2 ][ z 2 p
* Zr z\ -][z2 ][z2 + z 1 ]fz2 ][zs p = 0

1 1 1
or
()fz2 ]fz3 - z 1]fz2 ]fz2p )z + ( - ][z2]fz2 + zx y z2]fz2p ) z
— zx y z 2 y z2 + y z 2 yz2p + z x Vz2yz2 — ]fz2yz2p = o
or, multiplying both sides of the last equation by y z2 Yz2,
(1 — z\p)z + (—z2z3 + zxz2z2p)~z — zx + p + z xz2z2 — z2z2p = 0. (137)
In similar fashion we can write down the equation of the straight line sym
metric to line BP about the bisector BI of the interior angle 2? of A ABC:
(1 — z 2p) z + (—z2zx + zxz2z2p) z — z2 + p + z 2z2zx — z2zxp 0. (138)
The affix of point Q, the point of intersection of the lines (137) and (138)
that are symmetric to the lines AP and BP about the bisectors A I and BI
of the interior angles A and B of A ABC, may be found by solving the
system of equations (137) and (138). Since
A = 1 - * l/> — *2*3 + *1*2*3~P
1 z 2p Z3Zi + zxz2z2p
— — Z3 Z1 -f" (Top + Z2p — Z 2Z2p p + * 2*3 G*P Z*P + ZxZ2p p
= z2(z2 — Z j) - z2pp (z2 - Z j) = z2 (z2 — Z i ) (1 - pp).

it follows that A ^ 0 if and only if 1 — PP ^ 0, that is \p\ ^ 1; in other
words, when the point P does not lie on the circle (ABC).
(1) Assuming that \p\ ^ 1, that is that the point P does not lie on the
circle (ABC), we conclude that the system (137), (138) has a unique solu
tion for z and z : z =*•- q, z —q. This solution could be found, for example,
via the Cramer formulas, but we will take a somewhat different route;
namely, by substituting the value z — q, z = q into the equations (137),
(138) and subtracting the second equation from the first term by term,
we get
( z 2 — z ±)pq + z3 (zt — z2)q + z* - zt + z3 ( — — — ) + ^ ( ^ 1- ^ 2) = 0
V *1 *2 /
or
Z1 — Z2
pq + - z J - (zi - z2) - —— ( z l-z t)+ z 2p(zi—z2) = 0.
Z1Z2 Z1Z2
Cancelling zt — z2 and multiplying both sides by zxz29 we obtain
pq + <r9(p + q ) ^ a* = 0. (139)
Now this is the Morley relation. Indeed, passing to conjugate numbers,
we have
P q + 0z(P + q) “ 0 2 = 0. (140)
But <78 = l/a3, <r2 = 0 J 0 2 and so
P + q + 03Pq = 0 i - (141)
From the relations (139) and (141) it is easy to express q in terms of p.
From (139) we find
0 2 - pq _ -
q =
02
and (141) takes the form

p + q + p(0 2 -p q - 02 P) = 0 i
or
q (i - pp) = 02 P2 - 0 2 p —p + 0 i,
whence
= 09 p2 -
p - p +01
0 2
1 — pp
From the symmetry of this expression with respect to zl9 z2, z3 it follows
that the straight line C as well passes through the point Q.
(2) If point P lies on the circle {ABC) but does not coincide with any of
the points A ,B 9C9 then the straight lines (137) and (138) are collinear;
as will be evident from what follows, there is a third line collinear with
them: it is the line symmetric to line CP with respect to the straight line
C l (Fig. 40). The slopes of these lines are:
1 1
^2^3 02P Zi - P P - Zi 03
03 = 03 —09Pt
1 - z\p 1- Z i P p(zi- p ) P
Z
and if we take a Boutain point for the unit point, then x = —p. Setting
p = cos a + i sin a, we find
x = cos (7i — a) + i sin (n — a).
The affixes of the ends of a diameter of the unit circle (ABC), which has
such a slope, are
]/x = ± [ c°s + ' S'” ( t - t ) ] '
From this it follows that if the point P describes the circle (ABC), then
this diameter will turn through an angle n in a direction opposite that of
the radius OP (Fig. 41). If the Euclidean plane is completed to a projective-
Euclidean plane by points at infinity, or ideal
points, we can then assume that if the point P /
lies on (ABC) but does not coincide with any
one of the vertices A, B, C of A ABC, then it
will be associated with the ideal point of the
projective-Euclidean plane. The Morley relation
p + q + OzPq= <7i does not, of course, hold in
this case since no complex number can be as
sociated with ideal points.
Finally, from geometric reasoning it is clear
that if point P lies on line BC but does not
coincide with B or C, then point A will be
isogonally conjugate to it (the same goes for the
lines CA and AB). Finally, if point P coincides
with one of the vertices of A ABC, for example
P = A, then any point of BC will be isogonally conjugate to it (and the

same goes for the cases P — B and P = C). This can also be verified ana
lytically: let p = az2 + /?z8, where a and p are real and a + P = 1 (that
oc B
is, the point P lies on the straight line BC). Then p = ----- 1----- and,
Z2 z3
assuming g = zl9 q = — , we will have

zi
p + q + o-iP q = «z2 + Pz2 + zx + ztz,z3 (■“ •- + - - ) ™

V z2 zs / zt
— <*Z2 + PZZ + Z1 + aZ 3 + ^ Z2 — Z1 + (a + P)Z2 + (a + P ) Z3
= Z1 + Z2 + Z3 ~
Note that the centre / of the circle (/) inscribed in A /4#C, and the cen
tres Ia, /*, l c of the circles (/fl), (/fc), (Ic) escribed in the angles A, B, C of
that triangle are isogonal conjugates of themselves with respect to A ABC.
It will be proved below, in problem 41, that besides the points /, /a, Ib, Ic
there are no points that are isogonal conjugates of themselves with respect
to A ABC.
We also note that if we eliminate from the Euclidean plane the circle
(ABC) circumscribed about A ABC and eliminate the lines BC, CA> ABy
then the isogonal correspondence between the points P(p) and Q(q) will
be a one-to-one mapping (P *-* Q) or a one-to-one and involutory trans
formation (involution) described by the relation p + q h p qoz — ox.
If the Euclidean plane is completed to a projective-Euclidean plane,
and the straight lines BC, CA, AB are deleted, then the correspondence
between the points P(p) and Q(q) will again be one-to-one, but then the
relation p + q + pqoz — ox will refer only to the ideal points P(p) and
Q(q). Finally, throughout the projective-Euclidean plane, the mapping
of P(p) on Q(q) will no longer be a one-to-one mapping and the relation
p + q + pq<Tz= gx will again be valid only for the proper points P(p)
and Q(q) (ideal points do not have affixes).
Problem 38. The circle (O) = {ABC) circumscribed about A ABC is
taken as the unit circle; zl9 z2, z3 are the affixes of the vertices A, By C.
Find the affixes of the points that are isogonal conjugates of the following
points with respect to A ABC:
1°. The point G, the point of intersection of the medians of A ABC.
2°. The orthocentre H of A ABC.
3°. The centre 0 9 of the Euler circle of A ABC.
Solution. Let us take advantage of the formula of problem 37;
P + P<*2 ~ — <r»P*
PP - 1
which defines the affix q of point Q ; this is the image of point P with affix
p under an isogonal transformation with respect to A ABC. We have the
following.
1°. Since the affix g of point G is equal to <t3/3, it follows that the affix
/ of the image L of point G under an isogonal transformation with respect
to A ABC is
<*1 2 (7?
aL i. ? .\_ (7o (7i (7a — + --------------
3 3 _ 3 3 (To 9<73
9 9
2
-1- 2 '
3 1 + ~q 9<---
7o ^ 3(7j
?1<*1 <^1 - 9
9
The point L, which is the isogonal conjugate of point G with respect to
A ABC, is termed the Lemoine point. Thus, the affix / of the Lemoine
point I of A ABC is
I _2 3 o~i
- 9
2°. The affix of the point that is isogonally conjugate to the orthocentre
H with respect to the triangle ABC is
_ of
_ ^3_______ q
(TiO1! “ 1 <T101 — 1
that is, the orthocentre H and the centre O of the circle (ABC) are isogo
nally conjugate with respect to A ABC (the straight lines AO and AH
are symmetric about the line AI and similarly BO and BH are symmetric
about B l, CO and CH are symmetric about Cl).
3°. The affix of the point isogonally conjugate to the centre 0 9 of the
Euler circle (0 9) with respect to A ABC is
5i_ + a\ ____ ?i_
2 2 2 2<t3 4(t3*2
oî _ o1o1
4 4
ffi a\
2 4o3 a2cr1 — 2xx
_ j CTiffi - 4
T
Problem 39. Prove that the midpoint M of a segment whose ends are
points Px and Ql9 which are isogonal conjugates (with respect to A ABC)
of the ends P and Q of any diameter PQ of the circle (Q), which is con
centric with the circle (ABC), describes (as the diameter PQ rotates) a
circle that touches the tangents drawn from the orthocentre H of A ABC
to the circle (ABC), Triangle ABC is assumed to be obtuse, and the radius
p of the circle (£) is assumed to be less than OH [only in this case is it pos
sible to draw tangents from point H to (ABC)],
Solution, Take (ABC) for the unit circle; let zx, z2, zz be the respective
affixes of the points A, B, C, Let PQ be an arbitrary diameter of (Q), and
let p and —p be the respective affixes of the points P and Q.
Denoting by px and qx the affixes of points Px and Ql9 which are the res
pective isogonal conjugates of the points P and Q with respect to A ABC,
we will have (see problem 37)
2 _
02P ~ P + <*1
Pi
1 ~ pp
<*sP2 + 02P + p + <T1
<h
1 - pp
From these we find the affix m of the midpoint M of segment PXQX:
m = Pi + <7i °dP* + ffi
2 1 -P P
From this relation it follows that if point P describes a circle (&) which
is concentric with the circle (ABC), then 1 —/>/>=constant [1 — pp —
= —< 7 , where a is the power of the point P with respect to (ABC)] and,
hence, point M describes a circle (T), the affix of whose centre is

t=
1 - pp
and the radius is
°3P* OP2
Ri =
\ \ - pp\ 11 — o p 2 i
Knowing the affixes of the points H and T, we can find the complex number
corresponding to the directed line segment HT:
<*i 0-1P P ___ OP2
t — <7, — ----- 01 = ~ =T — 0V
1 - pp 1 - pp 1 - OP2
Associated with the directed line segment HO is the complex number
Hence
HT OP2
OP2 - 1
HO
From this relation it follows that point T is the image of the centre O of
the circle (ABC) under a homothetic transformation with centre 77 and
OP2
r a tio ----------- . Therefore the circle (T7) described by point M touches
OP2 - 1
the tangents drawn to (ABC) from the orthocentre 77 of A ABC (only
here is the condition used that A ABC is an obtuse-angled triangle).
Remarks. (1) The radius Rx of (T) is equal to the radius 1 of (ABC) if
and only if
OP2
1 - OP2
whence, obviously.
OP2
1 - OP2
(2) The radius Rx of (T) is equal to the radius of the Euler circle if and
only if
(3) The radius R± of (T) is equal to the radius of ((2) if and only if
Problem 40. Given in a plane an arbitrary triangle ABC. The circle

(ABC) = (O) is taken as the unit circle. Let t 0 be the affix of the centre 7
of the circle inscribed in A ABC, and let xi9 t 2, t 3 be the affixes of the
centres Ia, Ih, Ic of the circles (7fl), (Ib), (Ic) escribed in the angles A, B9 C
of the given A ABC. Prove that the orthopoles of the straight lines 01, OIa9
OIb9 01 c with respect to A ABC are, respectively, the Feuerbach points
<P0, <Pl9 &2, # 3 of A ABC: the points 4>0, 4>2, are, respectively, the
points in which the Euler circle of A ABC is tangent to (7), (7fl), (lb)9 (7C).
By proceeding from the foregoing, find the affixes q>09 <Pi, <P2 ><Ps of the
Feuerbach points # e, 4>l9 0 2, <J>3. The affixes of the vertices A , B9 C of
A ABC are equal to zl9 z2, z3 respectively (see Fig. 35).
Solution. Let t 0, t 15 t 2, t 3 be the respective affixes of the points 7, Ia9
Ihy Ic. On the basis of problem 7, the orthopoles of the straight lines OI9 OIa9
OIb9 01c with respect to A ABC have the affixes
(142)
(143)
<
p* = 044)
<Ps = (*! - ^ y ) (145)
since the slopes of the lines 07, 0 Ia, OIb, OIc are respectively,
T0 Tj_ T2 T3
Xq — ^1 — “ > ^2 — _ i ^3 ~
To Tx T2 Ts
We now have to prove that <po, <p[, </>2, </>£ are the affixes of the respective
Feuerbach points. Let us first consider the formula (142):
Let og be the affix of the centre of the circle (0 9) of A ABC. Since o9 =

= — , it follows that
2
/ / ^3To
(p0 “ ~ — 9 o °9
2 t0
whence
, , ._ 1
I ^0 °9 I — " “
2
and so the point lies on the Euler circle. By the Euler formula for the
distance between the centres of a circumscribed circle and an inscribed
circle of the triangle,
Ol2 = R2 — 2Rr = 1 — 2r (because we have R = 1),
where r is the radius of (/) inscribed in A A B C . Since 012 = t 0t 0, it fol
lows from the last formula that
The equation of the circle (7) is
(146)
4
We now prove that the point <P'0 with affix
lies on the circle (/). We have
<Po - *0 = - T0 ,
2 2t0
and since r0 is the fixed point of an isogonal transformation with respect to
A ABC [see problem 37, formula (141)], it follows that
2 t0 + a 3 fg — a x = 0,
whence
*i ~ T?>
t0 =
2
and therefore
, _ *i y:)To 0*1 , 0*3Tp
</>0 “ T0 —
2t 0 2 2
ffsio <r3f0
( t 0t 0 - 1).
2t0 2t0 2 t0
From this we get

_«TSTo
<Po - To : ( t 0t 0 1)
2f0
and, hence,
(<Po - T o) ( 9 i - T o) = (T ° T(> — = r 2.
4
That is, the point <P'0 with affix </>' lies on the circle (/), the equation of which
is of the form (146). Furthermore, the slope of the straight line I<Pq is
<P0 - T o ____ 2 f §
-/ - ^ 9
<Po ~~ T0 T0
The slope of the straight line 0 9<P'Q is

*i
<Po
2
- ~<Po
2
and this means the points 0 9, #o, /lie on one straight line. To summarize,
then: the point <P'0 with affix </>o lies both on the circle (/) and on the Euler
circle (0 9), and, besides, the points #o, h 0 9 lie on one straight line. Now
if two circles have one common point (in this case, $o) and their centres
(in the given case, 7 and 0 9) lie on the same line as that common point,
then they are tangent to one another at precisely that point. But the point
of tangency of the circles (7) and (0 9) is precisely the Feuerbach point <P0,
and so the points 4>9 and <P0 coincide and, hence, q>'0 = (p0.
To derive the other three formulas
<Pl — ^P19 ^2 = <p2, <Ps = <Ps,
where q>[9 <p2, (p's are given by (143), (144), (145), and <p1? <p2, <p3, are the
affixes of the Feuerbach points <Pl9 <P2, <P3, we first derive the formulas
Oil = R2 + 2raR9 (147)
Oil = R2 + 2rbR, (148)
01* = R2 + 2 rcR (149)
for the distances between the centres of the circumscribed circle and escrib
ed circle (it is of course sufficient to prove only the first one of them).
Let us consider the inversion [7fl, r%|, whose centre is the centre Ia of the
circle (Ia) escribed in the angle A of triangle ABC, and the power is equal
to ra. Let the circle (Ia) be tangent to the straight lines BC, CA, AB respec
tively at the points Pl9 P2, P3 (see Fig. 35). Under the inversion [7fl, r%]9the
points A, B, C go respectively into the points'^*, B*, C*9 in which the fol
lowing straight lines intersect: IaA and P2P3; IaB and 7>37>1; IaC and 7>17>2.
The points A*, B*9 C* are the respective midpoints of the segments R 2 R&
P3Pi,PiP2. Thus, the circles (ABC) and (A fB fC f) go into each other under
this inversion with the circle of inversion (7>17>27>3)=(7a).The circle (AfBfC *)
is the Euler circle of A 7>17>2P3 and therefore the radius of (AfBfC *) is
equal to R'a=rJ2. On the other hand, the ratio R'JR of the radii of (A?BfC*)
and (ABC) is equal to \kja\9 where a is the power of the centre Ia with
respect to the circle (ABC), and k is the power of the inversion at hand,
that is k = r*. Indeed, suppose M and M ' are corresponding points under
the inversion at hand [point M on the circle (ABC) and point M ' on the
circle (A*B*C*)]. Then
(IaM )(IaM') = r*a ( = k ) . (150)
Let M y be the second point of intersection of the line IaM with the circle
(ABC). Then
(IaM )(IaM1) = c. (151)
From the relations (150) and (151) we find
(hM ‘)
ChM 1) a a
Incidentally, this means that the circle (0*)=(A*B*C*) may be obtained
from the circle (ABC) by a homothetic transformation with the centre of
similitude (homothetic centre) I a and ratio k j a . (Of course, the correspon-
dence of points under this homothetic transformation and under the inver
sion that we considered is different.) Since a homothetic transformation
is a similarity transformation, it follows that
g
K a
R o il- *T
Since the point Ia lies outside the circle (ABC) [the midpoint of segment
IIa lies on the circle (ABC) — see problem 36], it follows that OIa > R
and, hence.
2R O il ~ R2 ’
whence
Oil = R *+ 2Rra.
We assumed /J = l. Besides, Oil = t 1t 1 since is the affix of point Ia;
hence, ra = (txt l — 1). We now prove that the point with affix
2
<Pi
lies both on the Euler circle and on the circle (/a). We have
r
<P1 *1
2t,
and since Ia is a fixed point of an isogonal transformation with respect to
A ABC, it follows that
2xx + <r3 ff — <rx = 0,
whence
*1 — *3*1
Tl =
2
and therefore
, <*i *1 — *3*1
2 tx 2
* 3*1 *3*1 *3*1
■ ( V i - l ) . (152)
2 2t 2 2t x
From this we have
? ;-? ! = — ( v i- i)
2ti
and so
(*1*1 - l)2
(<P'i - Tj) (ipl - Tj) =
4
That is, the point with affix <p[ lies on the circle (/fl) whose equation is
(z - Ti) (z — = r\.
Furthermore, 0 9 = aJ2 is the affix of the centre of (0 9). Therefore,
/ 1 / fi \ 1 a3t1
2 - -2 « . = -
whence
| ^ - o 9| = 1/2
and therefore the point <P[ lies also on the Euler circle, the equation of
which may be written down in the form |z — o9\ = 1/2. The slope of the
straight line Ia <P[ is
<Pl - T, = 2ti__________ = Tf
V' ~ ^ ^ Tl (t.T, - 1) 3
2fx
The slope of the straight line 0 9<P[ is
Consequently, the points 0 9, Ja and lie on one straight line. Thus, the
point also lies on the circle (0 9) and on the circle (Ia) and the centres 0 9
and Ia of these circles lie on the same straight line as their common point
<P’i\ hence, (Ia) and (0 9) are tangent at the point <P[ with affix q>[. But
the point of tangency of (/a) and (0 9) is the Feuerbach point &l9 and so
the points and <P± coincide and (p[ — (px.
In the same way we can prove that cp'2 and cp3, that is, the affixes of the
orthopoles of the straight lines OIb and OIc with respect to A ABC are
the affixes of the Feuerbach points ^>2 and tf>3. That is, <P'2 and are points
of tangency of the circles (Ib) and (Ic) with the Euler circle (0 9) of A ABC.
Problem 41. Let zl9 z2, z3 be the affixes of the points A, B, C in a system
of coordinates in which (ABC) = (O) is taken as the unit circle. It is re
quired to prove that:
1°. The affixes t (), zl9 r 2, t 3 of the centres of the circles inscribed and
escribed in A ABC are given by the equation
T4 — 2 (7 2I 2 + 8(73 T + <?2 — 4(72(73 = 0.
2°. The affixes ij/0, i//l9 ^ 2, <A3 of the points !P2, symmetric
to the orthocentre H of A ABC with respect to the Feuerbach points are
found from the equation
(4cf2 — a\)+ 4^1j/3 + 2a1a1ij/2 + 4al\// + 4<72 — a\ = 0.
These points *F09 *Fl9 *F29 X
V3 lie on the circle (ABC) (see Fig. 42).
Solution. 1°. Let T be the centre of the circle inscribed in A ABC (or
the centre of one of the escribed circles). Any one of these points has the
characteristic property that it coincides with the point that is isogonally
conjugate to it with respect to A ABC. It is easy to verify the sufficiency
of this condition geometrically; namely, any one of the indicated four
points T is fixed under an isogonal transformation. We will now prove that
there are four such points. Indeed, for point Twith affix t( |t| # 1) to
coincide with its conjugate point, it is necessary and sufficient that the
Morley relation
P + q + <r3 P ~q' =
hold for p = q = t, that is,
2t + (73 T2 = at.
For example, we will prove that it holds for the affix
T# = - 1f z 2 1 ^ 3 - If z 3 1! z t - ][z2
of the centre of the circle (/) inscribed in A ABC (see problem 36). We have
- = _ _ _ J________ }_________1 _ = _ + KZ2 + fcg
0 K*2 1^3 ]/zz fzj ]fz1 ]fz2 1lzx Vz2 )[zz
or
To = — [*1 + Z2 + Z3 + 2 (^z2 ]fz3 + ]/z3 ]/zx + /z j / z 2)].
<r3
From this we have
2t0 + o3xg = 2(— f z 2 ]/z3 — 1lz3 ]fz2 — ]!z2 j/z2)
+ zx + z2 + z3 + H]fz2 1fz 3 + 1fz3 j/zj + |/zj ]/z2) = Zj + z2 + z3 = alt
and so also for the other affixes zl9 t 2, t 3 [formulas (133) of problem 36].
Adjoining to the equation
2t + a3 t 2 — ax = 0 (153)
the equation obtained by equating to zero the conjugate number of the
left-hand side, we obtain
2t -f a3 z2 — ax = 0, (154)
16-810
Fig. 42
whence
T <7
X =
and the equation (153) takes the form

— 2 f f 1 ff 3 Ta + t 4 <r|
Lx + <x3 ---------------------------------- — G1 = 0
or
8r + t* + — t 4 - 4ox = 0
0*3 <T3
or
T 4 — 2(7 2 T 2 + 8 0 -3 T + — 4(7! (73 = 0.
Since this is a fourth-degree equation, it follows that there cannot be more

than four fixed points under an isogonal transformation. But the centre
of the inscribed circle in A ABC and the centres of the circles escribed
in this triangle are fixed under an isogonal transformation with respect to
A ABC. Hence, the fixed nature of a point under the isogonal transfor-
mation is the characteristic property of the centres of the circles (I), (Ia),
(Ib), (Ic) (there are no other fixed points under an isogonal transformation
with respect to A ABC).
2°. The Feuerbach point # is an orthopole, with respect to A ABC,
of the straight line OIk that passes through the centre O of the circle (ABC)
and the centre Ik of the corresponding circle tangent to the lines BC, CA>
AB. If the straight line is given by the equation
z — z0 = A (z — z 0),
then its orthopole with respect to A ABC has the affix
/‘ = - | f f i - y + Zo - ^ 0j
(see problem 7). In particular, if the line at hand passes through the coor
dinate origin, then
H = —■Oi — o^)-
The affixes q>k of the Feuerbach points <Pk corresponding to the centres Tk
of the four circles (Tk), each of which is tangent to the lines BC, CA, AB,
will be (see problem 40)
?* = ~ (155>
where zk is the affix of the point Tk (incidentally, it follows from this that
the slope of line OTk is equal to 2k = — and so — = Xk). We will
** **
denote the affixes xk and <pk of the points Tk and 4>k that correspond to
each other by t and cp. Then the relation (155) can be rewritten as
<P = — — (156)
Let us find the affix \j/ of point lF, which is symmetric to point H with respect
to point 0:
+ * _ 1 (n „ f \
2~ ~ " t r “ * 7 j'
whence
ik = - a 3 - . (157)
T
Since |^| = 1, the point W lies on the circle (ABC). Incidentally, this also
follows immediately from the fact that under the homothetic transforma-
tion (H, 2) the Euler circle of A ABC goes into the circle (O) = (ABC),
and since all Feuerbach points lie on the Euler circle, their images TV
*Pl9 ^ 2, under the homothetic transformation (H, 2) lie on the circle
{ABC).
Let us eliminate r and f from the equations (157), (153), (154). We
obtain an equation for finding if/. From the relation (157) we have
f = — ^ T dz
and, hence, the relations (153) and (154) take the form
2r + — (T1 — 0 (158)
—2 ^ t <t3 + <t3 t 2 — dx — 0 (159)
or
<r3 + 2 t — ax = 0, (160)
ovr2 — 2 \j/ x <x3 — = 0. (161)
It remains to write the resultant of these equations as equal to zero. But
this can be done differently: multiply Eq. (161) by —if/2 and add it to
Eq. (160) to get
(2^3 az + 2) t = (T1 —
whence
= ~ 1
2ip* <r3 + 2
Substituting this value of t , for example, into equation (161), we obtain
_ G\ — 2<71(X1^ 2 + C73^4
2^<t3 gi - — (Tj — 0
** 4(^«r3 + l)^ 2 + 1)
or
( ^ 1 — 4(T1ff3) \p4 — 4ff16:3^ 3 — 2cr1a 1^3^2 — 4<T1a 3i/' + — 4^ - 0
or, cancelling (j3,
(<rf — 4(71d:3)^4 — 4<71i/f3 — 2(71d:1^ 2 — 4(7^ + a2 — 4<t1(73 =- 0
or
(<t2 — 4<t2) 1/f4 — 4<r1i/f3 — 2 — 4a1ij/ + cr2 — 4<72 = 0.
Note that the coefficients of this equation that are equidistant from the
ends are conjugate in pairs: the first and the last, the second and the fourth
(an antireciprocal equation).
Problem 42. Let M be a variable point of a circle (O) of radius R circum
scribed about ABC\ P is a point symmetric to the orthocentre H of
A ABC with respect to the diameter of (O) parallel to the Simson line
for the point M with respect to A ABC; A', B \ C' are points symmetric
to M about the straight lines OA, OB, OC. Prove that the point Q, which
is symmetric with the orthocentre //' of A A'B'C' with respect to point P
describes a circle concentric with (O) and having a radius OQ equal to
O Q = ^ O I O I a-Orh O!c,
where / is the centre of the circle inscribed in A ABC, and h , h , lc are

the centres of the circles escribed in that triangle (Fig. 43).
Solution. Take (O) as the unit circle and let zly z2, z3, z0 be the respective
affixes of the points A, B, C, M. The equation of the straight line AB is
z + Z]Z2z = zx + z2. (162)
The equation of the perpendicular dropped from point M to line AB is
z — z0 = ZiZ2 (z — z 0)
or
z — ZjZ2z = z0 — ZjZ2z0. (163)
Adding the equations (162) and (163) termwise, we find the affix mz of
the projection Af3 of point M on line AB:
Similarly, the affix m 2 of the projection M 2 of point M on line AC:
m 2 = — (z0 + Z 1 + Z 3 — ZC3 Z0 )
2
The slope of the straight line M 2 M 3 is
m3 — m2 _ z2 — z 3 — ZjZpjzz — z3)
m3 — m2 z 2 — z 3 — z Yz3( z 2 — z 3)
Zq — Z1
(z2 - z3)
(z2 — z3)( 1 - Z3Z o)
(z2 - z 3) ( l - Z 1Z0) ZZ Z2 Z\ Zq ZQ
Z2 Z0 z1
and so the equation of the diameter of the circle parallel to the Simson
line of point M with respect to A ABC is of the form
z = - 3- z (164)
z0
or
ZqZ — gzz = 0. (165)
Incidentally, this equation could have been written at once by proceeding
from the equation of the Simson line and dropping the absolute term of
the equation [problem 3, equation (1)].
The equation of the perpendicular dropped from point H to the straight
line (165) is of the form
<r3 -x (166)
z ~ *i = -------(z — ffi).
zo
Adding the equations (164) and (166) term by term, we obtain
<7l<73
2 z — a1 = 9
Zo
whence
z= <r2
+
2 zo }
This is the affix of the projection of point I f on the straight line (164). The
affix p of point P is found from the relation
= X
2
whence
p — o-2z0.
Furthermore, the equation of line OA is
z = z\z or z — z\z — 0.
The equation of the straight line passing through point M perpendicular
to line OA is
z — Z0 = —z\ (z — z„).
Solving this equation together with the equation zz = 1 of the unit circle
(ABC), we obtain
One of the roots of this equation is naturally z = z0 (the affix of point M)

the other is
z — a' = zfz 0,
which is the affix of point A'. In similar fashion we find the affixes b' and
c' of points & and C':
b ' — zpo, c' = ziz0,
and, hence, the affix b! of the orthocentre H ' of A A'B'C':
h ' = (zf + z l + z|)z0.
The affix q of point Q, which is symmetric to point H ’ with respect to point
P, is found from the relation
q + h'
whence
q = 2p — h' — 2 <r2z 0 — (z\ + z\ + z§)z0
= z 0 [2<r2 — (zx + z 2 + z3)2 + 2(Z2Z3 + z3zt + zxz2)] = z 0(2<r2—a\+ 2a^
= z 0 (4<r2 - «rf).
Thus,
q = (4<t2 — a l)z0.
From this it follows that if point M with affix z0 describes a circle (ABC),
then point Q describes a circle (12), in the opposite direction, that is con-
centric with {ABC). The radius p of (£2) is

P = I 4<t2 — a l\.
The affixes of the points I, /„, Ib, Jc are:

T0 = — K Z 2 Y Z2 — 1 /Z3 f z l — ] f Zl ] ^ 2 = + *0 + Ah
Tl = — ]fz2 Kz3 + Zz3 Kzi + Kzi Vz2 = a0 — b0 — c0,

T2 = /z2 KZ3 —Kz3 + )fz1 KZ2 = —Oo + * 0 —Co,
T3 = f z 2 f z 3 + ^ Z 3 K Z1 — f z l Kz 2 = — Oo — ^ 0 + C0 . .
From this we have

T0TiT2T3 = [eg - (a0 + Z»o)2] [eg - (a0 - 60)2]
= (eg — ag — bl — 2a0b0) (eg — al — bl + 2a0b0)
= (eg - flg - b l f - Aalbl = (a% + b%- eg)2 - 4agig
= (flo + ô + eg)2 — 4 (fegcg + eg ag + eg&g)
= (zjz2 + z2z3 + z3Zi)2 — 4 (zgz^ + z 2z 2z 3 + 'z|z3Zj) = af — 4<r1a3
(see also problem 41). And so we get
ToTiT2t 3 = al — 4a,a3 — - - - 4y = 4° 2
a\ a\ ai
and, thus,
| V i T 2 t 3| = I to T ^ jT a j = |a \ - 4<t2| = p
or
P= | t (,1 | t ,| ;t 2| [ t j |
or
OQ = O IO Ia OIb OIc.
If the affixes of the points A 9 B9 C are taken in the form Rzl9 Rz29 Rz29
where \z±\ = \z2\ = |z3| = 1, then we obtain
OQ = 0 1 'O h '< > h 'O h
^ R3
Problem 43. The points A l9 A29 AZ9 A4 lie on a single circle (O) =
= (A1A2A3A4) which is taken as the unit circle. Denote by ipl9 (p29<p3, ty* the
affixes of the Feuerbach points <Pl9 tf>2, ^>3, ^4 that lie on the circles in
scribed in the triangles A2A3A49 AXA3A4, A XA2A49 AXA2A3. Prove that the
midpoints of the segments Ax<Pl9 A2<P29 A3<P3. A4<P4 lie on one circle.
Find the affix of the centre of this circle and its radius.
Solution. Let zl9 z2, z3, z4 be the respective affixes of the points Alf A2,
A3, A4. Then the affix <p1 of the Feuerbach point <P4 of the circle inscribed
in A A2A3A4 is computed from the formula (see problem 32, item 6°)
ZoZ3 | z2z4 ~b z*>z4
(Pi = ----------------- "—
Z2 + Z3 + Z4
and so the affix pL of the midpoint of segment Al<Pl is
__ <P i ~ h Z l z l z 2 4 ~ Z 1Z 3 ~ h Z l Z 4 ~t~ Z 2Z 9 Z2Z4 ~t~ Z3~4 &2
1 2 2(z2 + z3 + z4) 2(o1 —zty

where
er2 — z i z 2 + Z 1Z 3 + Zl Z\ + Z 2Z Z + Z ZZ \ + Z 3Z t*
(Tl = Z4 ~j~ z2 ~h z3 ~h z4-

We obtain similar expressions for the affixes jx2, /t3, j.i4 of the midpoints
of the segments A2<P2>A3<P3, A4<P4:
g 2 g 2 g 2
Pi ~ -------------- 5 Pz -- » Pt -- •
2 (g 4 — z 2) 2 {g 1 — z 3) 2 ( g 4 — z 4)
From this it follows that the points pk (k= 1, 2, 3, 4) lie on the circle ((2),
the equation of which is
u — ------------------- ,
2 g 4 — 2z
where z describes the unit circle.
Performing an inversion of this circle with the circle of inversion (0),
we obtain the circle
1 2 — 2z &x 2
v= — -------= 2 z.
u 2 2 2 g g g
This is the circle (0'), the affix of whose centre is 2 g J g 2 and the radius is
equal to 2/|<t2|. The circle (G) may be obtained from (0') by the same inver
sion and also by a similarity transformation. From this we can find the
affix of the centre of (G) and its radius.
But we can also take advantage of the formulas of problem 12 that
yield the affix of the centre,
bd — ac
(o = — —
dd — cc
and the radius
be — ad
of (Q) into which the unit circle (O) goes under the linear-fraction trans
formation
u = — Z tL t a d - b e * o. (167)
cz + d
Let us consider the transformation
(168)
20*! — 2z
Here [compare with (167)], a = 0, b — er2, c = — 2, d = 2a1 and so the
affix co of the centre of (£) on which the midpoints of the segments
Ai&i, A 2 & 2 9 A3*P39 A4<P4 lie is
co — 20-201 &2&1
4(01*1 ~ 1) 2(* 1 *1 - 1)
and the radius is
— 2at i ___*2
4axCi — 4 ~ ~2 *1*1 ~ 1
Problem 44. Let Bx and Cx be the points of intersection of the bisectors
of the interior angles B and C of A ABC with the circle (ABC) = (O).
Consider the sum of the directed line segments OBx + OCx — ON. Prove
that IN JL BC, where / is the centre of the circle inscribed in A ABC.
Also prove that the segment IN is equal to the radius of (ABC) = (O).
Solution. Take (O) as the unit circle. Let zl9 z2, z3 be the respective
affixes of the points A9B, C. The affixes al9 bl9 cx of the points A l9 Bl9 Cx
of intersection of the circle (ABC) and the bisectors of the interior angles
of A ABC are given by the relations
01 = — ] f Z2 V Z3, *1 = — V z 3 ) f ZU C = — fr , ]fz2 ,
where we take for the roots ]fzlt ]Jz2, J[z3 values such that ^ z 3\<.
< 1. The affix n of point N is
n = - ]fz3 ]jzl - /z j )Jz,,
and since the affix of point I is equal to
T0 = - jfz2 Vz3 - ]}z3 1(7X - fz j )[z2,
it follows that the directed line segment IN is associated with the complex
number ][z%]fz3, whence IN = \fz~3 ][z3\ = 1 = R. The slope of the
straight line IN is
The slope of line BC is equal to x ’ — — z2z3 and so % + x ’ 0, and

therefore the straight lines BC and IN are mutually perpendicular .
Problem 45. The angles of A ABC

form a geometric progression with ratio
2. Prove that the midpoints of its sides and
of the feet of the altitudes are six vertices
of a regular heptagon (Fig. 44).
Solution. Take (ABC) for the unit
circle and assign to point A (the vertex
of the smallest angle of the triangle) an
affix of 1. Then, assuming
a = cos(27t/7) + i sin(2;r/7),
we conclude that 1, a, a2, a3, a4, a5, a6 are
the affixes of the vertices of the regular
heptagon inscribed in the circle (ABC).
Under the conditions of the problem, the Fig. 44
2k
angles of the triangle are equal to A, 2A and 4A, whence A = ---- ; therefore
7
the affixes of the points B and C are equal, respectively, to a4 and a5.
Let Al9 Bly Cx be the respective midpoints of the sides BC, CA, AB,
and let al9 bl9 cL be the affixes of these points. Then
a4 + a5 a5 + 1 1 +a4
h =
The slope of line BC is equal to
* bc — — a4a5 = — a9 — r/2
(since a7 = 1) and, hence, the equation of the straight line passing through
point A perpendicularly to BC is
z — 1 = ct2(z — 1).
The equation of BC is
z— - a1
nr4 = - a2(z - a4).
Solving the system of equations
z — 1 = a2(z — 1)
z — a4 — — a2(z — a4),
we find
2z — 1 — a4 = — a2 + a2 = — a2 + a5,
whence
1 — a2 + a4 + a5
z = a2 = -----------------------,
which is the affix of the io o tA 2 of the altitude dropped from vertex A

to side BC. Then
1 —a5
xAC ^ = — a5.
A 1 - a5
The equation of AC is
z — 1 “ — a5(z — 1).
The equation of the altitude dropped from vertex B to side AC is
z — ola = <x5(z — a4).
From the last two equations we find
2z — 1 — a4 = a5 — a,
whence
, 1 — a + a4 + a5
2
And
1 - a4
AB
1 - a4
The equation of AB is
z — 1 — a4(z - 1).
The equation of the altitude dropped from vertex C to side AB is
z —a5 — a4(z — a5).
From the last two equations we find the affix c2 of the projection of point C
on line AB:
2 c2 — 1 — a5 -- a4 — a = a4 — a6,
whence
1 + a4 + a5 — a6
c2 = •
Consider the affix o9 of point Og obtained from point O under the homo-
thetic transformation ^<7,----—j, that is, the affix of the centre 0 9 of the
Euler circle of A ABC. Since the affix h of the orthocentre of A ABC
is equal to 1 + a4 + a5 and 0 9 is the midpoint of OH, it follows that
1 + a4 + a6
2
We now find
1 — a2 + a4 + a5 1 + a4 + a5
Oo oti
2 2 2
1 — a + a4 + a5 1 -f a4 + a5 a
bo On ---
2 2 2’
a4 + a5 1 + a4 + a5 1
0\ o9 —
2 2 T ’
ftC1
1 + a4 + a5 — a6 1 + a4 a°
+
c2 O9 —
2 2 2
1 + a4 1 + a4 + a5 _ a5
Cl On ---
2 2 T ’
1 + a5 1 4- a4 + a5 a4
b\ o9 —
2 2 2'
These numbers form a geometric progression with ratio 1/a; and the moduli
of all these differences are equal to — . This means that each of the directed
2
line segments
C9A2, 0 9B2, C?9A i , O9 C2 , 0 9Ci, 0 9B i ,
beginning with the second, is obtained from the preceding one by a rotation
in one and the same direction through an angle of 2n/7, that is, A2, B2r
A 1, C2, Clf 2?x are the six successive vertices of a regular heptagon inscribed
in the Euler circle of A ABC.
Sec. 2. Problems with hints and answers

1. The angles A, B, C of a triangle are connected by the relation C =
= 3B = 9A ; AA \ BB \ C C are the altitudes of the triangle, H is its ortho-
centre, / is the centre of an inscribed circle; r is its radius, (O) = {ABC)
is a circle circumscribed about A ABC ; R is its radius, and a>b, c are
the lengths of the sides BC, AC, AB. Prove that:
1°. H A' + HB' - H C = R/2.
2°. OP + OH2 - 5R2.
3°. be + ca + ab — R2^ 13.
4°. cos A + cos B f cos C = (1 + V13)/4.
5°. 0/* = J l* (5 -/l3 )/2 .
6°. r - if(]/f3 - 3)/4.
7°. OH2 - * 2(1 - 8 cos ^ cos 5 cos C) - /*2(5 + / 13)/2.
2 54 Problems in Geometry
8°. cos A cos 2? cos C = — (3 + /l3)/16.

9°. sin A sin flsinC = K(13 - 3 ^l3)/128.
10°. sin A + sin B + sin C = /(13 + 3^ 13)/8.
Solution. The points A, B, C may be regarded as the respective vertices
of a regular 26-gon inscribed in a circle of radius R y since A = nj13,
B = 37r/13, C = 9tt/13 (Fig. 45). First note that for any natural number,
cos(tt/2/13) = cos((26 — n) tt/13) = — cos((13 — n) tt/13)
and, furthermore.
i . 2tt , 4tt , 6tc , 8tt , IOtt 12tt
1 + COS-------bCOS--------bCOS----- + cos — - + COS + COS
13 13 13 13 IT ~13~
, 14tc , 16k 18tc , 20n , 227r , 24rc
+ cos--------b cos--------b cos,------- b cos--------b cos--------b cos----
13 13 13 13 13 13
2tt 4k 6k , 87T IOtt ,127c
cos---- + C O S -------- b C O S -------b C O S ------ b cos--------bC O S ------
13 13 13 13 13 13
147T , 167T , 1Sk 20k 22k
= COS --------bC O S ---------bC O S -------- b COS bC O S ------
13 13 13 13 13
24tc
+ cos = ~ 1/2.
13
Set
2k 6k 8K
X = c o s ----- + cos —
+ COS
13 13 13
(169)
4 k 107T 1271
y == c o s ----- + cos —
+ COS -----.
13 13 13
Then
cos A + cos B + cos C = —y.
Proceeding from the identities
2 cos2a = 1 + cos 2a, 2 cos a cos p = cos(a + fi) + cos(a 0,
*we obtain the equation
4x2 + 2x — 3 = 0,
whence
- 1 + /1 3
4 y = 4
and so
This can also be obtained from (169).

1°. ABC is an obtuse-angled triangle (C = 9tt/1 3), hence,
HA' = |2J? cos B cos C\ = — 2R cos B cos C
— — R[cos(B — C) + cos(fi + C)],
HB' = — 2?[cos(C — A) + cos(C + A)],
HC' = /?[cos(^ - B) + cos(A + £)].
From this we obtain
HA' + HB' - HC' = R ( - x - y ) = Rj2.
2°. OP + OH2 = R2 - 2Rr + 9R2 - a2 - b2 - c2,
__ . A . B . C
2Rr — 8R2sin — sin — sin —
2 2 2
= 2R2( - 1 + cos A + cos B + cos C) = 2R2( - 1 - y)T
a2 + b2 + c2 = 4 R2(sin2A + sirrfi + sin2C)

= 4/?2(3 - cos*A - cos3B - cos2C) = 2R\3 - x).
and, thus.
OP + OH2 = /{2(6 + 2jc + 2y) = 5/?2.
3°. fee + ca + ab = 4/?2(sin B sin C + sin C sin A + sin A sin B)
= 2rt2[cos(fi - C) - cos(fl + C) + cos(C - 4) - cos(C + A)
+ cos(/4 - B) - cos(/J + B)\ = 2/?2(x - >>) = IP /H .
The derivation of the other relations is left to the reader. In deriving
them, take advantage of the basic formulas relating to trigonometry and
the geometry of any triangle; also take advantage of the specific nature
of the given triangle (A = 7t/13, B = 37r/13, C ^ 97r/13).
It is also possible to introduce a complex system of coordinates by
26 __
taking {ABC) as the unit circle. Then all the 26 values of Î will constitute
the affixes of the vertices of the regular 26-gon mentioned earlier. These
affixes are
ak ^ cos-------b i sin----- (k ~ 0, 1, 2, 3 ,..., 23, 24, 25).

13 13
We can take point A as the unit point a = a0 — 1. The points B and C
will have the affixes
f 871 . . 8tt 6n , . . 6n
b — ag = cos-------b i sin , c = a 6^cos - ---- b i sm ----
13 13 13 13
(compare this with problem 45 above).
2. A quadrangle A BCD is inscribed in a circle (O). Perpendiculars AA2
and BB2 are dropped on CD from points A and B; perpendiculars BBX
and CCl are dropped on DA from points B and C; perpendiculars CC2
and DD2 are dropped on AB from points C and D; and, finally, from
points D and A we drop perpendiculars DDt and AAX on BC.
Prove that (see Fig. 46):
1°. The segments ALA2, B±B2, CXC2, DXD2 are equal and lie on straight
lines passing through point F, which is symmetric to point O with
respect to the centroid of the system of points A, B, C, D.
2°. The quadrangles AlB1C1D1and A2B2C2D2are similar to the quadrangle
ABCD and are inscribed in circles with centre F (I. Langer).
Hint. 1°. Take (O) = {ABCD) for the unit circle. Let a ,b ,c , d be the
respective affixes of the points A, B,C, D. The affix of point F is
f — Oi/2 {ox = a + b + c + d).
F ig . 4 6
The affixes of the projections A1 and A2 of point A on the lines BC and CD

are
+b + c — a ,-\{ a + c + d -
whence
fli — a2 = (a — c)(b — d), f — a, = ——{ad + be).

2a 2a
Furthermore,
— a2 _ / — _ bed
<*l — a* f — <*i a
so that the line ALA2 passes through point F. Besides,
AxA2 =--- I ax — a* I = --- ^C* BD

2
and similarly for CXC2,
2°. We have
—/ ------1— (ad + be), a2 = f — - (ab + cd).

2a 2a
From this (and from similar relations for bly b2y cly e2y dly d2) it follows
that the quadrangles >11J61C1D1 and A2B2C2D> are inscribed in a circle
----L>
with centre F(f). Suppose the quadrangle ABCD is convex. The transfor-
17 - 81
mation 2 ' = / — —- (ad+bc) 2 carries quadrangle ABCD into the quadr-

1
angle A ^ C ^ . The transformation 2 " = / ------ (ab + cd) 2 carries the
2
----- > V
quadrangle ABCD into the quadrangle A2B2C2D2. And so the quadrangles
AXBXC1D1, A2B2C2D2 are convex, have an orientation opposite that of
quadrangle ABCD, and all quadrangles ABCD, A^C -^D ^ A 2B2C2D2
are similar.
The radii of the circles in which the quadrangles A1B1C1Dl and A2B2C2D2
are inscribed are equal, respectively, to
— | ad + be | and — \ab + cd\

2 2
or
R\coz(AD, BC) | and R\ cos (AB, CD) | (R = 1).
Indeed, from the relations
_
——1 I/ a _L
+ b0A _L
+ b C1,\
+ cc ----—
2 V a j
we find
a i-d ! = (a - d) (ad + be)

2ad
AXDX= — AD | ad + be |,
2
whence
- | ad + bc\ = — (cos 04a £C)|.

2 y4Z)
Remark. The affix a3 of the projection A3 of point A on BD is
1
a3 — f~ {ac + bd).
2a
Hence,
— (ax + a2 + a3) - f ----- -,

3 6a
where a2 = ab + ac + ad + be + b d + cd. From this it follows that if

the quadrangle Q (that is, ABCD) is inscribed in the circle (O), then the
centroids of the projections of its vertices on the sides of the triangle, formed
by the other three vertices of Q, form a quadrangle Q' similar to Q but
with opposite orientation. Here, the quadrangle Q' is inscribed in a circle
whose centre is symmetric to the point O with respect to the centroid of
the quadrangle Q. The radius of this circle is equal to the distance from
the centroid of the quadrangle Q to the centroid of the six points at which
the circle (O) intersects the straight lines passing through some point of (O)
parallel to the sides and diagonals of the quadrangle Q (the solution of
Gourmagschieg).
3. Straight lines AP, BP, CP drawn from an arbitrary point P cut
the sides BC, C A , AB of A ABC in points A,, B„ C, and divide these sides
in the ratios:
v.
AjC B,A C,B
Find the ratios
CP
x = z=
PC,
Hint. Apply the Menelaus theorem to A AA,B and to the transver
sal CC,.
Answer, x = (1 + /I) v, y — (1 + /0 A, z = (1 + v) \i. Complex num
bers can also be used.
4. A triangle A,B,CXis inscribed in A ABC (that is, point A, lies on line
BC, point B! on line CA and point C, on line AB).
1°. Prove that A ABC and l ^ A ^ C , have a common centroid if and
only if the ratios
, Ta x Cfit AC,
X = ---------II = ------------------- V = ------------------- ----
A1C BXA C,B

are equal: A = ji = v.
2°. Assuming that A ABC and A A ,B ±C, have a common centroid
(A = n — v), prove that
A,B,C, A2 - A + 1
(A + l ) 2
ABC
3°. Suppose A A 2B2C2 is circumscribed about A ABC and has a common
centroid with A ABC. Prove that the ratio A', in which the points A, B, C
divide the directed line segments B2C2, C2A2, A2B2, is then equal to 1//,,
that is, the inverse (in magnitude) ratio in which the points Al9 Bl9 C1
divide BC, CA, AB. Find
A1BiC1
A2B2C2
and prove that
(ABC)2 = (A & C J (A2B2C2)
(R. Deaux).
5. The triangles ABC and A1B1C1 are homothetic under the homothetic
transformation with centre S and ratio
We know that A A'B'C' is inscribed in A ABC and circumscribed about

A îîQ . Prove that we then have
(A'B'C')2 = (ABC) ( A ^ C j)
(PilattPs theorem).
6. Given a triangle ABC. Its sides are oriented as follows: BC, CA, AB.
Prove that on the axes BC,CA, AB there is only one triplet of points
AUBX, Cx such that (BA2) ^ (CBX) = (AC2) and the straight lines AAl9
BBl9 CCX intersect in one point. Also prove that the points Al9 Bx, Ct
are the respective interior points of the segments BC, CA, AB.
Hint. Use the Ceva theorem. Investigate the resulting cubic equation
[(BA,) = (CBj) - (ACj) = x] for x (S. Vatricant).
7. Let I9Ia9Ih9Ic be the centre of the circle (/) inscribed in A ABC,
and the centres of the circles (Ia), (Ib), (7C) escribed in the triangle ABC.
Let R be the radius of the circle (O) = (ABC), and let A be a point
located on segment HHXat a distance of (2/3) HHt from H\ //is the ortho
centre of l\ABC , and H1 is the orthocentre of A îîC i, whose vertices
are the feet of the altitudes of A ABC. Denote by L the centroid of a
system of four points obtained from the points /, Ia, Jb9 Ic by inversion
with respect to the circle (O). Prove that the points O, N , L lie on one
straight line and that
OP OIa•OIb•OIc - 3R2•ON,
O N O L = ~ - R 2.
3
8. A triangle ABC is inscribed in a circle (O). Construct right-angled
------ ^ > ------ >
triangles AO A', BOB', COC' (in all three triangles, the angle O is equal
Fig. 47
to n/2) with the same orientation. Let the circles (A', A'A), (B',B'B),
(C', C'C) cut the sides AB and AC, BC and BA, CA and CB in the points
Ba, Ca; Cb,A b; AC,B C.
Prove that the triangles ABaCa, BCbAb, CAcBc have the same Euler
circle (Fig. 47) (R. Blanchard).
Hint. Take (O) as the unit circle. Let a, b, c, a', b \ c' be the affixes of
the points A, B,C, A', B', C . Then
a! = ia, b 'îb , c'=ic, AA' = ]f2.
The equation of the circle (A', A'A) is
(z — ia) (z + ia) — 2. (170)
The equation of AB is
z + abz = a + b,
whence
z + abz = a + b,
z = a + b — abz.
(z — ia) (a + b — abz + ia) — 2
or
(z — ia) (abz — a — b — ia) + 2 = 0.
One of the roots of this equation is z = a, the other is

ba = b + i(a + b).
Similarly, the affix ca of point Ca is
ca = c + i(a + c).
The affix ha of the orthocentre Ha of A ABaCa is found from the relation
OHa = OA' + A'Bm+ A'Ca + A 1A.

That is,
ha = ia + b + i(a + b) — ia + c + i(a + c) — ia -f- a —ia=al( 1+ i)—ai.
The affix a9 of the midpoint of HaA', that is, of the centre of the Euler
circle of AABaCa, is
(7i(l + i) — ai + ai 1+ i
(— bg — C9).
9. Given the affixes z1$z2,z 3 of the vertices AyB ,C of is.ABC. Find-

the affix z0 of the centre of the circle (ABC).
Answer * z0 = Zl J l ^ 3 ^ + Zg ~ + 73 ~ *i)
*1 z i 1
z2 z 2 1
Z3 Z 3 1
Hint. (z0 — z{) (z0 — z x) = (z0 — z2) (z 0 — z 2) = (z 0 — z3) (z 0 — z 3).

10. Let Al9 Bl9 Cx be points symmetric to the centre O of (O) = (ABC)
circumscribed about IsABC with respect to the sides j5C, CAf AB respec
tively. Prove that ISABC and lsA 1B1C1are symmetric about the midpoint 0 9
of segment OH, where H is the orthocentre of ISABC (Hamilton's
theorem).
Hint. Take (ABC) for the unit circle. Let zx, z2, z3 be the affixes of point
A, B, C. Then the affixes of the points Al9 Bl9 Cx will be z2 + z3, z3 + zl9
Zjl + z2. The affixes of the midpoints of segments AAlf BBX and CCX are
equal to (zx + z2 + z3)/2 = a j l , and this is the affix of the midpoint
of segment OH.
11. Given in the plane a similarity transformation consisting of a rotation
about a point O through angle a ^ nn (n an integer) and of homothetic
transformations with centre O and ratio k. Under this similarity trans
formation, let IsABC go into IsA'B'C'. Denote by Ax, Bl9 Cx the respective
points of intersection of the straight lines BC and B C , CA and C A ',
AB and A'B'. What must the position of O be with respect to A ABC
------- > -------->-
so that A ABC and A A 1B1C1 are similar and have the same orientation?
Hint. Let a ,b ,c be the affixes of the points A ,B ,C in an arbitrary

system of coordinates with origin at point O. Setting p = &(cos a +
+ i sin a), we find the affixes a' = pa, b' = /?&, c' = pc of points A \ B', C'.
From the equations of the lines BC and B'C',
(b — c) z — (b — c)~z + b~c — be = 0,
p(b — c) z —p(b — c) ~z + pp(b>c — — 0,
we find the affix z = ax of point Ax
be be p(p — 1)
a1 = -
b —c p —p
and similar expressions for b1 and cx:
^ ca — ca p(p — 1) ab — ab p(p —1)
~c — a p ~ P a —b p —P
The oriented triangles ABC and AXBiQ are similar and have the same
orientation if and only if
a at 1
b bx 1 = 0 .
c 1
Substituting into this equation the expressions for al9 bl9 cx in terms of
a, b9c, p and simplifying, we obtain
aa(b — c) + bb(c — a) + cc(a — b) = 0,
and this means (see the preceding problem) that O is the centre of the
circle (ABC).
Remark. If we now take (O) = (ABC) for the unit circle, then
p (p - i)
ax — (b + c),
p - p
P(P - - (c + a).
bx
P- P
(a + b).
P- P
and, hence, is the image of A A2B2C2 (which “ supplements”
A ABC) under the similarity transformation with centre O, angle of rotation
a rg fc ^
and proportionality factor

][k2 — 2k cos a + 1
2 P(P ~ 1)
P- P | sin a |
where k = |p (R. Blanchard).
12. Prove that if in a triangle ABC the median ma issuing from angle A
is equal to the length la of the bisector of the same angle A, ma = la, then
A ABC is an isosceles triangle and 6 = c. Solve the problem analytically
(by computation) and geometrically.
13. A1A2A9A4ASA6A7 is a regular convex heptagon. Prove that the circle
circumscribed about the triangle formed by the lines AXA2, A3A5, A4A7
passes through the vertex A6 (V. Thebault).
14. Prove that if the angles of A ABC are
n 4n
A= C=
79 7
and if a, b, c denote the lengths of the sides BC, CA, AB, then the following
relations hold true:
1°. a2 + 62 + c 2 = 7R2.
2°. OH = R)[2 (see problem 11).
3°. The angles of A A'B'C', whose vertices are the feet A \ B \ C’ of
the altitudes of the given triangle, are
A C, = A.
4°. cos A = — , cos B = —, cos C = ----—= ----- —.

2a 2b 2b 2c
5°. be = ^(6 + c)y be = c2 — a2, ac = b2 — a2 and, conversely, if the
first two relations hold, then /I = n il, B = 2njl, C = 4tt/7.
6°. cos ^4 cos 5 cos C = —1/8, sin /f sin 5 sin C = 1/7/8.
• 2 ^^ = 3--------
0 —c 36 a . 3c + 6
sm2 sin2 5 = ------, sin2 C — -----
4a 46 4c
8°. sin2 A sin2 B sin2 C + cos2 A cos2 B cos2 C — 1/8.
9°. cos2 A + cos2 B + cos2 C = 5/4.
10.° sin2 A + sin2 B + sin2 C = 7/4.
11°. cos A + cos B + cos C = - — ----- ~ ~-----
a 2 2(c —a)
12°. cos2 B cos2 C + cos2 C cos2 A + cos2 A cos2 B = 3/8.
13°. tan A tan B tan C = — Yl.
14°. ha = hb + h c.
15°. hi + h\ + hi = (a2 + 62 + c2)/2.
16°. BA’ A’C = CB’- B 'A = ab-, AB' C'B
4 4 4
17°. If Qa, Qb, Qc are the feet of the bisectors of the interior angles of
A ABC, and Q'a, Qb, Q'c are the feet of the bisectors of the exterior angles
of the triangle, then
b2 a2 be ac
AQa
1 QO a
f
/~%2
. ^L \/a T
i
be
I
D /^ 2 i
T
BQb
r * r \ 2 ___
a+ c
•
= C — a, CQc =
a+ b
= b — a.
19°. BQb CQc = (c — a) (b — a) — a \

20°. AQ'a = b + c.
21°. BQ'b2 = a2 - a b + b2.
22°. CQ'2 = b2 + be + c2.
23°. AQ'2 + BQ'b2 + CQ'2 = 4b2 + 4c2 - 2a2 = 8m2a.
24°. The radii of circles of Apollonius are equal to the sides of A ABC
(the circles of Apollonius of the triangle are (QaQ’a)> (QbQb)> (QcQ'c) cons
tructed on the segments QaQ'a, QbQ'b>QCQC as diameters).
Let A tBtCt — Tt be the triangle formed by tangents at points A, B ,C
to the circle (ABC) circumscribed about A ABC = T. Prove that:
------> ------>
25°. The triangles ABC and A tBtCt are similar and have opposite orien
tations; the proportionality factor is equal to 1/2.
26°. The altitudes A A \ BB', C C of A ABC bisect the sides B tCt, CtA t,
A tBt of AT,.
27°. The Euler circle of AT, passes through the orthocentre of AT.
------> *-----^
28°. The triangles IIbIc and ABC are similar, have the same orientation,
and the proportionality factor is equal to 1/2.
15. Given a triangle ABC, G is the point of intersection of the medians.
Let Au Bl9 Cx be the images of the points A, B ,C under the homothetic
transformation (G, — 2). Let
BCA', CAR, A B C (171)
be equilateral triangles having the same orientation; let
B C T ', CAB'', A B C ' (172)

also be equilateral triangles having the same orientation; note that the
orientation of any one of the triangles (171) is opposite that of any one
of the triangles (172). Prove that the triangles
B 'C A l9 C A 'B 19 A'B'Cx (173)

are equilateral and have the same orientation as that of the triangles (172),
and the triangles
B " C " \, C"A"Bl9 A"B"C x (174)
are equilateral and have the same orientation as that of the triangles (171)
(Eig. 48).
3 _____
Hint. If a is any one of the imaginary values of Y —I, then the affixes
of the points A \ B \ C', A ", B '\ C" are expressed in terms of the affixes
of the points A, B ,C by the relations
af = etc + a by
b' = cca + ac,
c' = CLb + a a;
a" = a c + a6,
b" — aia + ac,
c" = a6 + acr.
The affixes of the third vertices of the equilateral triangles constructed
------>-
on the segments B’C' and i?"C" and with orientation the same as ABC A"
------>
and ABC A', respectively, are
ac' + cub' = b + c — a
ac" + ab" = b + c — a
and, hence, these third vertices coincide with the point A1 (R. Deaux).
16. Prove that if the angles A, B ,C in A ABC are connected by the
relation
sin A = cos I? tan C,
then the altitude A A \ the median BB', and the bisector CC' intersect
in one point.
Hint. The angles B and C are acute. Hence, the foot A' of the altitude
AA' is an interior point of segment BC and
BA' c cos B
b cos C
A'C
From there on use Ceva’s theorem.
Supplementary question. What theorem results from what has been
proved if we apply an isogonal transformation to point M in which the
altitude A A \ the median BB', and the bisector CC' intersect? (after Morley).
Hint. Since, under an isogonal transformation, the centre O of {ABC)
goes into the orthocentre H of Is ABC (and conversely), we get the following
theorem: if angles A, B, C in Is ABC are connected by the relation sin A —
= cosf?tanC, then the diameter AO of the circle (O) = {ABC), the
bisector CC', and the cimedian BB* (the straight line symmetric to the
median BB' with respect to the bisector of the interior angle B) intersect
in one point.
17. Inscribed in the unit circle is a triangle ABC, the affixes of whose
vertices are respectively equal to zl9 z2, z3. Given a point P whose affix is
p =
azj + pz2 + yz2
--------------------------------- ,
a + P+ y
where a, p, y are real numbers. Let Al9 Bl9 Cx be points in which the straight
lines AP, BP, CP intersect the sides BC, CA, AB respectively. Find
(M C i)
{ABC)
Hint. The affixes of the points Al9 Bl9 Cx are
fiz2 + yz3 t yzs + a z j a z x + 0 z2
a, — ---------------, ol , cx = -------------- .
P + y y + « <* + P
Answer.-----------= -------------------------------.
{ABC) {P + y){y + * ) { x + P )
18. Prove that a necessary and sufficient condition for the orthogonal
projections Bx and Cx of the vertices B and C of A ABC on the sides
AC and AB respectively to lie on the same straight line as the centroid G
of A ABC is the equality 3a2 = b2 + c2 or a2 = b2 + c2, where a, b9c
arc the lengths of the sides BC9CA9AB of A ABC. _
Hint. Take {ABC) as the unit circle. Prove the equations z2z3 + z3z2=
2 — a2 and so forth, where zl9 z2, z3 are the affixes of the vertices of
the triangle, and a9b9c are the lengths of its sides.
19. Let zl9 z 2, z 3 be the affixes of the vertices of A ABC inscribed in the
unit circle. Prove that the equation z\ = z 2z 3 is a necessary and sufficient
condition for A ABC to be an isosceles triangle: AC — AB.
20. ABC is a right-angled triangle (C = n/2); (/) is a circle inscribed
in the triangle. Let Al9 Bl9 Cx be the orthocentres of the triangles IBC,
ICA, IAB. Prove that the length of the projection A$B$ of segment A1Bl
on the hypotenuse AB is equal to the diameter of (/). The length of the
projection B*C* of segment BXCX on leg BC is equal to the diameter of
the circle (Ih) escribed in angle B, and the length A£C?, the projection
of segment AXCXon leg AC, is equal to the diameter of the circle (Ia) escribed
in angle A.
Hint. Take (/) for the unit circle. Let P9Q, R be the points of tangency
of (/) with the sides BC, CA, AB, and let their affixes be 1, /, a respectively.
Then the affixes of the points A, B ,C are
2 2a/ , 2 2a
a = z------= -------- , b - ------== -------- ,
a + i a + / 1+ a 1+ a
2
c= - 1+ /
1+ 7
since the points A 9B9C are images, under inversion with respect to circle (/),
of the points A \ B \ C (the points of intersection of the straight lines IA y
IB, IC with lines RQ, PR, PQ, respectively) (Fig. 49). Now, the affix ax of
the orthocentre Ax of A IBC is ax = —----------- and so forth the
1+a
calculations are rather unwieldy).
21. ABC is a right-angled triangle (C = n/2). The circle {ABC) ={0)
is taken as the unit circle. The point M(z0) describes this circle. Let Q
be the midpoint of segment OM and let / be the Simson line for point M,
which line is constructed with respect to A ABC. Find the locus of points P
symmetric to Q with respect to /.
Answer. The segment A'B' obtained by a parallel translation of diameter
ABy the translation being defined by the directed line segment OC
(Fig. 50).
22. Given in the plane two distinct
points A(zj) and B(z2). Find the affix p' P
of point P' which is symmetric to point
P{p) about AB.
Hint. The affix /?' is found from the
equation (Fig. 51) Pt 5
! zi z\ 1
Z2 Z 2 = 0, where pL = (p + />')/2.
\ P i P i 1| P'
Answer. pr — (z, ~z2 — z2 z\ + p(z2 — Fig. 51
- z i))l(z 2 - z i)-
23. Two distinct points ^(zj) and B(z2) are given in a plane. Prove that
if the point M(X) describes the unit circle (with the exception of a point
with affix X = — 1), then the point M 'l | describes the mid-
V 1 + A )
perpendicular of segment AB.
24. Let Dy Ey F be the points of tangency of the circle (7) inscribed
in A ABC with sides BCy CA, AB. Denote by Al9 Bl9 Cx the midpoints
of the medians DA0, EB09 FC0 of the triangles ADI, BEI, CFI emanating
from the vertices Z>, E, F (A0, B0, C0 are the respective midpoints of the
segments AI, BI, Cl). Let H be the orthocentre of A DEF. Prove that
(Fig. 52)
III IH IH
IA i = IB ,= 1C, - -
4 sin(/4/2) f 4 sin(^/2) 4sin(C/2) *
Hint. Take (DEF) as the unit circle; denote by z,9z29z%the affixes of

the points D, E9F. Then
k il ___ 2 ^ _
IA ,= IA = k? I = k i I
2\z2 + Zz\ Iz2 + z s \
and so on.
25. ABC is an arbitrary triangle; (ABC) — (O) is the circle circum

scribed about it; Q is an arbitrary point lying on that circle. Through
point A draw a straight line parallel to line OQ (Fig. 53)4 Let A' be the
Fig. 53
second point of intersection of this line with the circle (ABC), and let P
be the point symmetric to point A' about the diameter of (O) parallel
to BC. Prove that:
1°. A similar construction carried out for the points B and C leads
to the same point P.
2°. The Simson line of point P with respect to A ABC is parallel to
the line OP.
26. Let A1,A 2,A 3,A 4 be four straight lines lying in the same plane;
none are parallel and no three pass through one point. Any three of them,
say Ax, A2, A2, form a triangle. Suppose
Ci2 3 is a circle circumscribed about this
triangle. Construct similar circles C124i,
C134, ^ 234- Prove that:
1°. The centres of the circles C234, C134,
C124, C123 iie on one and the same circle
C1234.
2 . The circles C234, C134, Ci24, C123, CA34
pass through the same point y4 (Fig. 54).
To the lines Ax, A2, Az, A4 add a fifth
straight line A5, but do this so that no
three lines of the five belong to a single
pencil. Eliminating one of these lines, say
A5, we construct circle C1234 and, simi
larly, eliminating lines Ax, A2, A3, A4 in
turn, we construct the circles C2345,
^1345? Cl245> Cl235 (Fig. 55). Prove that:
3°. The centres of these five circles lie on one and the same circle C12345.
4°. The circles C2345, C1345, C1245, C1235, C1234 pass through the same
point yB, which, however, does not generally lie on the circle C12345.
Prove similar statements for any number of straight lines Ax, A2, . . ., An.
Hint. Here, in brief form, is the solution to this problem given by a
9th form student of a Moscow secondary school.
Let xl9 x2, . . . , xn be the affixes of points symmetric to the point O (an
arbitrary point taken as the origin) with respect to the straight lines
Ax, A2, . .., An. Set tj = — Xj/xj. Consider the expression
a m==___________ xi A_____________ |___________ X2J2__________
(A h) (h A)* •• (A A) (A h) (h A)- • • (A A)
From this relation it follows that
t^n.i “f" i+1 A+l @n+l,i (175)
and
Qn,i= ( 1)” &n art.n-i-l> (176)
where an = txt2. .. tn. The equation of the straight line A} is
X = fA X ~ Xj).
Z2 — i — # 2 .1 — # 3 .2 ^ 3 # 3 |1
h ~h h~ h
and, similarly, the affixes of the points of intersection of the lines Al9 A3
and A29 J 3are a3t2 — f2#3.i and #3,2 — fi# 3,i- From this it follows that the
points of intersection of the lines Al9 A2, A3 taken in pairs lie on the circle
x = a3t2 — ^#3.i (|f| = 1), the affix of whose centre is a3 2 and the radius
is \a3 i|. This circle is said to be a circle associated with three lines Al9 A29 A3,
and its centre is a point associated with these three straight lines. Also
consider the straight line A4. From the relation a3 2 = a4 3 — t4a4t2 it follows
that the affix a3t2 of the centre of the circle associated with the lines Al9
A2, A3 may be taken in the form a4t3 — t4a4y2 and, similarly, the affixes
of the centres of the three circles associated with triplets of straight lines
(Al9 A2, d4), (Al9 A3, ^ 4,) (A2, 2I3, 2I4) will be tf4f3 t3a ^2, 04,3 ^*04,2>
^ 4,3 ~ ^ 4,2* From this it follows that the centres of these four circles
(circumscribed about the triangles formed by the straight lines (Al9 A2, A3),
(Al9A29A4)9 (Al9A39A4)9 (A29 A39 Aa)) lie on the circle x =~- a4%3 — ta4t2,
the affix of whose centre is a4 3 and the radius is equal to |a4f2|. We will
call this circle the circle associated with four straight lines Al9 A29 A39 A4
and its centre the point associated with these lines.
Continuing this reasoning, we arrive at the following theorem: if n
straight lines Al9 A29. .., A„ are given in a plane, then, by eliminating one
line at a time from this set An9. . . , A2, Al9 we obtain n groups of lines with
n — 1 lines in each group; the centres of the n circles associated with
these n groups belong to one circle r „:
X= „-i - ta„' „_2 (|f| = 1), (177)
the affix of whose centre is an, n_i and the radius is |a„t n_2\. We call this
circle r n the circle associated with n straight lines Al9 A2, . . ., An9 and its
centre is called the point associated with these lines. The equation of the
circle r„_x associated with the n — 1 lines Al9 A29. ..,
A„_x is of the form
X = an_1, „_2 - ta„ „ _ 3 (|r| = 1). (178)
The complex number
gn,n-3\
tn
an, n- 2/
has a modulus equal to 1 [this follows from (176)]. If this number is substi
tuted into (178) instead of /, we obtain
u n ,2
Jn n -1 n - 2>
a n t1
which is an expression symmetric in the indices 1 , 2 (this is because

the first position is always occupied by n). From this it follows that the n
circles associated with n systems of lines having n — 1 lines in each system
(these systems are obtained by a successive elimination of one of the n lines
from the system Al9 A29. .., An) pass through a single point with affix yn.
Remark. For n = 4, the modulus of the complex number
a n, 2 __ fl4 ,2
0/1.1 0 4 ,1
is equal to 1, hence the circle T4,

X = <74j3 f 0 4>25
associated with the four lines Al9 A2, d3, d 4 [that is, containing the centres
of the four circles circumscribed about the triangles (Al9 A29 A3), (Al9 A2t d 4),
1H 810
(Al9 A39 A4)9 (A29 A39 A4)] also passes through the point with affix
**4,2
y4 — tf4 ,3 **4,2‘
**4,1
A similar situation does not hold when n > 4, because, for example, the
modulus of the complex number C^ ~ is no longer (generally speaking),
**5,1
equal to 1.
27. AXA2A3A4 is an arbitrary quadrangle (not necessarily convex).
--------> --------> --------- > -------- >
Squares AXA2PXP29 A2A3P3P49 A3A4P$P39 A4AxPnP3 are constructed on
its sides; the squares have the same orientation; also constructed on its
sides are the four squares AXA2P\P29 A2A3P3P49 A3A4P3P39 A4AxP^Pg9
which also have the same orientation, but the orientation of any one of
the squares of the first group is opposite that of any one of the squares
of the second group. Let Bl9 B29B39 B4 be the centres of the squares of
the first group, and let B[9B29B3f B4 be the centres of the squares of the
second group. Prove that:
1°. Segments BXB3 and B2B4 are equal and mutually perpendicular
(the quadrangle B1B2B3B4 that satisfies this condition is termed a pseudo
square).
2°. B[B2B3B4 is also a pseudosquare.
3d. The segments B4B3 and B2B\ have the same midpoint C\; the seg
ments B[B3 and B2B4 have the same midpoint C2. Let C3 and C4 be the
respective midpoints of the segments AXA3 and A2A4. Prove that CXC2C3C4
is a square (Fig. 56).
[Solve this problem by using complex numbers; also solve it by the
methods of analytic geometry and the methods of vector algebra.]
28. A1A2A3A4A5A(iA7A8 is an arbitrary octagon (not necessarily convex).
On its sides, construct the squares
A,A2PXP29 A2A3P3P4, A3A4P5P6, A4A5P7Ps,
^ 5^ 6^VPio> d6/47Pi1PJ2, A7AsP13P14, A g A ^^P ^

so that all of them have the same orientation, and also construct another
eight squares
A1A2PlP2, A2A3P3P4, A3A4PbP69 A4A$P7P8,
A5AeP9P 105 A6A7Pn P 12, A7A8P13Pl4, AgA^lsP^,

also having the same orientation and such that the orientation of any one
of the squares of the first group is opposite that of any one of the squares
of the second group. Let Bl9 B2, B39 B49 B59B69 B7, Bg be the respective
centres of the squares of the first group, and B[9B29 B3, B4, B'h, B B 7, Bg
the respective centres of the squares of the second group.
Prove that:
1°. The midpoints Cl9 C2, C3, C4 of the principal diagonals * of the
octagon BxB2B3B4BbB8B7Bg form a pseudosquare.
2°. The midpoints C[9C2, C3, C4 of the principal diagonals of the octagon
B[B2B3B4B$BQBiBg also form a pseudosquare.
3°. The sets of points
A 1, A2, A3, A 4, A5, A6, A7, A8,
Bi, B2, B3, B49 i?5, B69 B79 Bg,
Ci» Q? c 3, C4,
c;, a , g , c;
have a common centroid.
4°. The midpoints of the segments CXC3 and C2C4 coincide respectively
with the midpoints of the segments C2C4 and C[C3.
* The principal diagonals o f the octagon B ^ B g B ^ B ^ B jB 8 are the segments BxBh9
B2B6, B3B~, B4B6 that join opposite vertices (and also the straight lines on which the
segments lie).
5°. A necessary and sufficient condition that C XC 2C 3C 4 and, hence,

also C [ C 2C 3C 4 be squares (with the indicated order of their vertices) is
that the two sets of points Ay, A 3, A-0, A 7 and A 2, A 4, A 0, A s have a common
centroid.
6°. The squares C 4C 2C 3C 4 and C [ C 2C 3C 4
coincide ( C x — C [ , C 2 — C 2, C 3 — C3, C4 = C4)
if and only if the two sequences of the principal
diagonals of the octagon A 1A 2A 3A 4A 5A 6A 7A a
are bisected by their points of intersection.
7°. If the points Cl9 C3, C 2, C 4 and the points
C [ 9 C3, C2, C 4 coincide, then the quadrangles
A 1A 3A 5A 7 and A 2A 4A 6A S are parallelograms,
and conversely.
Solve this problem through the use of
(1) complex numbers, (2) vector algebra, and
(3) analytic geometry.
29. Inscribed in a circle (O) is a triangle ABC. Prove that the points
P, Q, R of intersection of the tangent lines, drawn at the points A, B,C
to (0), with the straight lines BC, CA, AB, respectively, lie on one straight
line (Fig. 57).
30. Through points A = (zx), B = (z2), C — (z3) lying on the unit circle
draw straight lines parallel to a given straight line A , which has slope X.
Let A y , By, Cy be the second points of intersection of these lines with the
circle ( A B C ) ; A 2, B 2, C 2 are points symmetric to the points A lHB y , C y
with respect to the straight lines B C , C A , A B respectively. Draw through
points A 2, B 2, C 2 straight lines parallel to the line A , and denote the points
of intersection of these lines with the lines B C , C A , A B respectively by
P, Q, R. Prove that the points P, Q, R lie on one straight line, and set
up the equation of that line (Fig. 58).
Answer. X(2X3 + ax )? — 1) z + X2(2 f- cr1X — ) 3) z
= 26 + 0 2 X5 — Gy X 1 F (1 — Oy (J>) X3 — G 2 )} F GyX + 1.
31. Four points A y , A 2, A 3, A 4 are taken on a circle (O). Let P be the

point of intersection of the lines A 4A 3 and A 2A 4 ; let Q be the point of inter
section of the lines A 4A 2 and ^ 3^ 4; let R be the point of intersection of the
tangents to (O ) at the points A y and A 4 ; let S be the point of intersection
of the tangents to the circle (O) at the points A 2 and A 3. Prove that the
points P, Q, R, S lie on one straight line (Fig. 59).
Hint. Take (O) as the unit circle.
32. A triangle ABC is inscribed in a circle (O). Tangent lines are drawn
to (O) at the points A, B,C; these lines form (called the tangential
triangle of tsABC). Prove that:
1°. The points P, Q, R of intersection of the straight lines: P == (BC, BqCq)
Q = (CA, C0A0), R = (AB, A0B0) lie on one straight line m.
2°. The lines AA0, BB0, CCQpass through the single point M.
3°. The point M is the pole of the straight line m with regard to the
circle (O) (or, what is the same, m is the polar of point M with respect
to this circle) (Fig. 60).
Hint. Take (O) for the unit circle.
33. AXA2A3 is an arbitrary triangle lying in an oriented plane; P is an
arbitrary point lying in the plane of that triangle but not lying on any
one of its sides, nor on the circle {A4A2A3). Let S l9 B2, B3 be the orthogonal
projections of point P on the lines A2As, A3Al9 A1A2; Cl9 C2, C3 are the
orthogonal projections of point P on the straight lines B2B39 B3Bl9 BXB29
and Dl9 D2, D3 are the orthogonal projections of point P on the straight
lines C2C3, C3Cl9 CLC2 respectively. Prove that A A tA2A3 and A D 1D2D3
are similar and have the same orientation (Fig. 61).
Hint. Take point P for the coordinate origin. Let ak9 bk, ck9 dk (k =
== 1,2, 3) be the respective affixes of the points Akf Bk, Ck9 Dk. Express bk
in terms of ak; ck in terms of bk; dk in terms of ck.
34. On the circle (O) take six arbitrary points Al9 A2, A39 A4, A59 A6.
Prove that the three points P, Qy R ,
P — (A1A2f A4A5), Q — (A2A3, A5A6), R = (A3A4j A6A4),
the p oints of intersection of the straight lines, lie on one straight line
(Pascal's theorem) (Fig. 62).
Hint. Take (O) for the unit circle.
35. Prove that the points A(z1) and B(z2),
where zx and z 2 are the roots of the
equation
z2 + 2pz + q = 0
(pand q are complex numbers and p2—
— q 7^ 0), lie on the straight line passing
through the origin of coordinates if and
only if one of the following conditions is
valid:
(1) P = 0;
(2) p tÔ, qjp2 is a real number and
(qjp2) < 1; here, if p # 0 , 0 ^ qjp2 < 1,
then the points A and B lie on one ray
emanating from the coordinate origin 0 \
and if p ¥= 0, qjp2 < 0, then the points A and B lie on opposite rays
emanating from the origin.
36. Given a cubic equation:
a0z 3 + 3ax z2 + 3a2z + a3 = 0 .
Let z1? z2, z3 be its roots. Prove that a necessary and sufficient condition
that the points A(zx), B(z2), C(z3) be collinear is as follows: either
( 1) 2a\ — 3a0 axa2 + eft az = 0
or
(2) -------(Pojh---- <h)-------- which is a real number less th a n —1/4.
(2a\ — 3a0ax a2 + dba3)2
Hint. Under the translation transformation
z= C -^
ao
(Tschirnlraus transformation), the given equation takes the form
C3 + 3/?C + q = 0 ,
where
_ a0 a2 — a\ _ 2a\ — 3a0 ax a2 + a%a3
The sum of the roots t l9 C2, C3 of the last equation is zero, and since
(Ci + C2 + C3)/3 = 0 is the affix of the centroid of the system of points
A'(Ci), B'(Q , C'(C3), it follows that this centroid coincides with the coordi
nate origin. If the points A, B9C are collinear, then the points A \ B'y C'
are also collinear (and conversely). If the points A', B \ C' are collinear,
then, by the foregoing, they lie on a straight line that passes through
the origin.
37. Let a general Cartesian system of coordinates Oxy be introduced
in the plane. The affine transformation of the plane is a correspondence
under which the coordinates x', y' of the image M'(x', y') of a point M(x, y)
are expressed in terms of the coordinates *, y of the preimage M (x9y) of
the point M '(x', y') by the linear relations
x' = axx + hxy + cl9
y' = a2x + b2y + c2,
where al9 bly cl9 a29 b2, c2 are arbitrary real numbers and axb2 — a2b1 ^0.
Prove that:
1°. Every affine transformation may be written as
z' = otz + pz + y,
where z and z' are the respective affixes of the points M and M', and
|a| ^ \fi\. Conversely, if |a| ^ \P\9 then the relation z9 = ccz + p~z + y
defines a certain affine transformation.
2°. There exists an affine transformation (and only one) which carries
any three noncollinear points A(zx), B(z2), C(z3) into the three noncollinear
points A'(zi), B '(z 2), C'(z'3), respectively.
3°. A triangle ABC is said to be metaparallel to A A'B'& if the straight
lines that pass through the points A9B, C and are parallel respectively
to the lines 2?'C', C 'A \ A'B' intersect in one point. Let z9 = <xz +
+ P'z + y be an affine transformation under which A ABC goes into
AA'B'C'. Prove that A ABC is metaparallel to A A'B’C if and only if:
(1) a is a pure imaginary number or, what is the same,
*1 1
Zo 1
Z3 1
is a real number.
4°. Prove that the concept of metaparallelism is symmetric, but not
reflexive and not transitive.
----->
5°. Prove that if A ABC is metaparallel to two of the three triangles
------> ------> ------^
A'B’C , B'C 'A\ C'A'B\ then it is metaparallel to the third one as well
-----> ------>
(in this case we say that A ABC and AA'B'C' are three-times metaparallel).
6°. Given a triangle ABC in a plane:
A = (zj), B = (z2), C = (z3).
Find the affix z3 of point C' if we know that A ABC is three-times meta-
------>
parallel to A A'B'C' with vertices
A ’ = (0), = C' — (zg).
7°. A triangle ABC is said to be orthologic to A A'B'C' if the straight

lines that pass through the points A, B, C and are respectively perpendicular
to the straight lines B'C', C'A', A'B' intersect in one point. Prove that
-----> ------>
AABC is orthologic to A A'B'C' if and only if
(1) a is a real number, or, what is the same,
Z1 Z1 1
( 2) Z2 Z 2 1
*3 z* 1
is a pure imaginary number.

8°. Prove that the concept of orthologicality is reflexive and symmetric
but is not transitive.
9°. Prove that if A ABC is orthologic to two of the triangles A'B’C',
-----> ------>
B'C'A', C'A'B', then it is orthologic also to the third triangle (in this
case we say that AABC and A A’B'C' are three-times orthologic).
Zl - Z i z2 1
^2 — z2 z3 1
— Z3 Z i 1
Answer. 6°. 4 =
Zi ZX 1
z2 z 2 1
z3 z3 1
Chapter IV
INVERSION
Sec. 1. Inversion defined. Properties of inversion
Let us adjoin to the set of all points of the Euclidean plane a single ele
ment which we will call the ideal point or the point at infinity of the plane n.
Let us agree that any straight line of the rc-plane passes through the point
at infinity and that this point does not belong to any finite figure. The
Euclidean plane supplemented by a single point at infinity (with the indi
cated agreements) is termed a Euclidean circular plane or, simply, a circular
plane. Straight lines lying in the circular plane will sometimes be called
circles of infinite radius. We will also regard points as circles; such
“circles” will be called circles of zero radius or zero circles. Two inter
secting straight lines have two common points, one proper point and the
other the point at infinity. Two parallel lines have only one point in com
mon: the point at infinity; we will say that two parallel straight lines
or two circles of infinite radius meet at the point at infinity.
Suppose 0 is a fixed proper point of a circular plane, and k is a
fixed real number not zero. An inversion [0, k] with pole 0 and power k
of the rc-plane is a one-to-one transformation of that plane under which
each proper point M of the rc-plane distinct from point 0 is associated
with a proper point M f lying on the straight line OM and such that
O M O M ' = (OM) (<OM') = k.

With the pole O of inversion we associate the point at infinity 0 ', and
with the point O' we associate the point O.
Every inversion I is an involutory transformation, that is,
/ 2 ^ E,
where E is the identical transformation.
This follows from the fact that if under inversion / a point M ' is the
image of a point M, then M is the image of M'.
If the power k of the inversion [0, k] is positive, then the circle K with
centre O and radius ]fk is termed the circle o f inversion, and the inversion
itself is also called a symmetry with respect to the circle K. Under the
inversion [0, &], where k > 0, each point M of the circle of inversion is
invariant, that is, its image M' coincides with the point itself.
To construct the image M' of point M lying outside the circle K of
inversion [0, &], where k > 0, we draw from point M a tangent M T to
the circle K (T is the point of tangency); M' is the projection of T on
the straight line OM (two tangents may be drawn, M T and M T , the
point M' is the point of intersection of straight lines OM and TT')

(Fig. 63).
To construct the image M ' of point M lying inside the circle K of
inversion [O, k\9 k > 0, draw through M a straight line perpendicular
to the straight line OM. Let T be any one of the points of intersection
of this perpendicular with the circle K. Then the tangent to the circle K
at the point T will intersect the straight line OM in the point M ' (Fig. 64).
Thus, under the positive inversion [Oyk\9 k > 0, points lying outside
the circle of inversion K pass into interior points of that circle, and points
lying inside the circle K pass into points lying outside the circle.
For negative inversion [O, fc], k < 0, the circle K with centre O and
radius ^\k\ is invariant; however, each one of its points is noninvarian
and passes into a point diametrically opposite it. To construct the image M
of a point M under negative inversion [O, k]9 k < 0, we construct the
image M* of point M under the negative inversion [0, |&j], M ' is symme
tric to point A/* with respect to point O. From this it follows that under
negative inversion as well the set of all points lying outside the circle K
passes into the set of interior points of circle K and the set of all interior
points of circle K goes into the set of all exterior points of K (Fig. 65
and Fig. 66).
Let [0, k\ be an inversion with pole O and power k. If C is an arbitrary
circle that does not pass through the pole of inversion, then its image C'
under the inversion [0, k\ is again a circle that does not pass through
the pole of inversion.
Proof. Let M be an arbitrary point of the circle C, and let M ' be
its image under the inversion [0, k]9 that is, OM- OM' = k; here it is
well to point out that there is a frequently used notation OM -OM ' = &,
where we speak of the product of the lengths (OM) and (OM') of directed
line segments. Since the straight line OM has a common point M with
the circle C, this line has a second common point N (possibly M = N)
with the circle C. The product OM- ON — OM- ON = a is the power
of the pole O of inversion with respect to the circle C (Fig. 67). From
Inversion 283
the last two equations it follows that
Thus, the point M ' is the image of point N of the circle C under the
homothetic transformation (0, kja). But under the homothetic transfor
mation (0, k ja \ circle C goes into circle C'. If point M describes a
circle C, then the point N too describes the same circle C, and, hence,
its image M ' under the homothetic transformation (0, kja) describes
the circle C . Thus, the circle C is the image of C also under the homo
thetic transformation (0, kja), where a is the power of the pole 0 of
inversion with respect to the circle C* and under the inversion [0, k].
* Note that the inversion [0 , £ ] and the homothetic transformation (0 , fc/rr) carry
circle C into one and the same circle C'; however, the points lying on circle C are
transformed by the inversion [0 , k ] and the homothetic transformation (0 , kja) into
different points of circle C u n d e r the inversion [0, k], point M goes into point M ' of
circle C', and under the homothetic transformation (0 , kfo), point M goes into point
N 'y where N' is the second point o f intersection of straight line O M ' and circle C'.
If circle C passes through the pole of inversion 0 , then its image under
the inversion [0, A] is a straight line that does not pass through the pole
of inversion and is perpendicular to the straight line joining the pole of
inversion and the centre of circle C.
If the straight line does not pass through the pole of inversion 0 , then
its image under the inversion [0, A] is a circle that passes through the pole
of inversion.
If the straight line passes through the pole of inversion 0 , then under
the inversion [0, A] it passes into itself.
Under inversion, the tangency of the circles is retained.
This follows from the fact that under inversion, a circle passes into a
circle and the inversion is a one-to-one transformation (to explain this
more fully: under inversion, two tangent circles can go into two tangent
circles or into a circle and a tangent line to the circle, or into two parallel
straight lines).
If the pole of inversion [0, A] lies outside circle C, and C is the inverse,
then the set of points lying inside circle C goes into the set of points
lying inside circle C' (and conversely). The set of points lying outside (C)
goes into the set of points lying outside (C') (and conversely) (see Fig. 67).
If the pole 0 of inversion [0, A] lies inside circle C, and C' is the inverse
of (C), then the set of points lying inside (C) goes into the set of points
lying outside circle C', and the set of points lying outside (C) goes into
the set of points lying inside circle C'.
Remark. When investigating the mapping of regions under an inversion,
it is useful to bear in mind the following. Let us consider the inversion
[0, A] where A > 0. Let M be an arbitrary proper point of the circular
rc-plane, and let M' be its image under the inversion [0, A], that is,
{O^-OMT) = OM OM’ = A.
From this relation it follows that if the point M moves along the ray OM
receding from the pole, then the point M f will move in the opposite direc
tion since the product OM- OMr must remain constant and equal to A
(the points M and M ' will meet on the circle of inversion). The same
occurs in the case of the opposite ray to the one we considered. In concrete
cases (see below), this reasoning may be utilized profitably when investi
gating the mapping of regions under an inversion and when finding in
variant regions, that is, regions that go into themselves under the inver
sion in question.
The angle between two intersecting circles is preserved under an inver
sion, but the orientation of the angle is reversed (inversion is a conformal
transformation of the second kind). The proof of this proposition is given
below when we consider the inversion of space.
Inversion 285
Remark. A linear fractional transformation of the plane of a complex,

variable
z' = a d - b e * 0, c = 0
cz + d
may be rewritten in the form
, a A t A be — ad
z — -----\---------- —, where A = -----------.
c . d c2
c z H-----
c
It may be interpreted geometrically as follows:
d
1°. zx = z H-----is a translation.
c
2C. z2 = — is an inversion with respect to the unit circle with subse-
quent symmetry with respect to the x-axis *.
3°. z3 = A z 2 is a similarity transformation with centre O : a rotation
about the origin of coordinates through an angle arg A and a homothetic
transformation (O, jA ).
4°. z' = z3 + is a translation again.
c
From this it follows that a linear fractional transformation is a conformal
transformation of the first kind (that preserves orientation of angles), si nee
the transformations 1°, 3°, 4° preserve orientation of angles, and the inver
sion z2 — — with subsequent symmetry with respect to the x-axis also
preserves orientation of angles.
Sec. 2. Problem, involving inversion
Problem 1. Let A' and Bf be images of the points A and B under the
inversion [0, k\. Express the length of segment A'B' in terms of the lengths
of the segments ABy OA, OB and in terms of k. It is assumed that the points
A and B are distinct from point O.
Solution. Suppose that the points O, A and B do not fie on one straight
fine. Let k > 0. Then the points A' and B' fie, respectively, on the rays
OA and OB, and we have
OA •OA' = OB OB’ = k.
I*al lzil = L arg -2 = —arg z x (mod 2n).


OA’ _ OB’
OB ~ OA '
Hence, A OAB and A OB’A’ are similar (but with opposite orientations)
(Fig. 68). From the similarity of these triangles it follows that
A ’B’ OA’ OA’-O ^ OA O W _ k _ \k\

AB ~ OB~ OB-OA ~ OA-OB ~ OA-OB OA-OB*
and, hence,
OAOB’
This formula is also true if the points O, A and B lie on one straight line
and in the case of k < 0.
Problem 2. Prove that it is possible to circumscribe a circle about a
convex quadrangle ABCD if and only if the product of the diagonals of
the quadrangle is equal to the sum of the products of its opposite sides
( Ptolemy's theorem):
AC-BD = 'AB CD + BC-AD.
I. Suppose we can circumscribe a circle K about a quadrangle ABCD;

we will then prove that the relation AC-BD = AB-CD + BC-AD holds
true.
Consider the inversion [A , 1], Under this inversion, the circle K goes
into the straight line K ', and the points B, C, D go into the points B \ C",
D’ lying on that straight line. The point C will lie between points B’ and
D \ and therefore (Fig. 69)
B C + C D ’ = B ’D’
Inversion 287
or, on the basis of the preceding problem,

_ * c __. + CD _ BD
A B A C ^ AC AD ~ A B -A D ’
whence follows the relation AC-BD — AB-CD + BC-AD.
II. Suppose the relation AC-BD — AB-CD -\- BC-AD is valid. Let
us then prove that the quadrangle ABCD is convex and a circle can be cir
cumscribed about it.
We consider the inversion [A, 1]. Let B', C \ D' be the inverses of the
points B, C, D. From the relation
AC-BD = AB-CD + B C A D
we have
BC CD _ BD
AB AC AC-AD ~~ A B -A D ’
and from this relation and the result of problem 1 it follows that B'C' +
f C'D' = B’D’. Hence, the points B \ C', D' lie on one straight line
and the point C lies between the points B' and D'. From this it follows that
the points B, C, D lie on one circle (that passes through point A), which
is the image of the straight line K' under the inversion [A, 1]. Since the
ray AC lies inside A formed by the rays AD and AB, it follows that AC
is a diagonal of the quadrangle ABCD, and this means that the quadrangle
is convex (under the indicated order of its vertices).
Problem 3. Inscribed in a circle K is an equilateral triangle ABC. Let O
be a point not lying on the circle K. Prove that there is a triangle with sides
OA, OB and OC. Prove that if the point O lies on the circle K, then the
sum of two of its segments OA, OB, OC is equal to the third.
Proof. Suppose O does not lie on the circle K (Fig. 70). Consider the
inversion [ O, 1]. Circle K goes into circle K \ and the points A, B, C go
into the points A', B', C' that lie on K' and, hence, do not lie on one
straight line. On the basis of problem 1, we have
BC OA BC
OB-OC OA- OB-OC
CA O BC A
(i)
OC OA _ OA- OB-OC
AB _ OCAB
OA OB ~~ OA- OB-OC
and, hence,
B'C' : C fA' : A'B' = OA : OB : OC
since BC = CA = AB. Thus, the segments OA, OB, OC are proportional
to the sides B'C', C'A', A'B' of A A'B'C' and, hence, there is a triangle
with sides OA, OB, OC (this triangle is similar to A A'B'C').
Suppose point O lies on the circle K, for example, on the arc AC, that does
not contain B (Fig. 71). Under the inversion [O, 1], circle K goes into line
A", the points A, B, C go into the points A', B', C' that lie on K', and the
point B' will lie between points A' and C'; thus,
A'B' + B'C' = O'C'
and since relation (1) is valid in this case as well, it follows that
OB = OA + OC.
Remark. The theorem holds if point O is chosen arbitrarily in space.
The proof is analogous.
Problem 4. Two circles Cx and C2 with centres Ox and 0 2 are externally
tangent to each other. The straight line / touches both circles at distinct
Inversion 289
F ig . 71
points A and B. Construct a circle tangent to the two given circles and the
straight line /.
Solution. Let us consider the inversion [A, AB*]. Under this inversion,
circle C2 goes into itself because if an arbitrary straight line is drawn
through point A , the line intersecting circle C2 in points M and M \ then
AM -A M 1= AB2. Circle Cx goes into line C[ parallel to line / and tangent
to circle C2 (Fig. 72). Thus, the problem reduces to constructing a circle
that is tangent to circle C2 and to two tangent lines I and C[ parallel to it.
There are two such circles. Let K[ be one of these circles, let A[ be the point
of tangency of circles K{ and C2, and let B[ be the point of tangency of
K[ to line C[. Let Ax be the second point of intersection of line AA[ with
circle C2; point Ax is the image of point A[ under the inversion [A, AB*]. Let
Bx be the point of intersection of line AB[ with circle Cx (point Bx is dis
tinct from point A); point Bx is the inverse of point B[. The points Ax
and Bx are points of contact of the desired circle with the circles C2 and
Cx respectively. The centre Px of one of the desired circles is the point of
intersection of the straight lines 0 2AX and OxBl9 and the radius is equal
to PXAX= PXBX. The second circle is constructed in similar fashion.
F ig . 72
19—810
F ig . 7 3
Problem 5. Two circles Cx and C2 intersect in points A and B. A point C,

different from A and B and lying outside circles Cx and C2, is taken on a
straight line AB. Construct a circle that passes through point C and is
tangent to circles Cx and C2 (Fig. 73).
Solution. Consider the inversion [C, &], where k = CA-CB. Under this
inversion, each of the circles C1 and C2 goes into itself, and the desired
circle goes into a straight line tangent to the circles C1 and C2. Since the
two intersecting circles Cx and C2 have two common tangents K[ and K!ly
it follows the problem has two solutions. Let A[ and B[ be the points of
tangency of line K[ with circles Cx and C2. Denote by Ax the second point
of intersection of line CA[ with circle C2, and by Bx the second point of
intersection of line CB[ with circle C2. One of the desired circles passes
through points C, Ax and Bx. The second circle is constructed in similar
fashion (Fig. 73).
Problem 6. Given in a A ABC the radius r of an inscribed circle and the
radius R of a circumscribed circle. Find the distance d between their
centres.
Solution. If, under the inversion [O, &], circle C goes into circle C \
then circle C goes into circle C' also under the homothetic transformation
(O, k/a), where a is the power of the point O with respect to the circle C.
Consider the inversion [P, r2], where P is the centre of the circle inscribed
in the given triangle, and r is the radius. Under this inversion, the points
of the inscribed circle are fixed (since the inscribed circle is the circle of
inversion). The vertices of the given triangle under the inversion [P, r2J
go into the midpoints of the sides of A AxBxCXy whose vertices are the
points of contact of the sides of A ABC with the circle inscribed in A ABC
(Fig. 74). The radius of the circle passing through the midpoints of the
sides of the indicated triangle is equal to r/2. Thus, the circle {ABC)
Inversion 291
circumscribed about triangle ABC, whose radius is R, goes into a circle of

radius rj2 under the inversion [P, r2]. Since the power of the inversion
under consideration is k = r2, and the power of the point P with respect
to the circle {ABC) is equal to a — d2 — R2, it follows that
r/2 _ r2_ r2
1r ~ |d2 - * 2| ~ R2 - d2’
whence
d2 = R2 - 2Rr
(Euler’s formula).
Problem 7. N and S are two diametrically opposite points of a circle C;
I is a straight line tangent to C at point S. From an arbitrary point O lying
outside circle C but not lying on the tangent to C at N, we draw to circle
C tangent lines OA and OB (A and B are the points of tangency). Let O',
A f and Bf be the projections from point N on the straight line / of points
O, A and B. Prove that O' is the midpoint of segment A'B' (Fig. 75).
Proof. Under the inversion [N, N S 2], circle C goes into line / and circle K
with centre O and radius OA = OB, which is orthogonal to circle C, goes
into circle K which is orthogonal to line I; hence, the centre of K ' lies on
/; on the other hand, the centre of K f also lies on line NO, and for this
reason the projection O' of point O from point N on line / is the centre
of K'. Hence, A'B' is a diameter of K and O' is the centre of K ', and so
A'O' = O'B'.
F ig . 7 5
Problem 8. Circles C± and C2 do not have any points in common. Use

inversion to transform them into two concentric circles and indicate how,
under this inversion, the regions Du Z>2, L>3 into which the circles Cl, c 2
divide the plane, are transformed (Fig. 76).
Solution. Construct some circle K that orthogonally cuts both circles Cx
and C2. To do this, draw some circle M that cuts both circles Cx and C2
at the points P, Q and R, S respectively. Let O be the point of intersection
of the straight lines PQ and R S (Fig. 77). Then the segments m of tangents
drawn from point O to the circles Cx and C2 will be equal and, hence.
F ig . 7 6
Inversion 293
F ig . 7 7
the circle K with centre O and radius m will intersect both circles Cx and
C2 orthogonally. Denote by A and B the points of intersection of K with
the line of centres of Cx and C2. Then under the inversion [A, AB2],
the circles Q and C2 go into two concentric circles with centre B. Indeed,
the circle K goes into the straight line K ' that passes through point B and
since K is orthogonal to the circles C2 and C2, it follows that line K' will
be orthogonal to C2 and C[ into which circles Cx and C2 go; that is, the
centres of C[ and C2 must lie on the straight line K \ But they also must lie
on the straight line AB, hence, the centres of C[ and C2 coincide with
point B.
Furthermore (see Fig. 76), since the centre A of inversion lies inside circle
Cl9 the region Dx of points lying inside Cx goes into the region D[ of points
lying outside C[. Since the centre A of inversion lies outside C2, the region
D2 of points lying inside C2 will go into the region D'2 of points lying inside
C2. Hence the region Z)3 of points lying outside Cx and C2 will go into a
plane annulus bounded by the circles C[ and C2.
Consider the case where the circle C2 is put inside circle C2 (see Fig. 77).
In this case, the circle K that cuts C2 and C2 orthogonally goes into the
line K' = BC, where C is the second point of intersection of K and the
circle of inversion. If T and L are the points of intersection of K with Cx
and C2 and if the straight lines TA and LA intersect K' in the points T'
and L \ then the images of and C2 are the concentric circles C'2 with
centre B and radii B T and B L Furthermore, since point A lies inside C2,
it follows that the region D2 of points lying inside C2 goes, under the inver
sion [Ay AB2], into the region D2 of points lying outside C2. The region D3
of points lying outside Cx goes into the region D'3 of points lying inside C[,
and this means that the eccentric annulus Dx bounded by C2 and C2 goes
into annulus D[ that is bounded by the concentric circles C[ and C2

(Fig. 77).
Problem 9. Construct a circle tangent to three given circles Cl9 C2, C3
(Apollonian problem).
Solution. Consider only the case where each of the circles Cl9 C2, C3 lies
outside the other two. Let us perform an inversion under which Cx and
C2 go into two concentric circles C[ and C'2 (problem 8). Suppose that the
pole A of this inversion [A, AB2] lies inside Cl9 then R[>R2 where Ri and R2
are the respective radii of the circles C[ and C2. Since C3 lies outside Cx
and C2, it follows that its image C3 under the inversion [A, AB2] will lie
inside the annulus formed by C[ and C'2 since points lying outside Cx go
into points lying inside C*l9 and points lying outside C2 go into points lying
outside C2 (Fig. 78).
We split into two sets all the circles tangent to the concentric circles C[
and C2: circles of radius (/?( — R2)/2, each of which is externally tangent to
C2 and internally tangent to C[, and the circles of radius (i?( + R£!2,
each of which is internally tangent to C[ and C2. Only four circles K{, K'2,
K3, Kl belong to the first set, the first two of which are externally tangent
to circle C3, and K3 and Kl are internally tangent. To construct the circles
K[ and K f2 it suffices to construct their centres: these are the points of inter
section of the circles ( 0 \ (R[ — R2)/2) and ( 0 '\ Rl + (/?( — R 2)I2), where
O' and O" are the respective centres of C[ (or C2) and C3. The centres of
the circles K3 and Kl are the points of intersection of the circles (O', (R[~
-tf')/2 ) and (O", (R' - RQ/2 - R£.
Similarly, from among the circles of the second set that are internally
tangent to the circles C[ and C2 there are only four circles SJ, S'2, S3, S'4
that are externally tangent to circle C3 (5( and S%) and internally tangent
F i g . 78
Inversion 295
to C3 (S 3 and Si). The centres of the circles S[ and S'2 are points of inter
section of the circles (O', + Ri)/2) and (O", (Ri + R£)/2 + Ri), and
the centres of the circles Si and Si are the points of intersection of the
circles (O', (R[ + J?£)/2) and (O", (R[ + R'2) /2 - Ri). All these eight
circles K{, Ki, Ki, K4', S[, Si, Si, Si are constructed in Fig. 78. Their images
Kl9 K2, Ks, K±, Sl9 S2, S2, S4 under the inversion [A, ,42?2] will be tangent
to the three given circles Cl9 C2, C3. Thus, in the case at hand, the
problem has eight solutions.
Figure 79 gives the positions of the circles Kt and S t with respect to the
circles Cl9 C2, C3 on eight separate figures.
Problem 10. Prove that the Euler circle of A ABC is tangent to the circle
(I) inscribed in that triangle and is tangent to the three circles (4 ), (4 ), (/c)
escribed in that triangle (the points of tangency are <P0, <Pl9 <P2, # 3 and are
called Feuerbach points).
Solution. Let (/) and (Ia) be the respective circles (one inscribed in the
given triangle and the other escribed in Z_A). Denote by R and S the points
of tangency of (/) and (4) with side BC (Fig. 80). Then BS = CR (= p —c9
where p is the semiperimeter of A ABC). Let A', B', C' be the respective
midpoints of the sides BC9 CA9 AB. Denote by A", the orthogonal projec
tion of point A on the side BC, and by Q the point of intersection of side
BC with bisector IIa of /^BAC. The set of points A, Q, I, Ia is an harmonic
set of four points, that is, the double ratio of these four points is equal to
—1. Hence the points A ", Q, R, S, which are the orthogonal projections of
the points A, Q, I, Ia on line BC, will also be an harmonic set of four points.
Point A ' is the midpoint of segment R S (since A' is the midpoint of BC
and BS = CR); hence, A'Q -A'A" = A'R2. Let us consider the inversion
[A', A'R?\. Under this inversion, the circles (I) and (Ia) are invariant (since
the circle of inversion is a circle with diameter R S and is orthogonal to
both circles). From the relation A'Q-A' A" — A'R2 it follows that
under this inversion the point A" goes into point Q.
On the other hand, point A" (the foot of the altitude from A onto BC)
lies on the Euler circle, the point A' (the midpoint of side BC) also lies
on the Euler circle and, hence, under the inversion [A', A'R2] the Euler
F ig . 8 0
Inversion 297
circle goes into a straight line that passes through point Q and is anti-
parallel to the straight line B'C' with respect to /_ C'A'B'* or into a straight
line antiparallel to BC with respect to /_ CAB (because BC || B'C', CA\\
]|C'A', AB || A'B'). But the straight line that passes through point Q and
is antiparallel to line BC (with respect to BAC), is a straight line sym
metric to line BC with respect to the bisector of /_BAC, that is, the second
common tangent co' (interior) to the circles (/) and (7fl). Let and be
points of tangency of co' with (7) and (7fl). Under the inversion [A', A'R2],
the straight line co' is transformed into the Euler circle (co), and the points
#o and <P[ into the points # 0 and #1 that lie on (7) and (7fl). In these points
<2>0 and <Pl9 the Euler circle is tangent to the inscribed circle (7) and to the
escribed circle (7fl) because the straight line co' is tangent to (7) and (7a)
in the points d>o and <P[; # 0 and are the points of intersection of the
straight lines A'<P'0 and A'<P[ with the Euler circle or the second points of
intersection of these lines A'<P'0 and A'&[ with (7) and (7fl). Figure 80 depicts
the construction of thirteen points A', B', C , A", B", C", A" B" ', C"
<P0, # 2, # 3 (all of them lie on the Euler circle).
Sec. 3. Mapping of regions under inversion
Problem 1. Suppose a circle C lies outside the circle K of the inversion

[O, r2], where r is the radius of K. Let C' be the image of circle C under
the inversion [O, r2]. Then the set D± of all points lying inside C is mapped
one to one onto the region D[ of all points lying inside C'; the set 7)2 of
all points lying outside circles C and K is mapped one to one onto the set
D'2 of all points lying inside K but outside C'. The connected region D2 U D'2
is invariant under the inversion [O, r2] (Fig. 81).
* Straight lines that are symmetric with respect to the bisector o f / _ CAB are
termed antiparallel with respect to that angle.
F ig . 8 2
Problem 2. A circle C is externally tangent to a circle K of the inver

sion [0, r2]. Let C' be the inverse of C. The correspondence of the regions
under the inversion [0 , r 2] is similar to that considered in problem 1
(Fig. 82).
Problem 3. A circle C cuts the circle K of the inversion [0, r2] but the
centre 0 of the inversion lies outside C. The correspondence of the regions
under the inversion is indicated in Fig. 83. The regions D2 U D2, Ds u D'Zy
Z>4 U Di, Dq u Dq are simply connected and invariant under the inversion
[0, r2]. They are divided by the circle of inversion K into the regions D29
D2\ DS9 D%; Z)4, D^; 0 6, D'Qthat pass into one another under the inversion
at hand (Fig. 83).
Problem 4. A circle C passes through the centre of the inversion [0, r2]
and cuts the circle of inversion K. Under the inversion [0, r2], the image C'
of circle C is a straight line passing through the points of intersection of C
and K. The correspondence of the regions under the inversion is shown in
Inversion 299
Fig. 84
Fig. 84. The regions Dx U D[ and D5 u D'b are simply connected and are
invariant under the inversion [09r2]. The circle of inversion K divides them
into the regions Dx and D[, D5 and D 's, which go into one another under
the inversion at hand (Fig. 84).
Problem 5. The circle of inversion K lies inside circle C. The image of
the connected region Dl9 which consists of points lying outside circle C
is a simply connected region D[, which consists of points lying inside circle
C', where C is the inverse of C with respect to circle K. The image of region
D2, which consists of points lying outside K, but inside C, is the region D'29
which consists of points lying inside K but outside C'. The connected region
D2 U D2 is invariant under the inversion under consideration (Fig. 85).
Fig. 85
Problem 6. A circle C intersects the circle of inversion K orthogonally.

In this case the image C of circle C (under the inversion with respect to
circle K) coincides with the circle itself: C = C'. The region which con
sists of all points lying inside C but outside K goes into the region D[,
which consists of all points lying inside C and inside K. The region Dx UD'l9
which consists of all points lying inside C is invariant under the inversion
under consideration. Let P and Q be points in which the circles C and K
intersect. Draw rays OP and OQ. The region D 2, which is bounded by arc
PQ of circle C and the radii OP and OQ of circle K goes into the region Z)2,
which is also bounded by arc PQ of circle K (the complement of the first
arc of this circle) and by the prolongations of the radii OP and OQ beyond
points P and Q . Finally, the region Dz consisting of points lying outside K
and outside the angle POQ goes into region Dz, which consists of all points
lying inside K but outside /_ POQ . The region Dz u Dz is simply connected
and invariant under the inversion with respect to circle K (Fig. 86).
Problem 7. A circle C passes through the centre O of the circle of inver
sion K and lies inside K. The image C of circle C is a straight line that does
not intersect K.
The region Dl9 which consists of all points of the half-plane on the side
of straight line C that does not contain circle K , goes into a set D[ of all
points lying inside C. The region Z>2, which consists of all points lying out
side K and located in the half-plane (reckoned from line C ) that contains
AT, goes into the set D'2 of all points lying inside circle K but outside circle C.
The connected region D2 u D ’2 is invariant under the inversion at hand
(Fig. 87).
Problem 8. A circle C goes through the centre O of the circle of inver
sion K and is tangent to that circle. The image C" of circle C under the inver
sion with respect to circle K is the common tangent to C and K. The cor
respondence of regions Dx and D[, D2 and D2 is similar to the preceding
problem. The region D2 U D2 is simply connected and is invariant under
the inversion with respect to the circle K (Fig. 88).
Inversion 301
Problem 9. A circle C intersects the circle of inversion K and the centre

O of inversion (that is, the centre of circle K) lies inside C. The correspon
dence of regions Dx and D[, D2 and D'2, Dz and Z>3, Z>4 and Z>4 is shown in
Fig. 89. The regions Dx U D[, and Z)4 U D\ are simply connected and are
invariant under the inversion with respect to circle K.
Problem 10. Given three equal circles Cl9 C2, C3 that pass through one
point O and intersect at angles of n/3. Take point O as the pole of inversion
and take the circle of inversion K so that it intersects Cl9 C2, C3 and so that
the points of intersection lie inside K. The images C[9C2, C3 of the given cir
cles Cl9 C2, C3 are straight lines passing through the points of intersection
of circle K with each of the circles Cl9 C2, C3. The given circles Cl9 C2, C3,
their images C[9 C2, C3 (under the inversion with respect to the circle K),
and the circle K itself divide the plane into 24 regions. The correspondence
of regions under the inversion of the plane with respect to circle K is shown
in Fig. 90. The regions D7 U D?, D8 u D89 D9 u D9y D10 u A'o* A i U
U D[i, Dx2 U D'12 are simply connected and invariant under the inver-
Fig. 89
sion at hand. The regions Dk and D'k (k = 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11,

12) go into one another under this inversion.
Problem 11. Consider the cardioid
p = 1 + cos <p,
the equation of which is given in polar coordinates. Under the inversion
[ 0 , 1], where O is the pole of the polar system of coordinates, the image of
the cardioid is a parabola whose equation in that same polar coordinate
system is of the form
1
P ---------------------------
1 + cos q>
If a rectangular Cartesian system of coordinates is introduced and the
polar axis is taken as the x-axis, then the last equation becomes
y2 = 1 — 2x .
Figure 91 depicts a parabola constructed on the basis of this equation. The
cardioid p = 1 + cos <p, the parabola y2 = 1 — 2x, and the circle of inver
sion K divide the plane into six regions. The correspondence of these regions
under the inversion with respect to circle K is depicted in Fig. 91. The
regions Dx U D[ and D3 U D3 are simply connected and invariant under
the inversion with respect to circle K Region D2, which
is bounded by a part of the cardioid and an arc of the parabola, goes into
region D'2, which is made up of points lying outside the cardioid and the
parabola. Conversely, under the inversion at hand, region D2 goes into
region D2.
Inversion 303
Fig. 9
Problem 12. We consider Pascal’s limagon, the equation of which in

polar coordinates is of the form
p = m +72 cos cp.
We assume that 0 < n < m. Then we have p > 0 for all values of cp, and,
hence, the limagon considered here is a closed curve without self-inter
sections, inside which curve is the pole O of the polar system of coordinates.
Under the inversion [O, 1], where O is the pole of the polar system of coor
dinates, the limagon goes into a curve, whose equation in that same polar
1
P= m -------- i--------------------
+ n cos cp
or
P
P=
1 + e cos (p
where p = \jm , e = n/m. Since from the conditions 0 < n < m it follows
that 0 < e < 1, the last equation is the equation of an ellipse for which O
is one of the foci, p is a parameter, and e is the eccentricity.
For example, take m = 3/2, n = 1. Then the equation of Pascal’s li-
magon is
3 ,
p = --------- b COS cp.
2
and the equation of the ellipse, which is the image of this limagon
under the inversion [O, 1], is
1
--------- b COS cp
If a rectangular Cartesian system of coordinates is introduced, and the

polar axis is taken as the x-axis, then the last equation becomes
25 5
which is an ellipse with centre at the point (—4/5, 0) and with semi-axes
equal to a = 6/5, b= 2/^5. Figure 92 depicts the limagon under consid
eration and the ellipse, which intersects the limagon p = 1.5 + cos cp for
values of q> that satisfy the equation
3_ 1
+ COS (p =
2 3
-------- , C O S (p ’
\-
2
whose cos (p = ------- ; and, thus, assuming that 0^<p<2n, we have

2
cp = 2n/3, <p = 47t/3 (two points of intersection). For these values of (p
we have p = 1 and, hence, the circle of inversion K passes through the
points of intersection of the limagon and the ellipse.
The limagon, the ellipse, and the circle of inversion K divide the plane
into six regions. The correspondence of these regions is shown in Fig. 92.
The regions Dx U D[ and D2 U D'2 are simply connected and invariant
under inversion, and under the inversion at hand they pass into one anoth-
Inversion 305
er: D i ^ D[, Z>2 ^ The region Z)3, which is bounded by an arc of the
limaôn and an arc of the ellipse, contains the centre O of inversion (the
pole of the polar coordinate system), goes into the region D3, which consists
of all points lying outside the ellipse and the limagon. Conversely, under
the inversion [O, 1], the region Z>3 goes into the region Z>3.
Problem 13. Let us consider Pascal’s limagon once again in polar coor
dinates; it is given by the equation
p = m + n cos cp9
but it is now assumed that 0 < m < n. The equation p = 0 now has the
solution
cos cp = — mjn
and if we assume 0 ^ cp < 2k, then the last equation yields two values
for<p:
cp = arccos (—m/ri), cp = 2 n — arccos ( —mjn).
Therefore, when 0 < m < n the limagon passes through the pole twice.
In the case at hand, this is a curve with self-intersection; it forms a loop
inside the rest of the curve. Under the inversion [O, 1], where O is the pole
of the polar system of coordinates, the limagon goes into a curve whose
equation in the same polar system of coordinates is of the form
1
P = — -----------
m + n cos cp
or
H 9
1 + e cos cp
where p= 1/m, e = njm. Since we now have e > 1, the curve specified by
this equation is a hyperbola with eccentricity e and parameter p (half the
focal chord).
Let us now take, for example, m = 1, n = 2 (Fig. 93). The equations of
the limagon and its image (hyperbola) under the inversion [O, 1] are
p = 1 + 2 cos cp, p = ---------- .

1 + 2 cos cp
Transform the equation of the hyperbola by introducing a rectangular
Cartesian system of coordinates in which the x-axis is the polar axis. We
then have
9 3
2 0 -8 1 0
Fig. 93
This is a hyperbola with centre (2/3, 0) and semi-axes a = 1/3, b= l/]/3.

Since bja = |A3, it follows that the asymptotes of the hyperbola are inclined
to the #-axis at angles of ±7r/3. When cp = ±2n/3 in the equation of the
limagon p = 1 + 2 cos <p, the radius vector p vanishes. Therefore the asym
ptotes of the hyperbola are parallel to tangents to the loop of the limagon
at the coordinate origin. The limagon, the hyperbola, and the circle of inver
sion K divide the plane into 14 regions. The correspondence Dk^±D£ of
these regions is depicted in Fig. 93. The regions Dz UDz, Z>4 U D\,
D7 U D'7 are simply connected and are invariant under our inversion;
also the circle of inversion divides them into the regions Dz and Dz, Z>4
and Z>4, D7 and D7i which go into one another under the inversion. The
limagon and the hyperbola intersect in seven points; three of them lie
also on the circle of inversion; the other four P, P', Q, Q' do not lie on
the circle of inversion and correspond to one another {P ^l P \ Q Q')
under inversion (symmetry with respect to the circle K). Using the straight
lines OPP' and OQQ\ we can partition the regions Dx and D[ into three
* ** *** * ** ***
regions: Dl9 Dl9 and D[, D[, which will correspond to one another
under the inversion at hand. These same lines divide the regions Z>3, Dz
and Z>4, Z)4 into two parts each, and they correspond to each other under
the inversion [0, 1].
Problem 14. The equation of the cissoid of Diodes in polar coordinates
may be written as
_ 2 sin2 = lim (2 sin2 q>) = 2.
71 71 71
<p -» — —0 <p-» ------- 0 <p - » ------o
2 2 2
Under the inversion [0,1] (0 is the pole of the polar system of coordinates),
the cissoid is transformed into a curve whose equation in the same polar sys
tem is of the form
cos cp
2 sin2 <p
or
2'y2 = x.
This is a parabola for which the polar axis serves as the axis. The circle of
inversion K passes through two points in which the cissoid and parabola
intersect. (Note that the circle of inversion does not intersect the asymptote
x = 2 of the cissoid.) The cissoid, the parabola, and the circle of inversion
divide the plane into eight regions Dk9 D ’k (k = 1,2, 3, 4), which pass into
one another under the inversion at hand: Dk Dk.
The regions Dx U D[ and Z>4 U D\ are simply connected and are in
variant under the inversion [<9, 1] (Fig. 94).
Sec. 4. Mechanical inversors:

the Peaucellier cell and the Hart cell
Suppose MPM'Q is a rhombus. The point O is a point equidistant from
points P and Q. Also suppose the rhombus is hinged, point O being fixed
and joined by hinges with P and Q (the Peaucellier cell). Then if M describes
//' some curve L, the point M ' describes a curve L',
which is the image of L under the inversion
[O, OP2 - PM 2] (Fig. 95).
Proof, The points M 9M ' and O are at equal dis
tances from the points P and Q, and for that reason
lie on the midperpendicular of segment PQ, Let us
construct the circle (P, PM), The product OM-OM'
0 is equal to OP2 — PM 2. Indeed, let S be the mid
point of segment PQ ; then
OM-OM' = (OS + SM) (OS + SM')
= (OS + SM) (OS - SM)
= OS2 - SM 2 = OP2 - SP2 - (MP2 - SP2)
= OP2 - PM 2.
Note, in particular, that if point M describes a
circle passing through point O, then point M ' de
scribes a straight line. Thus, the Peaucellier cell makes
it possible, mechanically, to transform circular mo
tion into rectilinear motion.
Let ABCD be an antiparallelogram (that is, ABDC is an isosceles trape
zoid, BD and AC are its parallel sides, AD and BC the diagonals). Fix
a point O on segment AB and a point M on segment AD, and on segment
2?C fix a p o in t M 'so that the points O, M, and M f belong to a single
straight line parallel to BD ||AC. The points A, B, C, D, O have hinges;
the point O is fixed (this is Hart's cell). If
under these conditions we deform the
anti parallelogram ABCD, then the product
OM-OM' will remain constant, that is, if
point M describes a curve L, then point
M ' will describe a curve obtained from L
by an inversion with the pole O (Fig. 96)#
Proof. If the antiparallelogram ABCD
is hinged, then, as will readily be seen,
when it is deformed it remains an anti-
parallelogram (this stems from the fact
that an antiparallelogram is characterized
by the equalities AB = CD, and AD = Fig. 96
Inversion 309
= BC). On the other hand, the straight lines OM and O M \ which are
parallel to the bases BD and AC of the trapezoid in the original position
of the figure, will all the time remain parallel to them. Indeed,
AM _ AO BM' __ BO
~AD ~~ AB BC ~ A B '
When the antiparallelogram ABCD is deformed, these proportions are
preserved. In particular, from what has been said, the points O, M and
M ' remain on one straight line. Furthermore, from similar triangles we find
OM _ AO_ OM' __ AC
BD ~ AB 9 OB ~~ AB ’
whence
O M 'O M ' = -C^ - ? J L .B D A C .

AB*
But a circle can be circumscribed about the trapezoid ABCD, and so
AB2 + B D -AC = AD2
And, hence,
BD AC = AD2 - AB2.
Finally, we get
OM O W = 0A 0 B (AD* - AB*).
AB*
Therefore point M' is obtained from point M under the inversion [O, k],
where
_ OA OB(AD2 - AB2)
k = --------------------------------------------.
AB2
Sec. 5. The geometry of Mascheroni

We will assume that a straight line is specified if two distinct points
of the line are fixed in the plane. A circle is regarded as specified if the fol
lowing are given: the centre and also a point lying on the circumference of
the circle, or three points on the circumference. We will denote a circle with
centre O and radius OM as (O, OM). A circle specified by three points
A, B, C lying on the circumference will be symbolized as (ABC).
We now consider the solution of a number of basic problems involving
construction with a compass alone. Most of the solutions involve the use
of inversion. *
Starting mainly with these constructions, which were carried out without
the aid of inversion, the Danish mathematician G. Mohr (17th century)
succeeded in proving that with a compass alone it is possible to perform all

constructions that can be performed with compass and straight edge.
This proposition was independently proved at the end of the 18th cen
tury by the Italian mathematician L. Mascheroni, and it was precisely Mas-
cheroni’s work that became known in Europe (Mohr’s book had gone un
noticed). For this reason, constructions peformed with only a compass are
frequently associated with the name of Mascheroni (the geometry of Mas
cheroni, Mascheroni constructions and so forth).
We now consider several construction problems involving only a com
pass.
Problem 1. A line segment AB is specified by its endpoints. Construct,
on the extension of segment AB, beyond point B, a point C such that AC =
= n-AB, where n is a natural number.
Solution. We construct the circles (A, AB) and (B, BA). Let P be one
of the points of their intersection. Construct a circle (P, PA)\ let Q be
the second point of intersection of this circle with (B, BA). We construct
the circle (Q, QB)\ let R be the second point of intersection of this circle
with (P, BA). The point R lies on the extension of segment AB beyond
B, and AB = BR (Fig. 97).
True enough, the indicated construction yields the vertex R of a regular
hexagon; R is opposite vertex A and the hexagon is inscribed in (B, BA).
Continuing similar constructions, we can construct point C that lies on the
extension of segment AB beyond point B and such that AC = n-AB,
where n is any natural number.
Fig. 97 Fig. 98
Problem 2. Construct with compass alone the image M of a point M '
under the inversion [O, r2].
Solution. 1°. OM > r/2. If point M lies on the circle of inversion, then
its image M ' coincides with itself. Therefore, let us assume that point M
lies either inside the circle of inversion (Fig. 98) or outside it (Fig. 99).
In both cases we construct the circle (M, MO). Let X and Y be the points
of intersection of this circle with the circle of inversion. We construct the
circles (X , XO) and (F, YO); the second point of intersection of these circles
is the point M f. Indeed, since the points X and Y are symmetric about the
Inversion 311
Fig. 99
straight line MO, and the circles (X, XO) and ( Y, YO) are congruent, it
follows that the second point M ' of their intersection lies on the radial
line OM. Furthermore, if H is a point diametrically opposite point O on
the circle (M, MO), and E is the point of intersection of X Y and OM (E
is the midpoint of segment OM' ; neither point H nor point E need be con
structed, they are introduced merely to aid the proof), then
r2 = OX2 = O H O E = OMOM' .
2°. OM < r/2. In this case (M, MO) does not intersect the circle of inversion.
We construct on the radial line OM a point N such that rrO M = ON and
such that ON > rjl. Construct point N' obtained by the inversion [O, r2]
from point N (case 1°) and then construct on the radial line ON' a point
M ' such that OM' = rrON'; then
___ ___ ON ___
O M •OM' = ------ •nON' = ON•ON' = r2.
n
From the solution of this problem there follows the possibility of con
structing with compass alone a point C lying on segment AB and such that
n*AC = AB. Indeed, construct on the extension of segment AB, beyond
point B, a point C' such that AC' — rrAB (problem 1). Let C be the image
of point C' under the inversion [A, AB2] (problem 2). Point C is the desired
one. Indeed,
AC AC' = AB2, AC nAB = AB2, n-AC = AB.
Problem 3. Using only a compass, construct a circle C as the image,
under the inversion [O, r2], of the straight line C that does not pass through
the pole O of the inversion. The straight line is assumed to be given by two
distinct points X and Y lying on it.
Solution. The circle C' passes through

point 0. Let P be the second point of
intersection of the circles (X, XO) and
(Y, YO). Then P is symmetric to point
0 with respect to the line XY. Construct
(with compass alone) the image P' of
'i---=----t.t--h~--~-iP point P under the inversion [0, r 2]. The
desired circle C' is the circle (P', P'O)
(Fig. 100). Indeed, if E is the point of
intersection of the straight lines XY and
OP [E is the midpoint of OP and £' is
a point diammetrically opposite to point
Fig. 100 0 on circle (P', P'O)] then
r 2 = OP·OP' = OE·OE'.
Problem 4. Using only a compass, construct the point of intersection

of two straight lines, one of which is given by the points A and B, the other
by the points C and D.
Solution. Construct an arbitrary circle S in the plane. Construct (with
compass alone) the circles K 1 and K2 , which are inverses of the given lines
with respect to the circle S. Let M' be the point of intersection of the circles
K 1 and K 2 • Construct (with compass alone) the inverse M of point M'
with respect to the circle S. The point M is just the point of intersection of
the given straight lines.
Problem 5. Construct (with compass alone) the centre of a circle spe-
cified by three points A, B, C lying on that circle (the circle itself is not
drawn).
Solution. Let P and Q be the points of intersection of the circles (A, AB)
and (B, BA). The straight line PQ is the midperpendicular of segment AB.
Let L and M be the points of intersection of (B, BC) and (C, CB). The
line LM is the midperpendicular of segment BC. The centre 0 of the
circle (ABC) is the point of intersection of the straight lines PQ and LM
(problem 4).
Problem 6. On a straight line given by two points A and B, lay otr
from point B segments BC1 and BC2 equal to the given segment PQ.
Solution. Construct a circle (B, PQ). The problem reduces to seeking
the points of intersection of that circle with line AB. Construct (with
compass alone) the images K 1 and K2 of line AB and the circle (B, PQ)
under inversion with an arbitrary circle of inversion. Let c; and C~ be
the points of intersection of the circles K 1 and K2 • The inverses C1 and C~
of points c~ and c~ are the desired points.
Problem 7. Take three given line segments PQ, MN, RS and construd
a fourth, proportional to them, that is, such that
PQ· MN= RS· XY.
Inversion 313
Solution. 1°. PQ ,p MN. Construct circles (0, PQ) and (0, MN),
where 0 is an arbitrary point in the plane. On (0, PQ) take an arbitrary
point A and construct (with compass alone) a point Bin which the straight
line OA intersects (0, iVtN) (take the point B which lies on the radial
~
line OA). Construct a circle (0, RS) and on it take an arbitrary point C.
Construct a circle (ABC) (problem 5). Let D be the second point of inter-
section of this circle with line OC (it is constructed with a compass alone).
Then
OA· OB = OC· OD
or
PQ· MN = RS· OD;
OD is the desired line segment.
2°. PQ = MN. In this case we consider segments PQ and 2MN (seg-
ment 2MN is constructed with compass alone). Construct (with compass
alone) the segment X1 Y1 such that
PQ· 2MN= RS· X1 Y1 (case 1°).
The desired segment XY = X1 Y1 is constructed by compass alone (see

2
remark at the end of the solution of problem 2).
Problem 8. Given in the plane two points A and B. Construct (with
compass alone) a point C such that the angle ABC is equal to 90°.
Solution. On the straight line AB we construct (with compass alone)
the line segment BD = AB. Let C be any one of the points of intersection
or the circles (A, AD) and (D, DA). The point C is the desired point.
Problem 9. Given a segment AB with points A and B. Let n be an
arbitrary natural number not the square of any natural number. Construct
tlw segment ABVn.
Solution. Construct (with compass alone) a point C such that L ABC=
90°. Find the point P of intersection of BC with the circle (B, BA).
Then AP = ABV2. Then construct (with compass alone) a point D such
that L APD = 90° and construct the point Q of intersection of the straight
line PD with the circle (P, AB). Then PQ = ABV3 and so on.
From the results of this problem it follows that it is possible with compass
alone to construct on a circle (specified by three points) the vertices of
a regular triangle, a square, a regular pentagon and hexagon.
Sec. 6. Inversion of space

The inversion [0, k] of space (more precisely, of spherical space with
a single ideal point, or point at infinity, adjoined) is defined in the same
w;~y as the inversion of a plane. The inversions [0, k] of space, like those
of a plane, separate into positive inversions (k > 0) and negative inver

sions (k < 0) depending on the sign of the power k of inversion.
Under a positive inversion [<0, k], a sphere S0 with centre 0 and radius
]fk (termed the sphere o f inversion) consists of invariant points.
The set of points lying outside the sphere S0 is mapped into the set of
interior points of 50, and the set of all points lying inside S0 is mapped
onto the set of all points lying outside S0.
If k < 0, then a sphere with centre 0 and radius ^\k\ is invariant,
and each point of it is mapped into a point diametrically opposite on
the sphere S0.
If the inversion [0, k] is positive, then all spheres intersecting the sphere S0
of inversion orthogonally, and only such spheres, are invariant.
The inverse of a sphere not passing through the centre of inversion
is a sphere.
The tangency of spheres and the tangency of a sphere and a plane are
preserved under inversion.
If a sphere S with centre Q passes through the centre 0 of inversion,
then its image, under the inversion [0, &], is a plane orthogonal to the
straight line OQ, which plane does not pass through the pole of inversion,
and, conversely, a plane that does not pass through 0 goes into a sphere
that passes through 0.
The inversion of space is a conformal transformation, that is, under
the inversion, angles between tangent planes in the points of intersection
of two spheres are preserved.
A circle that does not pass through the pole of inversion goes into a
circle. Indeed, every such circle may be regarded as the line of intersection
of two spheres not passing through the pole of inversion.
Under the inversion of space, angles are preserved between intersecting
circles (which, generally speaking, lie in distinct planes). Indeed, let a
and b be tangents to the circles Cx and C2 in the point of their inter
section M. It suffices to prove that under the inversion [0, k] the angle
is preserved between the straight lines a and b. We consider only that
case where the straight lines a and b do not pass through the pole of
inversion 0. Let a* and b* be straight lines that are respectively parallel
to straight lines a and b and that pass through the pole of inversion 0 .
Under the inversion [0, £], straight lines a and b go into the circles a’
and b' that are respectively tangent to the straight lines a* and b* at
point 0 . Therefore, the angle between the circles a’ and b' is equal to the
angle between the straight lines a* and b*, that is, it is equal to the angle
between a and b because a* || a, 6*|| b.
* *
We now introduce a number of geometric constructions and analytic
derivations associated with stereographic projection in connection with
the use of this mapping of a sphere on a plane when making maps
Inversion 315
of the earth. We dwell solely on the geometric construction of the depiction

of the network of meridians and parallels and on the analytic investigation
of the nature of their distribution (under stereographic projections).
Let us recall the basic definitions and properties of a stereographic
projection in connection with the inversion transformation.
Let A and B be the ends of some diameter AB of a sp h ere^ ; ax andZ?!
are planes tangent to Qx at the points A and B. Let M be an arbitrary
point of Qx distinct from point A . Denote by M ' the point in which the
straight line AM intersects the plane bx. The correspondence under which
point M is associated with point M f is termed the stereographic projection
of the sphere Qx on the plane bx (Fig. 101).
Let us consider a section Q of the sphere Qx cut by the plane ABM
(Fig. 102). From the similarity of A ABM and A ABM' we find
AM- AM ' = AB2.
From this relation it follows that the image M f of point M under the
stereographic projection of the sphere Qx on the plane bx is also the image
of point M under the inversion / = [A, AB2] with pole A and the power
of inversion equal to AB2; in other words, point M ' is the image of point M
under an inversion with the sphere of inversion cox the centre of which
is point A and radius AB. To construct the image M ' of point M under
the inversion I it is first convenient to construct sections Q and co of
the spheres Qx and a>1 by the plane ABM. Then join point M with point A
and draw a straight line passing through point M perpendicular to straight
line AM. Let P be the point in which this line intersects the circle co;
then the tangent to co at the point P will cut line AM at point M'. This
construction is true of any point M lying inside the circle co. If point M
lies on co, then its image is the point M itself. If M lies outside co, then
to construct the image M ' under the inversion with respect to circles co,
we construct the tangent MQ drawn from point M to co (Q is the point
of tangency); the orthogonal projection M ' of point Q on line AM is
precisely the image of point M under the inversion I. All these construc
tions can be carried out in any section Q of the sphere Qx by a plane
passing through the straight line AB. Incidentally, they can also be carried
out in space by drawing the appropriate tangents not to the circles Q
and a), but to the spheres Q± and a>x.
From the foregoing we arrive at the following important conclusion:
on the set of all points of (with the exception of point A), a stereo-
graphic projection coincides with an inversion, the pole of which is the
centre of projection A, and the power of the inversion is equal to the
square of the diameter of in other words, a stereographic projection
on the set of all points of (with the exception of point A) coincides
with an inversion with the sphere of inversion col9 the centre of which is
point A, and the radius is equal to the diameter of Qv
From this it follows that the geometric properties of a stereographic
projection may be obtained from the familiar properties of inversion,
namely: under an inversion (and, hence, under a stereographic projection)
the circles that lie on the sphere and do not pass through point A
go into circles (lying in the plane b2). Since inversion is a conformal trans
formation (that is, angles are preserved), it follows that a stereographic
projection has the same property, so that using a stereographic projection
makes it possible to construct a conformal map of the globe (it is not,
however, possible to construct a map of the earth on which distances
are preserved).
Let us now see how one constructs the centre of circle C", into which
circle C, which lies on the sphere and does not pass through p oints,
is mapped. Let us first suppose that C is not a great circle of We
construct a cone K tangent to along the circle C. Let P be the vertex
of the cone. We construct a sphere T with centre P that passes through
circle C (Fig. 103). This sphere intersects Qx orthogonally. Under the
inversion I = [A, AB2], the sphere goes into the plane bl9 and the
sphere Tgoes into the sphere T \ which will intersect the plane bx ortho
gonally. From this it follows that the centre M ' of T' must lie on the plane
But, on the other hand, the centre of sphere T must also lie on line AP
Inversion 317
since sphere T is the image of A

sphere T under the inversion
[A, AB2]. Thus, the centre M ' of Q
sphere T' is a point in which
the straight line AP intersects
the plane bv Furthermore, the
circle C is the line of intersection
of the spheres Qx and T, hence
the image C" of circle C (both 6
under the inversion / and under Tf M T{ B
the stereographic projection Fig. 104
under consideration) is the intersection of the sphere T and the plane bx.
But since the centre M ' of the sphere T lies in the plane bl9 it follows
that C' is a great circle of sphere T', and therefore the centre of circle
C coincides with the centre M ' of sphere T , that is, it is the projection
M ' on plane b± of the vertex P of the cone K that is tangent to the sphere
Q1 along the circle C.
Note that the projection M ' of point P on the plane bx is not the image
of point P under the inversion / = [A, AB2].
To construct the centre and the radius of the circle C", let us consider
the section Q of sphere Q1 by the plane ABP (Fig. 104). Let T[9M and T%
be the centre projections of points Tl9 P, T2 from point A on the straight
line b (Tx and T2 are the points of tangency of tangents drawn from
point P to the circle Q). Then M is the centre of the circle C' and its
radius is equal to MT[ = M T2.
Let us now examine the case where the circle C is a great circle of Qx
and does not pass through point A. In this case, on the basis if the general
properties of inversion, the stereographic projection C of the circle C
on the plane bx will again be a circle. Geometrically, its centre and radius
are constructed as follows: the plane n in which circle C lies goes, under
the inversion I = [A9AB2], into the sphere n'9 the centre of which lies on
a straight line passing through point A perpendicular to plane n. On
the other hand, since the plane n is orthogonal to the sphere Ql9 it follows
that the sphere n must be orthogonal to the plane bx and, hence, the
centre M of the sphere n' must lie in the plane bv Thus, the centre M of
the sphere n and, hence, of the circle C' as well (which is a great circle
of the sphere n') is the point of intersection with the plane bt of a straight
line passing through point A perpendicularly to the plane n in which the
circle C lies.
Figure 105 shows a section Q of the sphere cut by a plane passing
through straight line AB and through the perpendicular dropped from
point A to the plane n in which circle C lies; DE is a diameter of circle C
along which the drawn plane intersects C. Joining point A with points D
and E, we obtain, in the intersection with line b, the points D' and E \
which are maps of the endpoints of diameter DE of circle C; the straight
line^L, which is perpendicular to the diameter DE, intersects (as indicated
above) the straight line b in the point M \ which is the centre of circle C".
Since M'D' = M 'E ' = M 'A, it follows that to construct the centre and
the radius of circle C' we need not draw the straight lines AD and AE:
by dropping the perpendicular AL from point A on the diameter DE of
circle C we obtain both the centie M ' of circle C" and the radius AM '
of that circle. Also note that the orthogonal projection L of point i on
the straight line DE lies on the circle with diameter OA (O is the midpoint
of line segment AB). This reasoning is used below in the geometric cons
truction of a spectrum of meridians when constructing maps of the western
and eastern hemispheres of the earth.
It remains to consider the case where the circle C passes through the
centre A of projection. In this case, the stereographic projection of circle C
is the straight line C' along which the plane b1 intersects the plane n (in
which circle C lies) (Fig. 106).
Let us now examine the construction of maps.
The geographic coordinates of a point lying on the earth’s surface are
called the latitude (p and the longitude 0. The lines on which the latitude (p
is the same, (p = constant, are termed parallels. These are sections of
the earth by planes perpendicular to the axis N S of the poles. The lines
on which the longitude 8 has the same value, 0 = constant, are termed
the meridians of the earth. They are semicircumferences of the great circles
of the earth, the boundary points of which are the poles N and S . The
network of meridians and parallels of latitude on the earth’s surface is
an orthogonal network. Figure 107 depicts, in axonometric projection,
the parallels of latitude and the meridians of a sphere.
Ordinarily, two methods are employed in the construction of a map
o f the earth with the use of stereographic projection: in one method the
map of the northern and southern hemispheres is done as follows. Let N
and S be the north pole and south pole, respectively, and let n and s
be the planes tangent to the earth at these poles. To construct a stereo-
graphic projection of the northern hemisphere, one projects this hemisphere
from the south pole S on the plane tangent to the earth at the north pole,
and to construct a map of the southern hemisphere, one projects that
Inversion 319
Fig. 107 Fig. 108
hemisphere from point N on the plane s. The result is a map made up

of two circles: one is a map of the northern hemisphere, the other is a
map of the southern hemisphere. Of course, this construction is subjected
to a similarity transformation. Figure 108 gives, in axonometric projection,
just such a construction of the map of the southern hemisphere.
In such a mapping of the earth, the parallels of latitude of the southern
hemisphere, for example, are projected into concentric circles lying in
the plane s and having the common centre S ; the equator o1 is projected
into the circle a inside which are the projections of all parallels. The semi
meridians of the southern hemisphere (or, similarly, of the northern hemi
sphere) are projected into radii of circles (on maps, this construction is of
course supplemented by a similarity transformation). If we change the
longitude 9 at equal intervals (from 0° to 180°), then the corresponding
radii of the circles a will turn in succession through one and the same
angle. Figure 109 depicts the construction of meridians under a change
Fig. 109 Fig. 110

780'
165'
15
105° 30° 75' 105° go - ^
F ig . I l l
in longitude 0 of 15°. When constructing parallels corresponding to

identical intervals of latitude (from 0° to 90°) one can do as follows:
construct the section of the sphere (Fig. 110) by some plane passing through
the straight line NS. Consider, for example, the southern hemisphere.
Let PSQ be the semicircle of the southern hemisphere along which the
drawn plane intersects the hemisphere, and let s be the straight line of
the intersection of this plane with the plane sx. Divide the semicircle PSQ
into several equal parts (in Fig. 110 they are indicated by points with
latitudes 0°, 15°, 30°, 45°, 60°, 75°, 90°) ; then the projections of the points
of division from point N on straight line s with the same latitudes (0°0°,
15°15°, 30°30°, 45°45°, 60°60°, 75*75°, 90°90°) yield the diameters of
the parallels. Construction of the parallels can be carried out together
with that of the meridians. Just such a construction has been carried out
in Fig. 110. Since the circle a (the representation of the equator) is
homothetic to circle (NS) with diameter N S (the homothetic ratio is
equal to 2), it follows that when constructing meridians and parallels, one
need not construct the circle (NS). That is the construction made in Fig. I ll
for both hemispheres.
This geometric method of constructing meridians and parallels brings
us to the conclusion that the distribution of meridians or, as we shall say,
the spectrum of meridians, is uniform and the spectrum of parallels expands
as we approach the representation a of the equator (for the time being
^ve note this property visually for parallels). We will give an analytic proof
that the spectrum of parallels expands near the equator.
Let M be an arbitrary point of the sphere Q, (p the latitude of point M ,
and M ' the stereographic projection of point M from point N on the
plane sv Let us now construct a section of the sphere Q1 by the plane N SM
(Fig. 112). Since
Inversion 321
Fig. 112 Fig. 113
and, consequently, (we assume that N S = 1)
Now let us prove that if we take three points with latitudes <p1 < q>2 < <P3
such that (p2 — <Px = <p3 — <p2, then the corresponding values of the func
tion x , that is,
will be connected by the relation x 1 — x 2 > x 2 — x3. This then is proof

that the spectrum of representation of parallels is expanded near the repre
sentation of the equator (Fig. 113). The inequality x± — x 2 > x 2 — x3t
which we want to prove, is equivalent to 2jc2 < x± + x3 or
or
1 — sin cp2 2 cos <p2
cos <p2 cos <Ps - <P1
2
That is,
1 — sin <p2 ^ _______ cosq)2_______
cos C0S<P?JZh-+ sin <p2
2
2 1 -8 1 0
— <Pi
Since cos <p2 > 0, cos - > 0, sin (p2 > 0, it follows that the inequa
lity is equivalent to the following:
(1 — sin q>2) ^cos <Pa~<Pi1 + sin q>2jc c o s 2 (p2
or
(D., (toy
cos —-----— + sin q>2 < 1 + sin <p2.
That is.
__ (P s-(pi
cos----- ^ , 1.
------<
2
This inequality is valid since 0° < <Pz - <Pl < 90°; the inequality x x —
— x 2 > x 2 — x3 is equivalent to it and, hence, it too is valid.
Remark. The very same result can be obtained by using derivatives.
We have
— (i+ i)-
X = — < 0.
- ( t +t )
Consequently, on the interval [0, n/2], the function x = cot ( - - + — |
V4 2)
is a decreasing function. Furthermore,
I t H )
* = >0, (pe [0, n/2].
and so on the interval [0, n/2] the graph of the function x =

is convex down and this means that on that interval the Jensen inequality
holds true:
X! + X2
x2 <
2
Inversion 323
That is, xx — x 2 > x 2 — x 3 > 0 (x± — x 2 > 0, x 2 — x 3 > 0 since 0° ^

< (Px < q>2 < <Ps < 90°, and on the interval [0, ic/2] the function x =
= cot ( — + — ) is a decreasing function).
V4 2)
The established fact of the spectrum of parallels becoming sparser as
one approaches the equator can also be substantiated approximately:
the accompanying table contains the values of the function x = cotl — +
+ for the following values of q>: 0°, 15°, 30°, 45°, 60°, 75°, 90° and
the values of Ax = x t — x t +1 (/ = 1,2, 3,4, 5, 6). We see that the values
of Ax diminish as the latitude increases, that is, as we recede from the equator
<p X Ax
0° 1,000
0.2327
15° 0.7673
0.1899
30° 0.5774
0.1632
45° 0.4142
0.1463
60° 0.2679
0.1362
75° 0.1317
0.1317
90° 0.0000
Let us now examine the stereographic projections of the western and

eastern hemispheres. Let a be the meridian from which we reckon the
longitudes 0, and W V the diameter of the earth perpendicular to the plane
in which the meridian a lies. The stereographic projection of the western
hemisphere (Fig. 114 shows the lower half of the sphere) is obtained by
projecting the western hemisphere from point W on the plane j8, which
is tangent to the western hemisphere at the point V. Since such projection
coincides with the inversion 1 = [W ,W V2\ for points of the sphere Ql9
it follows that the meridians and the parallels, as forming an orthogonal
network on the earth, will project into an orthogonal network of circles
on the plane Here, the meridians of each of the hemispheres (in Fig. 114
we have the western hemisphere) will project into arcs of circles passing
through the points N and S for which these points serve as boundary
points. Note that the spectrum of representations of the meridians (of
any one of the hemispheres) is exactly the same as the spectrum of parallels
F ig . 1 1 4
under the foregoing stereographic projection of parallels of the northern

and southern hemispheres. However, the centres of the arcs representing
the meridians of each of the hemispheres must be constructed in accord
with Fig. 115 which shows that construction (a map of the western and
eastern hemispheres). As for parallels, their construction should be based
on the following reasoning: since meridians and parallels form an ortho
gonal network, this network must also remain orthogonal in the repre
sentation of the sphere under a stereographic projection. In particular,
the parallels must be orthogonal to the representation of their principal
meridian a. Besides, the centres of parallels lie on the axis N S of the N
and S poles. From this we obtain the following method of a geometric
construction of a parallel passing through a given point M of the principal
meridian a : at point M draw a tangent to the circle a (to the representation
of the principal meridian). Suppose P is the point in which this tangent
line intersects the straight line NS; then the circle with centre at point P
and radius PM will be the representation of the parallel passing through
point M (Fig. 116). In Fig. 117 are constructed the representations of
the parallels (and meridians, in accord with Fig. 115, that correspond to
the values of latitude 0°, 15°, 30°, 45°, 60°, 75° for both hemispheres —
western and eastern).
From Fig. 117 it is evident that the spectrum of parallels expands near
the poles. We will now prove this analytically. Let q> be the latitude of
point M (Fig. 116). Setting N S = 2, we have PM = cot <p, OP = cosec q>
and, hence,
v = OK = OP — PM = cosec cp — cot cp = ----- CQS ^ = tan — •

sin cp 2
This function y = tan((p/2) is an increasing function on the interval [0, tz/2]
and as <p varies from 0 to n/2, the function y — tan(<p/2) increases from 0
to 1. We will prove that the values of the latitudes <pl9<p2 9<p3 , which
have equal increments (on the interval 0 ^ cpx < (p2 < cp3 ^ n/2; cp2 —
F ig . 11 7
— q>! = q>3 — (p2 > 0), that is, <p2 — (<px + <?3)/2, are associated with
the values of the function
Ti = tan (<pJ2), y2 = tan(<?>2/2), y 3 = tan((/>3/2)
such that y 2 — yt < y 3 — >’2 or that 2y2 < y\ + y3. Such is the proof
that the spectrum of parallels becomes less dense near the poles. Thus,
we have to prove that
2 tan(<p2/2) < tan(<pj/2) + tan(<p2/2),
that is,
2 sin(<jg2/2) < s in ^ ! + <p3)/2)
cos(<p2/2) cos(('/)1/2) cos(<p3/2)
or
2 sin(^>2/2) sin <p2
cos(q>2/2) cosOx/2)cos (2,
that is.
2
This inequality is valid since 0 < (<p3 — <px)/2 < n/2; the inequality
y2 < (yx + y3)/2 is equivalent to it and, hence, is true as well.
The same values can be obtained if one uses derivatives. We have
y = tan(<jo/2).
---------------> 0, <Pe [0, n/2].

2 cos2((p/2)
and so the function y = tan(<p/2) increases on the interval [0, nj2]. Fur
thermore,
n sin(W2) > o e [0, tt/2].
y
2 cos3(<p/2)
consequently, the graph of the function y = tan(<p/2) on the interval
[0, 7c/2] is convex down, and therefore on that interval the following Jensen
inequality holds true: y 2 < (ji + y 3)/2, where yl9y2,yz have the above-
indicated values.
Inversion 327
Finally, the fact that the spectrum of parallels expands near the poles
is confirmed by the following table too, in which are given the values of
the function y = tan(<p/2) at 15° intervals of latitude and the values of
Ay = J i +1 — y% O' = 152, 3,4, 5,6).
<p y Ay
0° 0.0000
0.1317
15° 0.1317
0.1362
30° 0.2679
0.1463
45° 0.4142
0.1632
60° 0.5774
0.1899
75° 0.7673 \
0.2327
90° 1.0000
Incidentally, all of this immediately follows from the fact that the spec
trum of the function y = tan(<p/2) repeats “in inverse order” , as q> varies
from 0° to 90°, the spectrum of the function
x = cot (p <P_
+
2 )
^compare the tables of the values of the functions y = tan —and x =
If we complete the representations of the meridians and parallels to full

circles when constructing maps of the western and eastern hemispheres
(in stereographic projection), then we obtain a set of all circles passing
through the points N and S (meridians) and a set of all circles orthogonal
to the representations of the meridians (parallels) (Fig. 118).
The set of all circles passing through two fixed points N and S is called
an elliptic pencil o f circles; the points N and S are then termed base points.
The set of all circles, each of which is orthogonal to all circles of the elliptic
pencil with base points N and S is termed a hyperbolic pencil o f circles,
which is the conjugate of the elliptic pencil; the points N and S are termed
limit points of the hyperbolic pencil, or Poncelet points of such a pencil.
When representing a network of meridians and parallels under a stereo-
graphic projection of the western and eastern hemispheres, we have to
Fig. 118
isolate from the elliptic pencil of circles and the conjugate hyperbolic pen
cil that part contained within the circle of the elliptic pencil, whose dia
meter is the line segment NS, where TVand S are base points of the elliptic
pencil.
A stereographic projection of a network of meridians and parallels may
be carried out by chosing, on a sphere, any two diametrically opposed
points A and B and projecting the sphere from point A on a plane tangent
to the sphere Qx at the point B. Since a stereographic projection is a con
formal mapping of the sphere on a plane, it follows that the orthogonal
network of meridians and parallels will be mapped into the same kind
of orthogonal network of two pencils of circles: the meridians are mapped
into an elliptic pencil with base points N and S, and the parallels into the
conjugate hyperbolic pencil with limit points N and S. The projection
of the principal meridian is some circle K that passes through the points N
and S, and N S is a chord of that circle. A map of one of the hemispheres
(western or eastern) will lie inside circle K. The construction of the map
of the other hemisphere is similar.
If we isolate any one of the circles (say circle a) from the hyperbolic
pencil of circles and regard it as a representation of the equator (and
regard the Poncelet point lying inside the chosen circle a as a represen
tation of the pole), then we obtain a map of the northern or southern
hemisphere. Representations of the parallels are all the ciicles of the hyper
bolic pencil that lie inside the circle a, and the semimeridians are arcs of
Inversion 329
the conjugate elliptic pencil that lie inside the circle a and emanate from the
pole (located inside a). In this situation, these semimeridians will join
up into an arc of one circle if the sum of the longitudes (east and west)
is equal to 180° (or, to put it differently, if the difference between the posi
tive and negative longitude is equal to 180°).
Such is the qualitative picture of representation of meridians and parallels
under an arbitrary stereographic projection of the western, eastern, northern
and southern hemispheres.
We will now show how an exact geometric construction is performed of
the spectrum of meridians and parallels under an arbitrary stereographic
projection.
Let us begin with the construction of the spectrum of meridians and
parallels when representing the western (or eastern) hemisphere; we
assume the latitude and longitude to vary in 15° intervals. Consider the
section Q = ANBS of the sphere by a plane passing through the dia
meters AB and SN (Fig. 119). Let N and S be projections of points#
and S from point A on a tangent to the section Q at the point B. Since
the meridians pass one into another (and the parallels are invariant) under
rotation of the sphere about the axis SN, it follows that any circle with
chord N S may be taken for the representation a of the principal meridian
with a supplement to complete the meridian to a full circle. Since a stereo
graphic projection is a conformal mapping of the sphere on a plane, in
order to construct the representation of the meridians in equal intervals of
longitude, one has to construct arcs of circles whose centres lie on the
midperpendicular of line segment NS, the centres lying inside the repre
sentation a of the principal meridian and forming a sequence of equal
angles. For example, if the longitude changes in 15° intervals, then we
obtain the representations of thirteen meridians (the circle a is a repre
sentation of two meridians with longitudes of 0° and 180°), which intersect
in succession at 15° angles. In Fig. 120, the constructions have been carried
out in superposed planes: the plane of the circle Q is brought to coincidence
with the plane tangent to the sphere Qx at the point B (the coincidence
is attained by rotating about the straight line along which these two planes
intersect).
For a geometric construction of the representation of these meridians,
note that the angle between two intersecting circles is equal to the angle
between their radii drawn to the point of intersection of the circles. For
this reason, to construct the radii and centres of the representations of
the meridians, turn the radius NO of the circle a about point # in a succes
sion of, say, 15° angles. To do this, construct the semicircle CDE with
centre N, whose diameter CE is tangent to the circle a at the point N
(see Fig. 119). Divide the arc CDE into 12 equal parts and project the points
of division from point N onto the midperpendicular of line segment NS.
These projections T are the centres of the representations of the meridians
(which correspond to 15° intervals of longitude), and TN are their radii.
Fig. 119
Inversion 331
Note that the idea of conformity of a stereographic projection that was

made use of in this construction could have been utilized earlier in the
construction of the spectrum of meridians in a simpler case.
To construct the representations of parallels at, say, 15° intervals of
latitude, one should take advantage of the method of constructing the
centre of a stereographic projection of a circle that was given above for

an arbitrary diameter AB of the sphere Qx with respect to the circle lying
on that sphere. Let OT be a radius of the circle Q perpendicular to S ^ .
Divide the arc S j^TNj^ into 12 equal parts. Let M be one of the points of
division (say, the closest one to Nx). The tangent to the circle Q at point M
intersects the straight line Sx in the vertex P of a cone tangent to
the sphere Qx along a parallel passing through point M. The projection P'
of point P from point A onto the straight line N S will be the centre of
the representation of that parallel. If PL is the tangent, for example,
to the representation a of the principal meridian (L is the point of tan-
gency), then PL is a radius of the representation of the parallel at hand
(and P is its centre). Thus, Fig. 120 depicts the representations of half
the parallels and half the equator (the centre of the representation of
the equator is the point of intersection of the straight line N S with the
straight line passing through point A parallel to NXS^).
Let us now consider the construction of the spectrum of meridians and
parallels under a stereographic projection, say, of the northern hemisphere
for an arbitrary position of the diameter AB of the sphere Qx with respect
to the equator.
Again construct the section Q of sphere by the plane AN1BS1 and
let DE be a diameter of the circle Q along which the plane of the section
cuts the plane of the equator. Let N and S be the projections of the
points N-l and Sx from point A on the tangent line to the circle Q at
the point B. To make a representation of the equator, drop from point A
a perpendicular to the diameter DE. Let M be the point of intersection of
this perpendicular with the tangent to the circle Q at the point B. Then M
is the centre of the representation of the equator, and MA is the radius
of that representation.
We will perform all subsequent constructions in the superposed planes:
the plane of the circle Q with the plane tangent to the sphere Qx at point B
(as above, coincidence is attained by turning about the straight line along
which these two planes intersect).
To construct the representations of parallels at, say, 15° intervals of
latitude, divide the arc EN1 of the circle Q into 6 equal parts. Let L be
one of the points of division (for example, the one closest to E). If Q is
the point of intersection of the tangent to the circle Q at point L with the
straight line SxNl9 then Q is the vertex of a cone that is tangent to the
sphere ^ along the parallel passing through point L. The projection R
of point Q on the straight line BN from point A is the centre of the repre
sentation of the parallel that passes through point L. Now the radius
of the representation of this parallel is equal to the line segment RE of
the tangent drawn to the representation of any meridian (that is, to any
circle passing through the points N and S). The foregoing method is
used to construct the parallels of the northern hemisphere for 15-degree
intervals of latitudes from 90° to 0°.
Inversion 333
To construct the representations of meridians, note (we have already

mentioned this) that when the sphere is rotated about the axis N ^ , the
meridians pass into one another (while the parallels are invariant). There
fore, in order to construct semimeridians we can take, for the two principal
semimeridians that correspond to longitudes of 0° and 180° and form a
single arc of the circle, any arc of the circle that passes through the points N
and S and lies inside the representation of the equator.
Construct a semicircle with centre N, divide it, say, into 12 equal
parts and project the points of division from point N onto the midperpen
dicular of the line segment SN (exactly the same approach was used above
to construct the representations of the meridians under a stereographic
projection of the western and eastern hemispheres). The projections
will be the centres of semimeridians emanating from point N and forming
a succession of 15-degree angles. The semimeridians, the difference of
the longitudes of which is equal to 180° (180° — 0° = 180°, 165° — (—15°)=
= 180°, 150° — (— 30°) = 180° and so on) are joined in the representation
into one single arc of the circle, which arc lies inside the representation of
the equator. This occurs, for instance, in the most elementary case as well
(it is shown in the figure), where the representations of the meridians (that
is, the radii of a circle) for the values of the longitudes (the difference of
which is equal to 180°) are joined into a diameter of the circle, which is
the representation of the equator. Of course, the same also occurs on the
sphere Q1 on any one of the two hemispheres (northern or southern).
CHAPTER V
BASIC DEFINITIONS, THEOREMS

AND FORMULAS
Sec. 1. Determinants of order three
A third-order determinant is introduced by the following relation:
<*i bx cx
A= 02 ^2 ^2 — 01^ 2^3 ^ 2 ^ 3 ^ 1 " 1“ ^ 3 ^ 1 ^ 2 “ 1“ Q J * 1 CZ ^ 1^ 3 ^ 2 > (0
a3 ^3 ^3
ah b i9 cv (i = 1,2, 3) are termed the elements of the determinant.
Basic properties:
1°. A determinant remains unchanged under a transposition, that is,
under an interchange of rows and columns:
a 2 03 Gl h Cl
bl b 2 b3 = bz C2
Cl c2 C3 b
a* 3 C3
2°. A determinant preserves its absolute value but changes sign when
any two columns are interchanged; for example:
b i ax c1 01 b± Cl
b% a<± C2 = — 02 b<2 C3
«$»
C0O
«?
#3 ^3 C3
A determinant remains unchanged under a cyclic permutation of its
columns:
0 i bx Ci bi cx 0i Ci 0i b±
02 b% C2 = b% C2 02 = c3 02 b%
CI3 b$ c3 63 C3 03 C3 03 63
and changes sign if the cyclic order of its columns is disrupted [see (3)].
3°. If two columns of a determinant are the same, the determinant is
equal to zero.
4°. If all the elements of some column of a determinant are equal to zero,
the determinant is zero.
Basic Definitions, Theorems and Formulas 335
5°. A determinant is a linear function of its columns. This means that

the following relation holds true:
Xax + jukx bx c1 01 *1 01 k x bx cx
Xa2 + fxk2 b2 c2 = X ci2 b2 c2 + ^ k 2 b2 c2
Xa3 + Hâ h c3 03 * 3 03 kz *3 cz
and analogous relations for the second and third columns.
6°. A determinant remains unchanged if a linear combination of any
two columns is added to another column, that is.
ax + Xbx + [icx bx cx 01 *1 01
a2 + Xb2 + nc2 b2 c2 = <x2 b2 c2
a2 -(- Xb3 -t- nc3 b3 c3 03 * 3 03
and analogously for the second and third columns. In particular, a deter
minant remains unchanged if any column is added to or subtracted from
another column.
7°. For a determinant to be equal to zero, it is necessary and sufficient
that its columns be linearly dependent, that is, that there exist numbers
A, n, v among which at least one is nonzero and such that
Xax + \ibx + \cx = 0,
Xa2 + i*b2 + vc2 — 0, (7)
Xaz + /x*3 + vcz = 0.
In other words, for a determinant to be equal to zero, it is necessary and
sufficient that one of the columns be a linear combination of the other
two, for example:
cx = Xax + \ibi,
c2 = ?m2 + \ib2, (8)
cz— kaz + fibz.
8°. Multiplication of determinants is carried out with the aid of the
following formula:
0i *1 01 *1 yi Z1
ct2 b2 c2 *2 y 2 Z2
03 *3 03 *3 y& z 3
01*1 + * 1 * 2 + ^1 * 3 01yi + * 1 ^ 2 + 0 1 J 3 axzx + b 1z 2 + c1zz

02*1 + *2*2 + 02*3 02^1 + b2y 2 + c2yz 0 2 Z 1 + * 2 Z2 + C2 Zz • (9)
03*1 + * 3 * 2 + 03*3 0s7i + bzy 2 + czyz azzx + bzz2 + czzz
Since a determinant remains unchanged under the transpose operation,

we can obtain another three modes of multiplication of determinants if
one of the factors of the left-hand side of (9) is transposed.
The cofactor of an element of a determinant is the coefficient of tha
element in formula (1).
For example, the cofactor Cx of element cx in determinant (1) is
Ci = a2b3 a3b2.
The cofactor A2 of element a2 of the determinant (1) is
A 2 = b3Ci b±c3
and so forth [in all, there are 9 cofactors A i9 Bh Ct (/ = 1,2, 3)].
A determinant is equal to the sum of the products of the elements of
any column into the corresponding cofactors:
A — aÂ-! + a2A2 + a3A3,
and similarly for the second and third columns.
A determinant is equal to the sum of the products of the elements of
any row into the corresponding cofactors:
A = a1A1 + b1B1 + cxCl9
and similarly for the second and third rows.
The sum of the products of the elements of any column into the corres
ponding cofactors of the elements of another column is equal to zerot,
for example:
d\Bx + a2B2 + a3Bs = 0
(and there are five other analogous relations).
The sum of the products of the elements of any row into the corresponding
cofactors of the elements of another row is equal to zero, for example:
axA 2 + b1B2 + 0 ^ 2 = 0
(and there are five other analogous relations).
The minor of an element of a determinant is the determinant obtained
by deleting the row and column at the intersection of which that element lies.
For example, the minor of element cx in determinant (1) is the determinant
a2 b2
— tt2b3 #3^2
#3 b3
and the minor of element a2 of determinant (1) is the determinant
^ Cl = bxcz — b3cl9 and so on.

^3 C3
The cofactor A of any element of a determinant is equal to the minor M

of that element multiplied by (— l)i+J, where i and j are, respectively.
the number of the row and the number of the column, at the intersection
of which that element lies:
A = ( - 1)i+J M.
Sec. 2. Vector algebra

A vector is a directed line segment. Vectors are symbolized as follows:
a, b, x, r ,. .. or AB, CD, FQ,. ..
A vector AB is said to be nonzero if the points A and B are distinct;
a vector AB is said to be a zero vector if the points A and B coincide.
The zero vector is designated by the symbol 0.
Two nonzero vectors AB and CD are said to be collinear if the straight
lines AB and CD are collinear, that is, either parallel or coincident. The
zero vector is assumed to be collinear with any other vector.
If the vectors a and b are collinear, then we can write a|| b. If the vectors
a and b are collinear and in the same direction, we can write a jf b, and if
they have opposite directions we write afj,b.
The magnitude (or absolute value) of a vector AB is the length of the line
segment AB. This can be symbolized as: \AB\, AB, |a|, or a.
The nonzero vectors a and b are said to be equal if affb and |a| = |b|.
The zero vector is assumed to be different from any nonzero vector. Zero
vectors are all assumed to be equal.
The nonzero vectors AB, CD, EF are said to be coplanar if the straight
lines AB, CD and EF are coplanar to one and the same plane (a straight
line and a plane are said to be coplanar if the line is either parallel to, or
lies in, the plane). If there is at least one zero vector among the vectors
— > — >—>
AB, CD, EF, then they are regarded as being coplanar.
To lay off a vector a from a given point A means to construct a directed
line segment AB equal to the vector a.
The vector —a = BA is said to be the opposite of vector a = AB.
The sum a + b of the vectors a and b is a vector which is constructed
as follows:
from an arbitrary point A lay off the vector a:
~AB = a
and from the point B lay off the vector b:
B C = b.
2 2 -8 1 0
Then
a + b = AC.
The properties of a sum of vectors are:
a + (b + c) = (a + b) + c,
a + 0 = a,
a + ( - a) = 0,
a + b = b + a.
The difference a — b between two vectors is a vector x such that b + x = a.
To construct the difference a — b, lay off vectors a and b from some
point O:
OA = a, OB = b.
Then a — b = BA.
The product Aa o f a number A ^ 0 by a vector a ^ 0 is a vector defined
as follows: |Aa| — \X\ |a|, and we have Aa ffa if A> 0, and Aa j j a if A< 0.
If either A = 0 or a = 0, then, by definition, Aa = 0.
The properties of a product of a number (scalar) by a vector are:
1 a = a,
A(â) = (Aju) a,
A(a + b) = Aa + Ab,
(A = Aa -f~ fia.
j|
If a || b ^ 0, then the ratio — is a number A such that Ab = a. If
b
a |al
0 # a||b # 0, then — = + — , the sign being if a ttb and the sign
b |b|
being ” if a | j, b. If a = 0, b # 0, then — = 0.
b
A linear combination of the vectors a, b, c, . . . is the sum Aa + pb +
+ vc + . . .
The vectors a, b, c, . . . are said to be linearly dependent if there are num
bers A, p, v such that at least one of them is different from zero and such that
Aa + pb + vc + . . . = 0. If this equation is possible only for A = p =
= v = . . . = 0, then the vectors a, b, c are said to be linearly independent.
For the vectors a and b to be collinear, it is necessary and sufficient that
they be linearly dependent.
For the vectors a, b, c to be coplanar, it is necessary and sufficient that
they be linearly dependent.
A basis el9 e2 in a plane is an ordered pair of noncollinear vectors el9 e2

coplanar to that plane. If the vectors of the basis are unit vectors and are
mutually perpendicular, the basis is said to be orthonormal. Ordinarily,
the vectors of an orthonormal basis are designated as i, j :
i ± i , |ii = ui = i.
Suppose a basis el5 e2 has been introduced in a plane; a is an arbitrary
vector coplanar to that plane. There exists a pair of numbers x, y (and only
one pair) such that
a - xeA+ ye2.
The coefficients x and y of ej, and e2 in this expansion of the vector a with
respect to the basis el5 e2 are called the components (or coordinates) of
vector a in (or with respect to) the basis el9 e2. If the vector a in the basis
el9 e2has components x, y , then we write a — {x, y) or a {x, y}.
In space, the basis el5 e2, e3 is formed by an ordered triplet of noncoplanar
vectors el9 e2, e3.
If the vectors of a basis are unit vectors and orthogonal in pairs, then
the basis is said to be orthonormal The vectors of an orthonormal basis are
ordinarily designated as i, j, k:
i l j , j l k , k_Li, |i| = |j| = |k| = l.
If we introduce a basis el5 e2, e3 in space and let a be an arbitrary vector,
then there is only one triplet of numbers x 9 y , z such that
a = xex + ye2 + ze3.
The numbers x, y , z are called the components (or coordinates) of the
vector a in the basis e2, e3. If the vector a has components x, y 9 z, then
we write a = {x, y 9 z} or a (x, y , z}.
Two vectors a and b are equal if and only if their corresponding compo
nents are equal.
If, in the basis el5 e2, we have
a = {x,j}, b = {x’,y'},
then
a - f b = {x + x', y + / } ,
a - b = {x - x \ y - y'}.
An = {Ax, Ay}.
If, in the basis el5 e2, e3, we have
a = {x, y, z}, b = {x\ / , z'},
then
a + b = {x + x', y + y', z + z'},
a — b = {x — x', y — y \ z — z'},
/la = {Ax, Xy9Az}.
A necessary and sufficient condition for two vectors a = {x , y } and

b = {x \ y'} to be collinear in a plane is the equation
x9 /
A necessary and sufficient condition for two vectors a = (x, y, z} and
b = {x' y \ z'} to be collinear in space is the condition
y z | |z x x y
/ z’ \ iz' x ’ x’ y
A necessary and sufficient condition for the coplanarity of three vectors
a = {x ,y , z}, b = {x ’, y z'}, c = {x", y", z"}
is the equation
x y z
x' y z 9 = o.
A scalar product a-b (or ab) of two nonzero vectors a and b is the product
of their magnitudes by the cosine of the angle between them:
ab = |a| |b| cos <p.
If a = 0 or b = 0, then, by definition, ab = 0.
The properties of a scalar product of two vectors are:
a2 = aa > 0 if a # 0, a2 = 0 if a = 0,
ab = ba,
(Aa) b - A(ab),
a(b + c) = ab + ac.
The scalar product aa (ordinarily we write a2) is called the scalar square
of the vector a.
If, in an orthonormal basis i, j in the plane,
a = {x,y}, b = { * ',/} .
then
ab = xx' + y y \
and if
* = {x>y}, b = { * ',/}
in an arbitrary basis e2, e2, then
ab = guxx' + gn (xy' + x'y) + g22yy'.
where
£ij ei e7*
The collection of scalar products = e, ey- of the basis vectors e2
is termed the fundamental tensor of that basis.
If, in the orthonormal basis i, j, k,
a = {x, y, z}, b = {x', / , z'},
then
ab = x*' + yy' + zz\
and if
a = {x, y, z), b = {jt\ / , z'}
in the arbitrary basis el9 e2, e3, then
ab = gn xx' + g12 yy' + g33 zz' + g12(xyf + x'y)
+ gniyz' + y ’z) + g*i(zx' + z’x \
where gxi = e; ey (the fundamental tensor o f the basis el5 e2, e3).
A plane is said to be oriented if a basis el5 e2(with a fixed order of vectors)
is introduced in the plane and the whole set of pairs a, bof noncollinear
vectors is partitioned into two subsets in the following manner. If an ordered
pair a, b of noncollinear vectors a and b have the same orientation as that
of the basis ex, e2;
a>MT eJL>e2,
that is,
OAB 1 1 OEi#,, where OA = a, OB = b, OEx= elf OE2 = e2.

then this ordered pair a, b is called the right-hand pair, or the pair with posi
tive orientation. But if a, b and el5 e2 have opposite orientations, „
a, b U e 1? e2,
------ ^ ------ >
that is, (with notation the same as above) OAB f [ OExE2, then the ordered
pair a, b is called the left-hand pair, or the pair with negative orientation.
Suppose the noncollinear vectors a and b lie in an oriented plane. The
cross (or pseudoscalar) product (a, b) of vector a and vector b is a number
whose absolute value is equal to the area of a parallelogram with sides
OA = a, OB = b and which is positive if the ordered pair a, b is a right-
hand pair, and negative if that pair is a left-hand pair. If a j| b, then, by
definition, ab = 0.
The properties of a cross (or pseudoscalar) product are:
(a, b) = — (b, a),
(a, b + c) = (a, b) + (a, c),
(A a, b) = A(a, b).
Suppose, in an oriented plane, there are given two noncollinear vectors

a and b, the angle between them being equal to a. The angle cpformed by
vector a and vector b is the angle cp = a if a, b is a right-hand pair, and the
angle cp = —a if a, b is a left-hand pair.
If the vectors a and b are nonzero vectors and a ] f b, then, by definition.
<p = 0, and if a b, then, by definition, cp = n. The angle b = { * ',/}
in the orthonormal basis i, j, then
xx' + yy9 . xy9 — x'y
COS cp = . sm cp = ■■■■■■■
]fx2 + y 2 ]fx'2 + y'2 ]fx2 + y2 J/V2 + y '2
If the nonzero vectors a and b are given by their components in an arbitrary
basis el9 e2, then
gn x x 9 + g12(xy' + x'y) + g22yy'
COS (p = ‘ r......................................... ,7......... ■— ■ = ; ,
Vgn X2 + 2g12 xy + g22y2 Vgn x ’2 + 2g12 x 'y ’ + g22y '2
sm cp — .-----------------------—
Ygr-(xy'- - x'y)._ .—.. . Zy
Yg n x 2 + 2g12 xy + g22y 2 Yg n x'2 + 2g12 x' / + g22y '2
where g is the Gram determinant:
£ll ^12
g= = («i e2)2.
g 21 <§22
Our formulas specify a set of values of the angle cp of the form {cp +
+ 2k 7i}, where k assumes all integral values and cp is some value of the
angle formed by vector a and vector b.
Suppose vector a lies in a plane oriented by the basis el9 e2. The vector
Up is a vector obtained by rotation of vector a through the angle cp (if
0 < cp < n, then the pair a, a^ is a right-hand pair; and if —n < cp < 0,
then it is a left-hand pair; if cp = 0, then a^ = a, if cp = n9then a^ = —a).
The vector obtained by rotation of vector a through the angle n/2 is
symbolized thus: [a].
We have the following formula:
(a, b) = [a] b.
If, in the orthonormal basis !, j,
a = {x,y}, b = {x9, / } ,
then
a— y sin q>, x sin <p + y cos <p},
[a] = { - y, *},
(a,b) = * y
x' y'
If the vectors a = {x, y } and b = { x \ / } are specified in an arbitrary
basis el9 e2, then
(a, b) = \x y
Two bases el9 e2 and e1, e2 in a plane are said to be dual bases if
when i = j,
[0 when / / j.
The vectors of dual bases are connected by the relations
W [ej ^ [e2] _ [e1]
(ea.ej)' (el5 e2) ' 1 (e2, e1) ’ 2 (e1, e2)
In other notation: the vectors of dual bases a, b and a*, b* are connected
by the relations
(a,>*) = (a, b*) = 1, (a, b*) = (a*, b) = 0,
[b] b* _ [a] [b*] [a*]
, b
[b, a] (a, b) (b* a*)' (a*, b*)
An expansion of an arbitrary vector a in terms of the bases e1; e2.and e1, e2
is of the form
a = (ae^-ej + (ae2)-e2,
a = (aei)^1 + (ae2)-e2,
where in parentheses we have scalar products of the vectors ( Gibbs' for-
mulas).
The numbers a1 = ae1, a2 = ae2, that is, the coefficients of ej and e2
in the expansion of the vector a in terms of the vectors ej, e2 are called the
contravariant components of the vector a with respect to the basis ex, e2,
and the numbers a± = aex, a> = ae2, that is, the coefficients of e1, e2 in
the exapansion of vector a in terms of the basis e1, e2, are called the cova
riant components of the vector a with respect to the basis ex, e2.
If one of the vectors is specified in the basis el5 e2 by its contravariant
components.
a = x e f+ ye2 = {x, y).
and the other by its covariant components,

b = jc'e1 + / e * = [x\ / ] ,
then
ab = xx' + yy'.
[a] = f g [ - y , x ] .
= U ? n cos + Yg sin (p) + yg.,2 cos <p]
(when <p = 7i/2 we get the preceding formula).
A space is said to be oriented if a basis e1? e2, e3 has been introduced and
the set of triples of noncoplanar vectors has been partitioned into two
subsets. If an ordered triple a, b, c of noncoplanar vectors has the same
orientation as basis el9 e2, e3, then it is termed a right-hand triple, or a
triple with positive orientation. If the ordered triples a, b, c and el5 e2, e3
have opposite orientations, then the triple a, b, c is termed a left-hand
triple, or a triple with negative orientation.
The triple scalar product (a, b, c) of an ordered triple a, b, c of nonco
planar vectors lying in an oriented space is a number whose absolute value
is equal to the volume of a parallelepiped with edges OA = a, OB = b,
OC = c (O is an arbitrary point) and which is positive if the triple a, b, c
is right-hand, and negative if the triple a, b, c is left-hand. If the vectors
a, b, c are coplanar, then, by definition, (a, b, c) = 0.
The properties of a triple scalar product of three vectors are:
(a, b, c) - (b, c, a) - (c, a, b) - - (b, a, c),
(Aa + pb, c, d) = A(a, c, d) + ju(b, c, d)
(and analogously for the second and third factors).
If the space is oriented by means of the orthonormal basis i, j, k and,
in this basis,
a = {x, y, z}, b = {*', / , z'}, c = z"),
then
x y z
(a, b, c) = x ' y* z ’
x" y" z"
and, in an arbitrary basis,
(a, b, c) = Yg x y
//
where g is the Gram determinant

£ n £12 £13
£= £21 £22 £23 (ei, e2, e3)2.

£31 £32 £33
Suppose two noncollinear vectors a and b are specified in an oriented

space.
The vector product [a, b] of vector a by vector b is a vector defined by
the following conditions:
(1) |[a, b]| = |a || b| sin <p, where cp is the angle formed by vector a and
vector b;
(2) [a, b] _L a, [a, b] l b ;
(3) the ordered triple a, b [a, b] is a right-hand triple.
If a || b, then, by definition, [a, b] = 0.
A vector product has the following properties:
[a, b] = - [b, a],
[2a, b] = 2[a, b],
[a, b + c] = [a, b] + [a, c],
[a, b] c = a[b, c] = (a, b, c).
If, in an orthonormal basis i, j, k,
a = {x, y, z}9 b = {*', / , z'}.
then
[•>»]=IKz,
liy *
z x
z' x'
■ ' I}.
' / If
Two bases el9 e2, e3and
id e1, e2, e3are
; said to be dual bases if
when i = j 9
when i ^ j.
The vectors of dual bases are connected by the relations
_ [e2e3] _ [e3e J _ _
(ex e2 es) [ei e2 e3) (ei e2 e3)
_ [e2 e3] _ [eê1] _ [e1 e2]
1 (e1 e2 e3) ’ 2 (e1e2 e3) ’ 3 '(e1 e2 e3)
Other notation: the dual bases a, b, c and a*, b*, c* are defined by the
relations
(a, a*) = (b, b*) = (c, c*) = 1,
(a, b*) = (a*, b) = (b, c*) = (b*, c) = (c, a*) = (a, c*) = 0
and, furthermore,
a * = b * = c* =
(abc)’ (abc)’ (abc)
[b* c*] [c* a*] ^ [a* b*]
(a* b* c*) 5 (a* b* c*) ’ (a* b* c*)
Any vector a is expanded in terms of the bases el9 e2, e3 and e1, e2, e3 as
follows:
a = (ae1) •ex + (ae2)*e2 + (ae3)-e3,
a = (aej)•e1 + (ae2)-e2 + (ac3)-e3,
where the numbers in parentheses are scalar products of vectors. These are
the Gibbs formulas.
The numbers a1 = ae* are called the contravariant components (coordi
nates) o f the vector a in the basis el9 e2, e3, and the numbers at = aef are
called the covariant components (coordinates) o f the vector a in the same
basis.
If the vectors a and b are specified by contravariant components in an
arbitrary basis el5 e2, e3,
a = xej + ye2 + ze3 = {x, y, z),
b = x'Ci + y'e2 + z'e3 = {x \ y ’, z'},
then the covariant components of the vector product
IZ X x y\]
[ • , b ] = y * f K z,
(z' x' ’ X' y 'W
We note here a number of other identities:
ac ad ac ad I
(a, b) (c, d) = ,[«, b][c, d] =
be bd be bd|
[a[b, c]] = b(ac) - c (ab), [a, b]2 + (a b)2 = a2b2
(the parentheses in the last two formulas contain scalar products);
[[a, b] [c, d]] = c(a, b, d) - d(a, b, c) = b (a, c, d) - a(b, c, d),
a(b, c, d) + b(c, a, d) + c(b, d, a) + d(a, c, b) = 0,
[a [b, e ]]+ [b [c , a]] + [c [a, b]] = 0,
ax ay az
(a, b, c) (x, y, z) = bx by bz
ex cy cz
(a, a) (a, b) (a, c)
(a, b, c)2 = (b, a) (b, b) (b, c)
(c, a) (c, b) (c, c)
([a, b], [b, c], [c, a]) = (a, b, c)2.
There are problems in this text that also involve what are known as slid
ing vectors.
A sliding vector is also a directed line segment; however, the equality
of such vectors is defined as follows: two nonzero sliding vectors AB an
CD are said to be equal if the lengths of segments AB and CD are equal,
if AB and CD are in the same direction, and if they lie on the same straight
line. A sliding vector co is given in a plane by its components x , y and by
the moment z = (r, co) with respect to some point O (r = OM, where
M is an arbitrary point of the straight line); (r, to) is the cross (or pseudo
scalar) product.
Sec. 3. Analytic geometry

Let us fix a point O in the plane and a basis el9 e2. Lay off vectors el5 e2
from the point O :
OE1 = el9 OE2 = e2.

The collection of straight lines Ox = OEl9 Oy = OE2 on which vectors
e1? e2 of the basis are laid off from O is termed a general Cartesian system
o f coordinates in the plane. The line Ox is the axis o f abscissas, or x-axis,
and the line Oy is the axis o f ordinates, or the y-axis. The point O is the
origin o f coordinates'. Ox and Oy are the axes o f the coordinate system.
A general Cartesian system of coordinates Oxyz in space is defined in
similar fashion.
If an orthonormal basis i, j (respectively i, j, k) is introduced in a plane
(respectively in space) and a point O is fixed, then the corresponding system
of coordinates is said to be rectangular Cartesian.
The radius vector r of point M is a directed line segment:
r - OM.
The general Cartesian coordinates x ,y o f a point M in the general Carte
sian system of coordinates (<0, el9 e2) in the plane are the coordinates of
its radius vector r = OM in the basis el9 e2:
r = xeL + ye2.
The coordinates o f point M in the rectangular Cartesian system
of coordinates (O, i, j) in the plane are the coordinates x, y of its radius
vector r = OM in the basis i, j:
r = xi + y\.
Similarly defined are the general Cartesian and rectangular Cartesian

coordinates x 9y 9z of point M in space:
r = OM = xe± + je 2 + ze3.
r = OM = xi + y] + zk.
If the point M has coordinates x 9y , then we write M(x, y) or M = (x, y)
[in space: M(x, y , z) or M — (x, y, z)].
If rx and r2 are the radius vectors of points A and B, then
AB = r2 — r2.
In coordinates:
= { x 2 — x l9 y 2 — yi} (in the plane),
AB = {x2 — xl9 y 2 — yi, z2 — Zi} (in space),

where A = (x l9 yx)9 B = (x 2, y 2) (in the plane); A = (xl9 yl9 zx), B =
(x2, y 2, z2) (in space).
The length d o f a line segment AB whose endpoints are specified by radius
vectors rx and r2 are computed from the formula
d = |r2 - Til.
In a' rectangular Cartesian system of coordinates:
d = ] f( x 2 — xx)2 (y2 — y xf (in the plane),
d = ][(x 2 - x2)2 (y2 - }\)2 + (z2 - Zj)2 (in space)
In a general Cartesian system of coordinates:
d =]fgn (x2 - Xx)2 + 2g12(x2 - xx) (y2 - yx) + g22 (y2 - y x)2 (in the plane)
d2 = gn (x2 - xx)2 + g22 (y2 - y j 2 + £33 (^2 - ZiY
+ 2g12 (x2 - xx) (y2 — y x) + 2g23 (y2 — y\) (z2 - z^
+ 2^31 (z2 — zx) (x2 — xx) (in space),
where gl7 are the components of the metric tensor introduced in Sec. 2.
In particular, the distance r from the point to the origin of coordinates
in a rectangular Cartesian system of coordinates:
r = ]/x2 + y 2 (in the plane)
r = 1/x2 + y 2 + z2 (in space).
and in a general Cartesian system of coordinates:

r = ]fgnx2 + 2g12 xy + g22y2 (in the plane)
r = U n x 2 + g22yz + gi3 z2 + 2g12xy + 2g2syz + g31zx (in space).
Suppose a rectangular Cartesian system of coordinates has been intro
duced in the plane. Let us consider an arbitrary point M different from
the coordinate origin. The first polar coordinate of point M is the length
of segment OM. The second polar coordinate of the point M is the angle q>
formed by vector i and the radius vector OM of the point M.
If x, y are the coordinates of point M in a rectangular Cartesian system
of coordinates, then
x = p cos cp, y = p sin q>;
x x . y y
p = ][x 2 + J 2, cos -
line segment MXM2 is the number
. m Jm
mm2
No matter what the number X ^ — l and no matter what the nonzero direct-
------>
ed line segment MXM29 there is one and only one point M which divides
the line segment M XM2 in the ratio X.
If ii and r2 are the radius vectors of the points Mx and M 2 then the radius
vector r of point M is defined by the relation
r = ri + ^ .
1+ X
In particular, the radius vector of the midpoint of the line segment is equal
to the half-sum of the radius vectors of its endpoints:
2
In a general Cartesian system of coordinates, the coordinates of the point
M are expressed in terms of the coordinates of the points M x and Af2 by the
relations
*i + Xx2^ y_i+ 4 v2(in the plane).
i + ;/’ i + x~
+ Xx2 yx + Xy2 zx + Xz2
--------------------------- 9 V— ------------------------------ 9 Z= ------------------- (in space),
1+ X 1+2 1+2
and the coordinates of the midpoint of the line segment are:

x= y = (in the plane),
*1 + X2
------------------------9
+ z 2 (in space).
2 2
Points A, B,C, . . . are called collinear if there is a straight line on
which they all lie.
Points A, B ,C ,D , . . . are said to be coplanar if there is a plane in which
all the points lie.
If rl9 r2, r3 are the radius vectors of the points A, B,C , then a necessary
and sufficient condition for their collinearity is of the form
(*i - r3, r2 - r3) = 0
and, in a general Cartesian system of coordinates,
“ *3 7i — 73 0
X2 *3 y 2 73
or
*i yi 1
*2 72 1 = 0 ,
*3 ys i
where A = (xl9 7i), B = (x29y2), C = (x3, y9).
If rl5 r2, r3, r4 are the radius vectors of the points A, B,C , D, then a
necessary and sufficient condition for their coplanarity is of the form
(fi — r4, r2 - r4, r3 — r4) = 0,
and, in a general Cartesian system of coordinates,
*i “ *4 7i - >4zi - *4
*2 “ *4 J 2 — 7 4 *2 — *4
-o .
*3“ * 4 73 “ 74 *3 “ * 4
where
A = Oi> 7i> Zi)9 B = ( x 2 , y 29 z2), C = ( x 3 , 73>Z * ) 9 D = ( x A9 yA9 z4).
A triangle ABC is said to be a collection of three points A, B,C . An
oriented triangle ABC is an ordered set of three points A ,B ,C . If the
points A ,B ,C are noncollinear, then A ABC is said to be nondegenerate,
and if they are collinear, then the triangle is degenerate.
The area (ABC) of a nondegenerate oriented A ABC lying in an oriented
plane is a number whose absolute value is equal to the area of A ABC,
it is positive if A ABC has a right-hand orientation (that is, the ordered
pair CA,CB has a right-hand orientation), and is negative if A ABC has
,
Basic Definitions Theorems and Formulas 351
a left-hand orientation. If the points A, B, C are collinear, then, by defi

nition, we agree that {ABC) = 0.
The area (ABC) of the oriented A ABC is computed from the formulas
1 1
(ABC) = — (CA, CB), (ABC) = — (rx - r3, r2 - r3)
(r1? r2, r3 are the radius vectors of the points A , B, C).

In a general Cartesian system of coordinates.
h * i- * 3 yi - ys _V9 *1 7i 1
2 x*— * 3 y* - 2 *2 72 1
*3 73 1
In a rectangular Cartesian system of coordinates
1 j y% i
(ABC) = — * 1 ~ * 3 J ' l - J ’j ~ X.2 y 2 i
* 2 — *3 72— 7S
*3 7 3 1
To compute the area S of A ABC, take the right-hand sides in absolute
value:
Vg *1 - *3 J l - J3 Xi yi 1
S = -~- mod = T -m o d
* 2 - *3 yz — y* *2 7s 1
*3 7s 1
7i 1
S = — mod = —- mod
2 ^ 2 -^ 3 ^ 2 -^ 3 2 x2 j>2 1
*3 7a 1
In these formulas, A = (xlf jj), 5 ; (* 2, 72) , C = (*3, J 3).
The ratio
of the nondegenerate A PQR and A ABC lying in a plane is a number whose

absolute value is equal to the ratio of the areas A PQR to the area of A ABC
and which is positive if A PQR and A ABC have the same orientation (that
is, RP, RQ | t CA, CB) and negative if A PQR and A ABC have opposite
orientations. If PQR is a degenerate triangle and ABC is a nondegenerate
triangle, then, by definition,
PQR
0.
ABC
We have the formula
PQR (PQR)
> (ABC)
ABC
{if A PQR and A ABC lie in an oriented plane).

The barycentric coordinates , p, y of a point M with respect to a non
cl
degenerate oriented triangle are the numbers
MBC n AMC ABM

CL = , P = --------- - 9
ABC ABC ABC

A tetrahedron ABCD is a set of four points A , B, C, D in space. An orient
ed tetrahedron ABCD is an ordered set of four points A, B, C, D in space.
------>
If the points A, B,C, D are noncoplanar, then the tetrahedron ABCD
is said to be nondegenerate and if they are coplanar, it is said to be dege
nerate.
------>
If a nondegenerate tetrahedron ABCD lies in an oriented space, then
it has a right-hand, or positive, orientation if ordered triples of vectors DA,
DB, DC and e1? e2, e3 have the same orientation. But if these ordered
------>
triples of vectors have opposite orientations, then the tetrahedron ABCD
has a left-hand, or negative, orientation.
------>
The volume (ABCD) of a nondegenerate oriented tetrahedron ABCD
lying in an oriented space is a number whose absolute value is equal to
the volume of the tetrahedron ABCD and which is positive if the te
trahedron ABCD has a right-hand orientation and is negative if the
------re
orientation is left-hand. If ABCD is a degenerate tetrahedron, then, by
definition, we assume that (ABCD) — 0.
------y
The volume (ABCD) of the oriented tetrahedron ABCD lying in an ori
ented space is computed from the formula
{ABCD) = — (DA, DB, DC).

6
If rl5 r2, r3, r4 are the radius vectors of the points A, B,C, D, then
(ABCD) = — (rL - r4, r2 - r4, r3 - r4).

6
In a general Cartesian system of coordinates,

xx - x t y1 - yx zx - z4 . 71 *i i
(ABCD) = — ^ - ^ y 2 - y 4 z8 - z4 = ^ *2 J'a *2 1
*s — *4 y» — y* zz — z* 6 *3 1 ’
*4 >-4 Z4 1
and in a rectangular Cartesian coordinate system,
*i > 4 Z1 Z4 *i zx 1
_1_
(ABCD) = — *2 - *4 J2 - y4 *2 — *4 *2 y2 z2 1
6 6
IX3 — Xx ^ 3 y 4 Z-3 ^4 *3 y» z3 1
*4 y* z* 1
In these formulas,
A = (xu y lt Zi), 5 = (*2, y 2, z8), C = (x3, y3, z3), Z) = fo , y t, z4)
The ra/io
P gi?5
ABCD
of a nondegenerate oriented tetrahedron PQRS to a nondegenerate oriented

------>
tetrahedron ABCD is a number whose absolute value is equal to the ratio
of the volume of the tetrahedron PQRS to the volume of the tetrahedron
------> ------>
ABCD and which is positive if PQRS and ABCD have the same orientation
(that is, the ordered triples SP, SQ, SR and DA, DB, DC have the same
------ > ------ >
orientation), and negative if PQRS and ABCD have opposite orientations.
------> ------>
If PQRS is a degenerate tetrahedron and ABCD is a nondegenerate one,
then, by definition, we assume that
PQRS
0.
ABCD
If the tetrahedrons PQ RS and ABCD lie in an oriented space, then
PQRS (PQRS)
(ABCD) '
ABCD
2 3 -8 1 0
3 54 Problems in Geometry
The barycentric coordinates a, /?, y, <5 of a point M with respect to a non

degenerate oriented tetrahedron ABCD are the numbers
MBCD n A M CD ABMD c ABCM

« = ---------------- , P = y = — -, <5 = -
ABCD ABCD ABCD ABCD

The area 5, in space, of A ABC whose vertices are given by the radius
vectors r2, r3 is computed from the formula
S = y Kt(ri - *») (** - r3)]2.
In a rectangular Cartesian coordinate system:
S = ---]f A\ + A\ + A2,
where
}’i - y t a - z3 5 A% — Z1 — ^ 3 *1 — * 3 *i — X a y , , —
Ax = j ^3 —
y s — y 3 - z3 z2 Z Z X2 -*3 *2 — xs y 2 — y3
and in a general Cartesian coordinate system:
S —— Yg11 ^1 + g 22^ i + 3+ 2g 12 ^1^2 + 2g23A2A3 + 2g31A3Av

In these formulas,
A = (*i, yl9 zO, £ = (*2, .y2, ^2), C = (*3> z3)> £,v = cV .
If we introduce a general Cartesian coordinate system in the plane, then
the equation o f any straight line lying in the plane is the first-degree equation
Ax -f By C — 0
and, conversely, any first-degree equation
Ax + By + C = 0
(where A2 + B2 # 0) in any general Cartesian coordinate system is an
equation of that straight line.
The direction vector a = PQ of a straight line p is any nonzero vector,
collinear with that straight line (vector PQ ^ 0 and the straight line p
are said to be collinear if the straight lines PQ and p are collinear, that is
either parallel or coincident; the zero vector is assumed to be collinear
with any straight line).
For a straight line specified in a general Cartesian coordinate system by
the equation
Ax + By -j- C — 0,
the vector a = {—B, A} is a direction vector.

A necessary and sufficient condition of the collinearity of a vector a =
= {l,m} and a straight line A x+ B y+ C = 0 in a general Cartesian coordinate
Al -j- Bm = 0.
For the coordinates of all points (x, y) lying to one side of the straight
line given by the equation
Ax -j- By -f C — 0
with respect to a general Cartesian coordinate system, the following in
equality is valid:
Ax By -\- C > 0
(positive half-plane); and for the coordinates x , y of all points (x, lying
to the other side of that straight line, the followng inequality is valid:
Ax -J-* By -f C <! 0
(negative half-plane).
The vector n = {A, B} is termed the principal vector o f the straight line
specified with respect to a general Cartesian coordinate system by the equa
tion Ax + By + C = 0. If it is laid off from any point M 0 of the straight
line in question, M qP = n, then the point P will lie in the positive half
plane.
If A and B are regarded as covariant components of the vector, then the
vector [A, B\ is said to be normal to the straight line specified with respect
to a general Cartesian coordinate system by the equation Ax + By + C = 0.
In a rectangular Cartesian coordinate system, the principal vector
n = {A, B} of a straight line Ax + By + C = 0 is the normal vector to
that line.
The equation of a straight line passing through point determined by
the radius vector r± and having the direction vector a is of the form
(r - rl9 a) = 0.
If in a general Cartesian coordinate system, a = {/, m j, M 1 = (x^ jx),
then the last equation becomes
x - x 1y - y ±
= 0
I m
or
x - *i _ y —y±
l m
(if / = 0, then this notation is to be understood as x — x2 = 0, and if
m = 0, then as y — y x = 0).
If / 7* 0, the last equation may be rewritten in the form

y — yi = k(x — Xx),
where k — m/l is termed the slope o f the straight line (in a general Carte
sian coordinate system). The slope of the straight line that passes through
two points (xl9 yx) and (x2, y 2) and is noncollinear with thej-axis is com
puted from the formula
* = -^ = 2 L
*2 -
(in a general Cartesian coordinate system).

In a rectangular Cartesian coordinate system, the slope is
k = tan a,
where a is the angle from the x-axis to the straight line under consideration.
The equation of the straight line cutting the jy-axis in the point (O, b)
and having slope k is, in a general Cartesian coordinate system, of the
form
y = kx + b.
The equation of a straight line that does not pass through the coordinate
origin and cuts the axes in points (a9 0) and (0 , b) is of the form
( intercept form o f the equation o f a straight line).

If a straight line passes through a point M x given by the radius vector rx
and has the direction vector a, then the radius vector r of any point of
it can be represented as
r = rx + fa,
where t takes on all real values. This equation is called the parametric equa
tion o f the straight line. The number t is the coordinate of point M on the
line in question provided that Mx is the coordinate origin and a is the
basis vector:
t r — r, M XM
a a
Suppose in a general Cartesian coordinate system we have r = {x, y}9
ri = {*i> yi}> a = {/, w}; then we obtain the parametric equations of the
straight line in the form
x = x x + It, y = yi + mt.
Let rx and r2 be the radius vectors of two distinct points Mx and Mt.
Then the equation of the straight line M ±M 2 is of the form
(r - rl9 r2 - rx) = 0.
In a general Cartesian coordinate system.
x - *1 y - y i = o
*2 - *1 y* yi
-
or
x y 1
*i yi i = 0
*2 ^2 1
or
X — x1= y — y1
x* ~ x 1 y2 — yx
(if any one of the denominators is equal to zero, then this notation is to
be understood in the sense that the numerator is also zero).
The equation of a straight line passing through point M x given by radius
vector rx and having the normal vector n is
n(r — rj) = 0.
If in a rectangular Cartesian coordinate system rx = {xl9 y x}, n =
= {A yB}y r{x, y }, then the equation n(r — = 0 becomes
A(x - Xj) + B(y - yf) = 0.
Such also is the form of the equation n(r — = 0 in general Cartesian
coordinates if t x = {xl9 y x}9 r = {x9 y] but n = [A, B] (covariant coor
dinates).
A necessary and sufficient condition that two straight lines specified
by the equations
Ax -f By + C = 0,
A 'x + B'y + C = 0
with respect to a general Cartesian coordinate system intersect, are parallel,
or are coincident is that the system of equations have only one solution,
no solution, or an infinity of solutions, respectively.
If these straight lines intersect, then to find the coordinates of their point
of intersection we have to solve the following system of equations:
Ax “f- By -|- C — 0,
A'x + B'y + C' = 0.
A pencil o f straight lines is the set of all straight lines passing through
the same point S (proper pencil; the point S is called the centre o f the
pencil) or the set of all parallel straight lines (improper pencil).
If general Cartesian coordinates are used to specify the equations of
two straight lines
Ax + By + C = 0
A 'x + B 'y + C ' = 0
that intersect in a point S, then the equation of the pencil with centre S
is of the form
oc(Ax + By + C) + P(A'x + B'y -f- C') — 0,
where a and ft take on all values with at least one nonzero value.
If the straight lines
Ax -f By -f~ C — 0,
A 'x + B'y + C' = 0
are parallel, then the preceding equation is the equation or an improper
pencil to which the given lines belong; note that all possible values are
taken for a and P but such that at least one of the coefficients of x and y
is nonzero:
ocA -f PA' y ccB -f- BB .
A necessary and sufficient condition that three straight lines given in
a general Cartesian coordinate system by the equations
Ax 4“ By + C — 0,
A'x + B'y + C ' - 0 ,
A "x + B"y + C" = 0
belong to a single pencil is the equality
A B C
A' B' C' = 0.
A" B" C"
Here, the straight lines in question belong to a single proper pencil if at
least one of the determinants
A' B’ A" B" A B
5 9
A" B" A B A' B'
is different from zero, and they belong to one improper pencil if all the
determinants are equal to zero.
If in a rectangular Cartesian coordinate system two straight lines are

given by the equations
Ax -f- By -J- C = 0,
A 'x + B'y + C' = 0,
then the cosines of the angles (plt2 between them are determined from the
formula
, A A '+ B B '
COS <pl f 2 = ± --^===r-•
YA2 + B2 Y ^ '2 + B'2
In general Cartesian coordinates,
8 u AA' + g12(AB' + A'B) + g22BB'
COS (p 1 2 = ± —7— ----y - •
I^ 2 + 2 g12AB + g22B2 YgnA'2 + 2 g12A'B' + g22B'2
The angle <p formed by the straight line Ax + By + C = 0 and the

straight line A'x + B'y + C' = 0 in rectangular Cartesian coordinates
is found from the relations
AA' + BB' AB' - A'B
cos (p » sm <p —
Y a 2 + B2 ][A'2 + B'2 Y a 2 + b 2 y A 't+ B '*
and if the straight lines are not perpendicular, then it suffices to know
only the value of tan<p:
AB' - A'B
tan cp = -----------------
AA' + BB'
In general Cartesian coordinates we have
g n AA' + g12(AB' + A'B) + g22BB'
cos q>
YgnA2 + 2 g 12AB + g22B2 ]fguA '2 + 2g12A'B' + g22B'2
= gnAA' - 2g12(AB' + A'B) + gnBB'_______________
Yg2A2 - 2 g 12A B + g nB2 ] f ^ A ^ ~ 2g12A'B' + g llB’2
AB' - A'B
sm cp = f f. ■ -— ,
]fg y gllA2 + 2g12AB + g22B2 ]fguA’2 + 2g12A'B' + g 22B'2
= _________________Vg(AB' - A'B)_________________
y g22A2 - 2 g 12A B + gllB2 ][g22A'2 - 2g12A'B’ + gn B'2 ’
and if the straight lines are not mutually perpendicular, then
AB' - A'B
tan <p = —7=--------------- -------------------------------
Yg(gllAA' + g12(AB' + A'B) + g22BB'
= _________ fg {A B ' - A'B)_________ .
g22A A ’ - 2g12(AB' + A'B) + gu BB'
Ordinarily, for the angle formed by the straight line Ax + By + C = 0

and the straight line A'x + B'y + C' = 0 one takes the set of values
{(p + kn} (k assumes all integral values), where cp is one of the values
of the angle formed by the straight lines.
If, in a rectangular Cartesian coordinate system, the straight lines p
and p' have slopes k and k ' and if they are not mutually perpendicular,
then the tangents of the angles q>12 between p and p' are computed from
the formula
k' ~~k
tan <plf2 = ±
1 + kk'
and in a general Cartesian system of coordinates
V g fr - k)
tan q>li2 = ±
' g u — g u ( k + k ’) + g z jc k '
The tangent of the angle <p formed by straight line p and straight line p' is
k' — k
tan <p =
1 + kk'
and in a general Cartesian coordinate system
tan <p = 1fg(k’- k )

gn - gi2(k + k') + g22kkf
If the straight lines p and p ’ have, in general Cartesian coordinates,
the slopes k and k \ then a necessary and sufficient condition for their
collinearity is that
k = k\
A necessary and sufficient condition of perpendicularity is that
£n — g^{k + k f) + g22k k f = 0,
and, in rectangular Cartesian coordinates,
1 + kk’ = 0 or k k ' = - 1.
If a straight line is given by the equation Ax + By + C = 0 in rectangu
lar Cartesian coordinates, then the distance d from point M0(x0, y0) to
this straight line is computed from the formula
\Ax0 + By0 + C|
a = -----------
r - .................
9
][A2 + B2
and if the vector n = {A, B}, which is normal to the straight line under
consideration, is a unit vector: A2 + B2 = 1 (in this case the equation
Ax + By + C = 0 is said to be normal), then
d = \Axq + By0 + C|.

j ______\Ax0 + By0 + C\ _y — \Ax0 + Byp + Cj\
~ g™AB+g**B* ~ 18 M - 2 g12A B + g n B* ’
and if the vector [A, B] is a unit vector, then
d = \Ax0 + Byo + C |.
If the straight line is given by the equation
(a, r - r 1) = 0,
then the distance d from the point M 0 with radius vector r0 to this straight
line is
d = )(a, rt - r0)| ^
|a|
and if a is a unit vector, then
</= |(a, rx — r0)|.
If in rectangular Cartesian coordinates, a = {/, m}, ro = {*o>Jo}>
fi = {*i, Ji}, then
mod *i - *o Ji - Jo
I m
d=
Vi* + mr
and if the vector a is a unit vector (12 + m2 = 1), then
mod - *o y ± - yQI
/ m I
where the symbol mod x denotes the absolute value of the number x. In
general Cartesian coordinates,
fgm od * i — * o y± — y n
I m
d=
]fgul* + 2gn lm + g2Zm2
and if a = {/, m} is a unit vector, then
d = ] fg mod J i-J .
/ m
If the straight line is given by the equation
n(r - rO = 0,
then the distance d from the point M 0 with radius vector r0 to this straight
line is computed from the formula
j !n(r, - r0)|
a = --------------- f
and if the vector n is a unit vector, then

d = Info - r0)|.
If in rectangular Cartesian coordinates r0 = {x0, j 0}, rx = {xx, j x),
ft = {A, B}, then
\A(xl — x0) + B(yx — jo) |
d ~ )[A*+~B*
and if the vector {A, B} is a unit vector, then
d = |A(x0 - xx) + B(y0 - j x)j.
d= Ign^fa — *o) + gizU(y! - Jo) + B(x\ — x0)] + g22B(yL — j 0)j
VgiiA2 + 2g12AB + g22B2
and if \{A,B}\ = 1, then
d= ~ x0) + g12[A(y1 - j 0) + B(x1 - x0)] + g22B(yx - j 0)|.
If the vector n is given by covariant components, n = [A, B], then
d= M(xx - x0) + B(y1 — j 0)| _ \A(X! — x0) + B(yl — j 0)| ^
][gÂ* + 2g12AB + g22B2 8 ]fg22A2 - 2gn AB + g nB* *
and if |n| = 1, then
d = \A(xt - x0) + B(y\ - jo)|.
The area {ABC) of an oriented A ABC whose sides are specified with
respect to a general Cartesian coordinate system by the equations
(BC): Axx + Bxy + Cx = 0,
(CA): A 2x -f- B2y -j- C2 = 0,
(AB): A 3x + B3y + C3 = 0,
is computed from the formula
Ax Bl Cx 2
A2 B2 C2
VF A3 B3 c 3
(ABC)
2 a 2 b21 A2 Bg A Bt
A3 B3 U i Bx A2 B2
and, in rectangular Cartesian coordinates, from the formula

Ai Bi Ci 2
A2 B2 C2
1 A3 B3 C3
;;2 A2 B2 ^3 &3 A x
A3 B3 A ! B, A2 B2
The area S of A ABC is
Ai B± Ci
A 2 &2 C2
A 3 B 3 C3
S = Vg
B2 A 3 Bs Ai B
" m o d (r2
\ \ A Z B3 Ai Bi
A 2 BiII)
Ax Bx Cx
A 2 B 2 C2
1 A 3 Bs C3
S = T mod I
IIA 2 B
A 3 B2
2 A 3 B3 Ai B}
Ai Bx A 2 Bi il)
If a general Cartesian coordinate system Oxyz is introduced in space,
then the equation of any plane in the first-degree equation
A x -f- By -(- Cz -{-.0 = 0,
and, conversely, any first-degree equation
Ax -}- By -f- Cz -{- O = 0
(where A 2 + B 2 + C2 ^ 0) in any general Cartesian system of coordinates
is the equation of the plane.
For the coordinates of all points (*, y, z) lying to one side of a plane
that is specified with respect to a general Cartesian system of coordinates
by the equation
A x 4" By -f- Cz -f- 0 = 0,
the following inequality is valid:
Ax -f- By -J- Cz -j- O > 0
(positive half-space), and for the coordinates of all points (x ,y ,z) lying
to the other side of that plane, the following inequality is valid:
Ax -{- By -f Cz -f- O <: 0
(negative half-space).
The vector n = {A, B,C } is called the principal vector o f the plane speci
fied by the equation
Ax "j- By -j- Cz -f- 2) — 0.
---- **
If it is laid off from any point M0 of the plane, M0P = n, then point P
lies in the positive half-space.
If A, B ,C are regarded as covariant components, then the vector [A, B, C]
is normal to the plane
Ax + By + Cz + D = 0.
In a rectangular Cartesian system of coordinates, the principal vector
n = {A, B, C} of the plane specified by the equation Ax + By + Cz + D =
= 0 is a vector normal to that plane.
The equation of a plane passing through point M 1 defined by the radius
vector rx and normal to the vector n is of the form
n(r - rj) - 0.
If in general Cartesian coordinates rx = {xl9 y l9 zx}9n[A, B, C], r = {x9y 9z}9
then the equation n(r — r2) = 0 becomes
A(x — A'i) + B(y —y j + C(z — zL) = 0.
In rectangular Cartesian coordinates,
d = {A9B9C} = [A9B9C].
The vector PQ =£ 0 and the plane n are said to be coplanar if the straight
line PQ is either parallel to, or lies, in the plane n. The zero vector is
assumed to be coplanar with any plane.
For a vector a = {/, m9n} and a plane Ax + By + Cz + D = 0 to
be coplanar, it is necessary and sufficient that the following equality be
valid:
Al -f- Bin -f- Cn — 0
(the coordinates are general Cartesian).
The equation of a plane passing through a point and coplanar
with two noncollinear vectors a and b is of the form
(r - r 1? a, b) = 0.
y - yi z —>
H
H
1
k mi = 0,
k m2 n2
where a = {ll9 ml9 n ^ 9 b = {/2, m2, «2}, tx = {xl9 yl9 z j , r = {x9y 9z).
The equation of a plane passing through two points Afifo) and Af2(r2)
------>
and coplanar with the vector a Ik M XM2 is of the form
(r - rl9 r2 - rl9 a) = 0.
In general Cartesian coordinates.
x - x x y — y\ z - Zi
*2 — >’2—y± z2— = 0,
I m n
where M x = (xl9 y l9 Zj), M 2 = (x29y2, z2), a = {/, m9n}.
The equation of a plane passing through three noncollinear points
M2(t2), M3(r3) has the form
(r — r3, rx — r3, r2 — r3) = 0.
and, in general Cartesian coordinates,
x — x 3 y — ys z — z3
*i — *3 T1 - T 3 z1 — z3 = 0,
* 2 — * 3 >J2 —y* z2— z3
or
x y z 1
^1 J i 1 _ Q
*2 T2 ^2 1
x 3 x 3 z3 1
where M x = (*l5 z*), M2 = (x2, y 2, z2), M3 = (x3, >>3, z3).
The equation of a plane passing through a point Mjfo) (and coplanar
with two noncollinear vectors a and bean be written in parametric form:
r = r± + wa + fb.
In this equation, u and v are the coordinates of the point M in a general
Cartesian coordinate system in the plane at hand, in which the origin is
the point Af1(r1) and the basis is a, b.
The parametric equations of a plane in a general Cartesian coordinate
system are:
X = Xt + + vl2.
y = yi + umX + vm2,
z = zx + uitx + 1m2y
where rx = {xx, y u z j , a = {/x, mu wx}, b = {/2, m2, «2}.
The parametric equation of a plane passing through points M^rri, M2(i2)

and coplanar with the vector a Ik MJM2is of the form
r = rx + «a + v(r2 — r^.
X = Xi + Ul + v(x 2 — Xj),
y = y i + um + v(y 2 — ji)-
2 = zx + un + v(z 2 — Zj),
where Mx = (x1; y lt zx), M 2 = (x2, y 2, z2), a = {/, m ,«}.
The parametric equation of a plane passing through three noncollinear
points M fa ), M2(r2), M 3(r3) is of the form
r = r3 + — r3) + v(r2 — r3)

x = x3 + u(xj, — x3) + v(x2 — x3),
y = y* + u(yi - y 3) + v(y2 - y a),
z = z3 + u(z1 — z3) + v(z 2 — z3),
where M x = (x3, y u zx) M 2 = (x2, y 2, z2) M = (x3, y 3, z3).
If the plane does not pass through the coordinate origin of a general
Cartesian coordinate system and cuts the coordinate axes in the points
(a, 0,0), (0, b, 0), (0,0, c), then its equation can be written as
-^ + ^ + ^ = 1
a b c
(the intercept form o f the equation o f the plane).
For two planes given by the equations
Ax By -J-' Cz -f- D — 0,
A x - \~ B y - \- C z - { - D — 0
with respect to a general Cartesian coordinate system to intersect, it is
necessary and sufficient that the vector
(15 C C A
||5 ' C C A' ' B' I]
be nonzero: a ^ 0. In this case (a # 0), the vector a is the direction vector
of the straight line along which the planes in question intersect.
For the two planes
Ax -f- By -f Cz -f- D — 0,
A'x + B'y + C'z + D' = 0
Basic Definitions, Theorems and Form ulas 367
■5T
to be parallel, it is necessary and sufficient that the vector a be equal to
zero but that at least one of the determinants
A D B D C D
A' D' ’ B' D' C D'
be different from zero.
For the two planes
Ax -I- By -J- Cz -f- £) — 0,
A 'x + B'y + C'z + D' = 0
to be coincident, it is necessary and sufficient that the corresponding
coefficients of the equations of the planes be proportional:
A ' = kA, B' = kB, C = kC, D9 = kD.
Three planes specified with respect to a general Cartesian coordinate
system by the equations
Ax -f" By -f- Cz -f- D — 0,
A'x + B'y + C z + D ' = 0 ,
A"x + B"y + C"z + D" = 0
have only one point in common if and only if the following inequality
is valid:
A B C
A' B' C' * 0 .
A" B" C"
To find the coordinates of this point it is necessary to solve the system of
equations of the three given planes.
A pencil o f planes is a set of all the planes passing through one straight
line / (proper pencil). The straight line / is called the axis o f the pencil.
An improper pencil of planes is the set of all parallel planes.
If two planes given by equations with respect to a general Cartesian
system of coordinates.
Ax -f- By -j- Cz -f- D — 0,
A'x + B'y + C 'z + D ' = 0
intersect along a straight line /, then the equation of the pencil of planes
with the axis / is of the form
oc(Ax + By + Cz + D) + fi(A'x + B'y + C'z + D') = 0,
where a and /? take on all possible values with at least one of them being
different from zero.
If the planes are parallel, then the last equation is the equation of the
improper pencil of planes to which they belong, and for a and P one
takes all possible values with the exception of those for which all the
coefficients of x , y , z, that is,
ocA + p A \ <xB + PB', aC + p C
are equal to zero.
A bundle (or sheaf) o f planes is the set of all planes that pass through
the same point S (proper bundle) or the set of all planes coplanar with
one straight line (improper bundle). The point S is termed the centre
o f the bundle.
If three planes specified by equations in general Cartesian coordinates,
Ax + By + Cz + D = 0 ,
A 'x B'y -f- C'z -}- D; = 0,
A"x + B"y + C"z + D" = 0,
have the same common point 5, then the equation of the bundle of planes
with centre S is of the form
<x(Ax + By + Cz + D) + P(A'x + B'y + C z + D)
+ y(A"x + B"y + C z + D") = 0,
where a, p , y take on all values, at least one of which is nonzero.
If the three indicated planes are coplanar with one and the same straight
line, but do not pass through one straight line, then the last equation is
the equation of the improper bundle to which the three given planes belong;
and for a, p9y one takes all possible values, with the exception of those
for which all the coefficients of x, y, z, that is,
ocA + PA' + yA", ctB + p V + yB", <xC + fiC' + PC"
are equal to zero.
The distance d from the point M 0(x09y09 z0) to the plane given by the
equation
Ax -f- By -J- Cz -f Z) — 0
in rectangular Cartesian coordinates is computed from the formula
j _ 1Ax0 + By0 -j- Cz0 + D I
“ y A 2 + B 2 + C 2
and if the vector n = {A, B, C} is the unit vector (the equation Ax + By +

+ Cz + D = 0 is then said to be normal)9 then
d = |A xq + By0 + Cz0 + D\.
,

____________ 1Ax0 + By0 -f- Cz0 + D 1 _____
" + g2*B2 + g33C2 + 2g12AB + 2g23BC + 2g31CA
_ Vg\Ax$ + By0 -f~ Cz0 + ^1
£ll gl2 gl9 A \
_ g%l g22 #23 B
gz\ #32 #33 C
A B C 0 /
and if the vector [A, B, C] is the unit vector, then
d = |Ax0 + By0 + Cz0 -f- D\.
The distance d from the point M0(r0) to the plane given by the equation
(n, r— rj) = 0 is computed from the formula
d _ [nfa-Tp)!
M ’
and if n is the unit vector, then
d = \n(rx - r0)|.
If the vector nis given by covariant components in a general Cartesian
system of coordinates: n= [A, B, C], and the vectors r0 and are given
by contravariant coordinates (components):
*0 = { * o > J o . Z0} , *1 = { * 1, J j , Z l } .
then
d = ____________ \A (x i - *o) + B (yi - Jo) + C (Z i - Z0) ____________
IfgnA2 + g225 2 + g33C2 + 2g12AB + 2g™BC + 2g31CA

_ fg \A(xi - *o) + B(yt - Jo) + C(z1 - 2 0) |
/ gll gl2 gl3 d \

_ &21 H>22 g23 &
g$l &32 g23 C
V A B C 0 /
and if the vector n = [A, B, C] is the unit vector, then
d = \A(xx — X q) + B(yl — j o ) + C(zx — z0)|.
d= ~ *o) + -S(ji - Jo) + C(z1 — z0)|
VA2 + £ 2 + C2
24 —810
and if the vector n = {A, B , C] = [A, B, C] is the unit vector, then

d = \A(x± - x 0) + B fa - yo) + C(zi - z0) |.
The cosines of the angles <plt2 between two planes specified by the
equations
Di(r — ri) = 0, n2 (r — r2) = 0
are computed from the formula
iDiDol
cos <plf2 = ±
I«il |n2|
If the planes are given by the equations
A x -f- By -j- Cz -f- D =■ 0 ,
A 'x + B'y + C z + D' = 0
with respect to a rectangular Cartesian coordinate system, then
AA' + BB' + C C
COS <p12 = + -7-........................... -- .
][a 2 + B2 + C2 Y a '2 + B'2 + C'2
_ , ([A,B,C], [ A \B \C '] ) .
cos <î 2 — ± ------------------------------;
I[A, B, C] I |[A', B', C ) \
([A, B, C], [A\ B', C']) = g*AA' + g*BB’ + g™CC’
+ g12(AB’ + A'B) + g™(BC' + B'C) + ga,(CA' + C'A)
gll Sit Sit A
_ _.A. Sti Stt Sts B
E £*i Stt Stt O
A' B' C' 0
IN, B, C]\ = YguA2 + gl2B2 + g™C2 + 2g '2AB + 2g*BC + 2g31CA
/ 8ll 8 l 2 8 l 3 A 1/2
_ 1 821 822 823 B

Ye # 3 1 &32 833 C
\ A B C 0
IN', B', C']|
= YgnA'2 + g*F* + gMC'2 ^ 2 ^ 'B '~ + 2 g ^ B ’Cr :\-2g2lC'A'
/ 811 S12 813 A' \ 1/2
_ 821 822 822 B'
£31 £32 8%3 C'
V A’ Bf C 0
.
The parametric equation of a straight line passing through point

and having the direction vector a is of the form
r = fi + fa.
where f is the coordinate of the point A/(r) on that line if we take point
for the origin and vector a for the basis vector:
M XM
t = 1— .
a
In a general Cartesian coordinate system, the parametric equations of
a straight line are written as follows:
* = *i + tl ,
y = yi + tm,
z = zx + tn ,
where Mx = (xl9 yl9 zx), a - {/, m, n}.
The parametric equation of the straight line passing through two
points A f ^ ) and M2(r2) is of the form
r = Ti + f(r2 - rx),
X = X \ 4 - t ( X o — A'x),
y = yi + t(y i- yi),
z = zx + t(z2 — Zx).
Two straight lines are said to be coplanar if they lie in one plane.
For two straight lines
r = rx + fa, r2 = r2 + fb
to be coplanar, it is necessary and sufficient that the following equality
hold:
(r2 — rlf a, b) = 0
or, in general Cartesian coordinates,
*2 — *1 y* — yi z2 zx |
/i m± *i = 0,
k m2 no
where M1 = (xly yu zt), M2 = (x2, y2, z2), a = {/„ mu n j , b = {/2, nu, n2).
Remark. The equations of a straight line are often written in the form
x — x x _ y — y x _ z — zx
I m n
(canonical equations o f a straight line). If one of the denominators is zero,

then this notation is to be understood as follows: the numerator is also
equal to zero. For example, the system of equations
x —2 _ y —5 _ z —1
3 0 ~ 4~~
is to be understood as follows:
y - 5 = o,
x — 2 __ z — 1
3 _ 4
The distance d from the point M0(r0) to the straight line given by the
equation r = rx + /a is computed from the formula
d = l(ri - r0, «)l
|a|
]fA2 + A\ + A\
|/72 + m2 + n2
where
yi-yo zi—zo Ao — Z1 Z0 %i -Xo * i-x o y i-y o
A± =
m n I i m
and, in general Cartesian coordinates.
d= + g22A\ + g™A% + 2g12A1A2 + 2g™A2A3 + 2gÂ3A1
]fgiJ2 + 8 2 2 m2 + g 3 3 n2 + 2 g12lm + 2 g23mn + 2 g91nl
/ 8 ll 8l2 8 13 At \ 1/2
_ 8 21 8 2 2 822 A2
831 832 #33 ^3
A\ A 2 A 3 0 /
\fgiJ2 + gi3m2 + g33n2 + 2g12lm +2g23nm+2g3lnl

The shortest distance d between the noncollinear straight lines
r = rj + fa, r = r2 + fb
is
d = Kr2 ~ ri>a. b)l
l[ab]|
If in rectangular Cartesian coordinates we have

*x = {*1, yi, Zi}, r2 = {x2, y 2, z3}, a = {/1( mlf nx}, b = {l2, m2, n2},
then
* 2 - * i J 2 - J 1 ^ 2 - ^ 1
mod h mx nx
d= 12 M2 ^2
][d\ + S*+ SI
where
mx nx »1 h > <^3 = h "ii
*1 = s* =
m2 n2 n2 l2 h m2
and, in general Cartesian coordinates.
X2 — y 2 - yi z2 — zx
mod h mx
h m2 «2
d=
U ]lgn Sl + g*b\ + g™b\ + l g liS1S2 + 2 g™S2S,3 + 2g31<53(51
y z - y i z2
*2 — * 1
mod h m1
_ h m2 n2
gll gl2 giz <5i 1/2
g2l g22 #23
&31 gs2 £33 ^3
< <53 0
A straight line in space can be specified by the equations o i two inter
secting planes:
Ax + By + Cz + D = 0,
A 'x + B’y + C'z + Z>' = 0
(general Cartesian coordinates).
Its direction vector is
C A
-! C' A'
To reduce the equations of the straight line
\
Ax + By + Cz + D = 0,
A 'x + B'y + C z + D' = 0
to parametric form, we have to find some solution x l9 y l9 zx of the system;

and then the parametric equations of the given straight line are
B C
X = Xt + t
B' Cf ’
C A
y )’i + t
C Af ’
A B
z — zt + I
A9 B'
and the canonical equations are
X — Xj y -y i Z - Zi
B C C A A B
B' C' C' A' A' B'
A necessary and sufficient condition for the plane Ax + By + Cz -f- D =
= 0 and the straight line x = x x + It, y = y x + mt, z = zx + nt, given
by equations with respect to general Cartesian coordinates, to intersect,
be parallel, or for the line to lie in the plane is as follows:
| condition
intersection A l + Bm + Cn # 0
parallelism A l + Bm + Cn = 0
Ax, + By, + Cz, + D * 0
straight line lies in plane Al + Bm 4- Cn = 0
Axx + Byx + C zx 4- D = 0
The angle between the straight line

r = rA+ fa
and the plane
n(r — r0) = 0
is given by the formula
Iani
sin <p =
a In!
If, in rectangular Cartesian coordinates, a straight line is given by the
equations
x = Xi + It,
y = y1 + mt,
z = zx + nt
and a plane by the equation

Ax + By + Cz + D — 0,
then
Al + Bm + Cn
sm q> —
+ B2 + C2 V/2 + ro2 + /I2
Al + Bm + Cn
sin <p = -- — ------- ,
Vk Vt s
where
7i = gJM2 + g22£ 2 + g^C 2 + 2g'2AB + 2g23BC + 2g31CA
gll gl2 glZ d
g21 g22 #23 &
g £31 £32 #33 c
A B C 0
T2 = gul2 + g22m2 + gZ3n2 + 2g12lm + 2g2Zmn + 2g31nl.
A necessary and sufficient condition for a line and a plane to be parallel
is of the form
[A,B, C) || {Urn, n}
or
{Agu + Bg12 + Cg13, Ag21 + Bg22 + Cg“ Ag31 + Bg32 + C^33}|| {/, m, n}
or
Agn + Bg'2 + Cg13 - £/,
v4g21 + #£22 + Cg23 = km, k ^ 0,
>te31 + *£32 + Cg33 = *n,
or
£ 11/ + = kA,
g2il + £ 22™ + £ 23" = kB, k ^ 0,
gzil + £32™ + g^n = kC9
A — A7, 5 = km9 C = kn.
The equation o f a circle (C, r) with centre C(a, ft) and radius r in rec
tangular Cartesian coordinates is of the form
(x — a)2 + (y - £>)* — r2 = 0,
and if the centre of the circle is the origin of coordinates, then

x 2 + y2 + z2 = 0.
Sometimes the point C(a, b) is regarded as a circle of zero radius (the
zero circle). The equation of a zero circle C(a, b) in rectangular Cartesian
coordinates is of the form
(x - a)2 + ( y - b)2 = 0.
For the coordinates x, y of any point (x, y) lying outside the circle
(x - a)2 + ( y - b ) 2 - r 2 = 0
we have the inequality
o = (x — a)2 + (y — b)2 — r2 > 0,
and for all points M(x, y) lying inside that circle we have
a = (x — a)2 + (y — b)2 — r2 < 0.
The number a is called the power o f the point M(x, y) with respect to the
circle (C, r) and is equal to
G = d2- r \
where d is the distance from point M to the centre C of the circle (C, r).
If an arbitrary straight line intersecting the circle (C, r) in two points A
and B is drawn through M, then
a = (MA, MB) = MA-MB.

The equation o f a sphere (5, r) with centre S(a, b, c) and radius r in
rectangular Cartesian coordinates is of the form
(x — a)2 + ( y — b)2 + (z — c)2 — r2 = 0,
and if the centre of the sphere lies in the coordinate origin, then
x 2 + y 2 + z2 — r2 = 0.
The equations of zero spheres are:
(x — a)2 + (y — b)2 + (z — c)2 = 0,
x 2 + y2 + z2 = 0
For the coordinates x 9y , z of any point M lying outside the sphere
(x - a)2 + ( y - b)2 + (z - c)2 - r2 = 0
we have the inequality
g = (x — a)2 + (y — b)2 + (z — c)2 — r2 > 0,
and for all points M (x, y , z) lying inside the sphere we have
g = (x — a)2 + (y — b)2 + (z — c)2 — r2 < 0.
The number a is called the power o f the point M with respect to the sphere
(S, r) and is equal to
cr = d2 — r2 = (MA, MB) = M A-M B,

where d is the distance between the points S and M, and A and B are
points in which an arbitrary straight line passing through point M inter
sects the sphere (S, r).
Sec. 4. Complex numbers

In the set of complex numbers x + yi (x and y assume all real values),
the sum and product are defined as follows:
(x + yi) + (*' + y'i) = (x + s') + (y + ? ) i,
(x + yi) (pc' + y'i) = (xx' — yy') + (xy' + x'y) i;
here, the real number x may be written as x + 0-j>; in particular, 1 = 1 h
+ 0/, 0 = 0 + 0 - /. It is obvious that
0 + z = z, 1 • z = 1 and 0 • z = 0
for all complex numbers.
The set of complex numbers contains all the real numbers (x + 0 - / =
= x + 0 = x) and also the number i (0 + 1• i = 1 • / = /), the square
of which, by virtue of the definition of a product of complex numbers,
is equal to —1:
i2 = - 1 .
The operations of subtraction and division are defined as the inverses
of addition and multiplication; if z = x + yi, then —z = (—x) + (—y) i.
The following properties are obvious:
2 + (Z' + 2") = (2 + 2 ') + 2 " ,
2 + ( - 2) = 0,
z + z' = z f + z.
z(z'z") = (zz') z",
22' = Z’Z
Z (z ' + Z " ) = ZZ' + 2Z " ,
where z, z', z" are arbitrary complex numbers.

The modulus (absolute value) \z\ = p of the complex number z — x + yi
is the principal root (positive real root) Y x2 + y 2:
W = P = V*1 + 7*-
The argument (or amplitude) q>= arg z of the complex number z = x +

+ yi ^ 0 is the number (p defined by the relations
JC x y_ ____y _ _
cos cp = — sin (p
P \[x* + y2’ P j/*2 + y*
The argument of the number z ^ 0 has an infinity of values. If cp is one
of the values of the argument of z ^ 0, then all the values arg z are con
tained in the formula
arg z = q> + 2kny
where k takes on all integer values.
This relation is frequently written as follows:
arg z = cp(mod 2 n )
(read: “the argument z is congruent to cp modulo 2rc”).
If z = 0, then p = 0, and cp is any number.
For two complex numbers z ^ 0 and z' # 0 to be the same, it is necessary
and sufficient for their moduli to be equal,
1*1 = l*'l,
and for their arguments to be congruent modulo 2n:
arg z = arg z'(mod 2n).
From the preceding formulas it follows that
z = x + yi = p (cos '),
then
zz' = pp' [cos((p + <p') + i sin(cp + <p%
— = — [cos(')], z' / 0,

z p
1*1 I* I,
arg(zz') = arg z + arg z'(mod 2n),
J*l
izi
arg — arg z — arg z'(mod 2n), z' ^ 0,

z
zn = pw(cos ncp + / sin n(p)
(n an integer); in particular, if |z| = p = 1, then

(cos ) (the de Moivre formula).
If z ^ 0 and n is a natural number, then
2k + l .s l.„ <p
_ +_ 2A:7u
_ j \.
)jz = y p(cos <p + / sin (p)- ■ f c ( c o s ^ ± _
fc = 0, 1,2,. .. , n — 1.
Two complex numbers z = x + yi and z = x — yi are said to be conj
ugate. They have the following properties of conjugacy:
z = z,
z + z' = ~z + z',
z — z' = z — z',
zz' = zz'.
If m= — (z' 7^ 0), then u = —

z' z'
M = 1*1,
arg z — — arg z (mod 27r).
Suppose a rectangular Cartesian coordinate system is introduced in a
plane. With every complex number z = x + y i we associate a point Af(jc, y).
This correspondence is one-to-one.
The number z is called the affix of the point M. A point with affix z
is here symbolized thus: M{z) or M = (z).
If z1 = x ± + yfi, z2 = x 2 + jv , z 3 = x 3 + are the affixes of the
points A, B ,C in rectangular Cartesian coordinates, then the area {ABC)
of an oriented £±ABC is computed from the formula
*1 Z l 1
(ABC) = — Z2 Z2 1
4
Z3 *3 1
In particular, a necessary and sufficient condition for the collinearity
of three points A{z1), B(z2), C(z3) is of the form
Zi z\ 1
z2 z 2 1 - 0 .
z3 z 3 1
The transformation under which a point M(z) is associated with a point

M \z '), where
z' = az + b.
is a similarity transformation o f the first kind (that is, a transformation
that does not change the orientation of the plane). Indeed,
and transition from point M(z) to point M f(z') is performed as a translation

— z -j-----and a transformation z' = azl9 which consists in a rotation
a
about the origin through an angle arg a and a homo the tic transformation
with centre O and ratio \a\. All these instances are similarity transformations
of the first kind.
A transformation under which a point M(z) is associated with a point
M'(z'), where z' = az + b, is a similarity transformation of the second kind
(that is, a transformation that reverses the orientation). Indeed, this trans
formation consists in a symmetry zx = z with respect to the x-axis and
a transformation z' = az± + b that reduces to a translation, a rotation,
and a homothetic transformation. Of all these transformations, only
symmetry with respect to the x-axis changes the orientation of the plane.
From the foregoing it follows that if two triangles ABC and PQR are
given via the affixes of their vertices,
A = fe), £ = (zg), C = (z3),
P = («i), Q = (w2), R = (w3)>
then a necessary and sufficient condition that A ABC and i\PQ R be similar
and have the same orientation is
zx 1
z2 m2 1 = 0,
z3 *^3 i
and a necessary and sufficient condition that A ABC and A PQR be similar
but have opposite orientations is
«i 1
Zg Ug 1
z 3 «3 1
Let us consider two distinct points M x(z^) and A/2(z2). On the basis of
the foregoing, point M(z) lies on the straight line MXM2 if and only if
z z 1
Zi *i 1 = 0.
z2 z2 1
This equation can therefore be called the equation of the straight line MXMV
It can be transformed to
Z2 Z1 /— — v
z zi ~ -------- ZT (z - z l)*
Zo Zi
We will call the ratio

Z2 Zj
x=
z2 zx
the complex slope o f the straight line M XM2. Note that
Z2 ~ Zl = 1.
*2 - *1
Thus, the equation of the straight line M XM2can be written down as
Z — Zx = X (Z — Z7! ) ,
where |x| = 1.
Conversely, any equation of the form
Z — Zx = x(z — Zj),
where |x| = 1, is an equation of a straight line. Indeed, since \x\ = 1, it
follows that
x = cos <p + i sin <p.
Setting up the equation of the straight line passing through two points
with affixes zx and zx + cos - - + / sin — , we obtain the equation
2 2
Z ~ Zjl = x (z — z’j).
In particular, note the equation of the straight line that passes through
the coordinate origin:
z = xz,
where
x = cos (p + / sin (p = —
zi
and zx ^ 0 is the affix of any point (zx) of the straight line under consi
deration.
Note that the straight line
z = xz
passes through points of the unit circle (the unit circle is a circle with centre
at the coordinate origin and with radius equal to 1) with affixes ][x (J/x
always has two values: they are the affixes of the endpoints of the diameter
of the unit circle). Indeed, if ]/x is either of the values of this radical,
then for z = ]fx the equation z = xz becomes an equality (the left-hand
side is ]/x; the right-hand side is, x]fx = J/x).
The two straight lines
Z — zx = x(z — Zj), \x\ = 1,
z — z2 = x'(z — Z2), \x'\ = 1,
are collinear if and only if x = x'. True enough, these lines are collinear
if and only if the system of equations
z — xz = z1 — x z l9
z — x'z = z2 — x z 2
in z, z either has no solution or has an infinitude of solutions, and this
occurs if and only if
1 —x
= 0 or x = x \
1-x '
The two straight lines p and q given by the equations
z — Zx = x(z — Zj), |x | = 1,
z — *2 = x'(z — Z2), |x'| = 1,
are perpendicular if and only if
x + x' = 0.
Indeed, let us consider the straight lines p ' and q \ which are collinear
respectively with p and q, but which pass through the coordinate origin :
p': z = xz,
q': z = x ’z
The straight lines p and q are perpendicular if and only if the lines p '
and q' are perpendicular. Suppose p ‘ and q' are perpendicular. On p ’
take a point with the affix z0 ^ 0. Then by virtue of the relation
arg(izG) = arg i + arg z0 = ----- f- arg z0(mod In)

2 ,
the point with affix zz0 lies on the straight line q\ We have
z0 — x z o, iz0 = x iz0
or
z0 = xz'o, iz0 - — x f i z 0
or
Zq — xz o, Zo — X Z q.
Hence, x z 0 — — x'z0 and since z 0 ^ 0, it follows that x = — x \ whence

x + x' = 0.
Conversely, let x + x f = 0. Let us prove that the straight lines p' and qf
are mutually perpendicular. Draw through the origin a straight line p*
perpendicular to q', and let x* be the complex slope of p*. Then x * +
+ x' = 0, and since x + x' = 0, it follows that x = x* and, hence, the
straight lines p* and p' are coincident, that is p' J_ q', whence p _L q.
Remark. In analytic geometry, the slope k of a straight line that is non-
collinear with the y -axis is the tangent of the angle of inclination of that
line to the jc-axis. If the complex slope of a straight line that is noncollinear
with the y-axis is equal to
x = cos cp + / sin <p,
then the angle a of inclination of that line to the *-axis is equal to (p!2
(since a straight line passing through the coordinate origin and having
a complex slope x passes through points with affixes Yx).
Now we find
1 —x . <p
----- = — i tan — = — ik ;
1+ x 2
consequently,
, A —x
k = z ------ .
1+ x
Conversely,
i —k
x = ------ .
i ~f~ k
The angle formed by the straight line
Z — Zl = x(z — Z j)
and the straight line

z — z2 = x \ z — z?)
is equal to
y x*
arg — (mod n).
yx
Indeed, the straight lines z = xz, z = x'z, which are collinear with the
given lines, pass through points with affixes |/x and |/V respectively
and so the angle formed by the two straight lines is equal to
arg 1/V — arg jfx = arg - (mod n).

yx
The equation of any straight line can be written as
Az + Bz + C = 0,
where C is a real number and B — A ^ 0. Conversely, any such equation
is an equation of a straight line provided C is a real number and B = A # 0
Proof. Let Px + Qy + R = 0 be the equation of a straight line in rec
tangular Cartesian coordinates. Since
X = (z + z ) 9
y = — (* - 2) = — (Z - Z),
21 2
it can be rewritten thus:
“ P(z + Z) + — Q(z - 2) + R = 0
or
P - Qi
2 z + 2 '.' z +R = 0
or
A z+ A z + C = 0 (C = 2R).
Conversely, setting
A — P — Qi, A = B = P + Qi,
we can rewrite the equation
Az + B z + C = 0
as
(P — Qi) (X + yi) + (P + Qi) (X — yi) + C = 0
Basic Definitions. Theorems and Formulas 385
or
2Px + 2 Q y + C = 0,
which is a first-degree equation.
The equation
Az + A z + C = 0
is called a self-conjugate equation o f a straight line since the left-hand side
of the equation is a real function of x and y:
u = Az -f- A z -j- C,
u = A z + Az + C = u.
If a straight line is given by the self-conjugate equation
Az -f- Bz C — 0,
where B = A ^ 0 and C is a real number, then the distance d from point
(z0) to this line is
, \Az0 + B z0 + C\
a = ----------------------.
2\A\
Indeed, the equation of the straight line passing through point (z0) per
pendicularly to the given straight line is of the form
Z — z0 = — ( z — z0)
or
Az — Bz — ZqA + z 0B = 0.
From the system
Az + B z + C = 0,
Az — Bz — Az0 + B z o = 0
we find the affix of the projection of point (ze) on the given straight line:
Az0 — B~z0 — C
z =
2A
whence
Az0 -j- B zq -j- C
z9 — z =
2A
and so
\Az%+ B z %+ C|
d = \ z Q- 2 ' I =
2\A\
21—SIS
If a straight line is given by the self-conjugate equation

Az -f- Bz -f- C — 0,
where B = A ^ 0 and C is a real number, then for all points (z) lying to
one side of that straight line we have
Az -f Bz -j- C > 0
(positive half-plane), and for all points (z) lying to the other side of that
line we have
Az -f- Bz -j- C 0
(negative half-plane).
If (z0) is any point lying on the straight line
Az -f- Bz -{- C = 0,
where B = A ^ 0 and C is a real number, then the point (z0 + B) lies in
the positive half-plane since
A(z0 -f- B) + B (z0 + B) + C — Az0 + B zq + C + AB + BB
= AB + BB = BB + BB = 2BB = 2\B\* > 0.
If the sides of A ABC are given by the self-conjugate equations
(BC): Axz -j- B±z -j- C\ = 0,
((7/4) : A2z BoZ -f- C2 = 0,
(AB): A3z + B3z -f- C3 = 0,
where Ck are real numbers and Bk = A k ^ 0, then the area (ABC) of the
oriented A ABC is computed from the formula
Ai B L Cx 2
A 2 B2 C2
i A3 ^ 3 C 3
(ABC)
4 | a 2 b 2 A3 B3 j Ax Bt
\ a 3 b 3 A l Bx \A , B2
The result remains the same if the left-hand sides of the equations (BC),
(CA), (AB) are multiplied by any complex numbers different from zero.
The area^S of A ABC is computed from the formula
A i Bx Cx 2
mod A2 B2 C2
CO
1
S'
A 3 Bz j A\ j\
4 mod I \ A2 B%
' \ A a B3 A-l B- j a 2 b 2 J
LIST OF SYMBOLS
vector
AB a vector with origin at point A and terminus at
point B: a ray (radial line) with origin at A and
passing through B
0 zero vector
1*1 = a magnitude (absolute value) of vector a
\AB\ =- AB magnitude of vector AB
a ij b vectors a and b are collinear
■TM vectors a and b are collinear and in the same
aU b/ direction
•Tib vectors a and b are collinear and in opposite
directions
*9 a vector obtained from vector a by a rotation
through an angle <p in an oriented plane
[*] a vector obtained from vector a by a rotation
through the angle n/2 in an oriented plane
(a, b)l a pseudo scalar (or cross) product of vector a by
a x bf vector b in an oriented plane
(a, b, c) triple scalar product (or cross product or mixed
product) of three vectors in oriented space
ab = a-b scalar product of vectors a and b
a2 scalar square of vector a
[a, b] vector product of vector a by vector b in oriented
space
general basis of vectors in a plane
e1, e2 dual basis of the basis el5 e2
general basis of vectors in space
e1, e2, e3 dual basis of the basis e1? e2, e3
i, J orthonormal basis of vectors in a plane
i, j, k orthonormal basis of vectors in space
a*, b* dual basis of the basis a, b in a plane
a*, b*, c* dual basis of the basis a, b, c in space
(O, e1? e2) general Cartesian system of coordinates in a plane
(O, e2, e3) general Cartesian system of coordinates in space
( O , i, j) rectangular Cartesian system of coordinates in a
plane
(O, i, j, k) rectangular Cartesian system of coordinates in

space
{* ,y} vector in a plane specified by contravariant com
ponents (or coordinates) x, y
{*> y, z) vector in space specified by contravariant com
ponents (or coordinates) x , y, z
[*, jO vector in a plane specified by covariant compo
nents (or coordinates) x, y
[x, y, z] vector in space specified by covariant components
(or coordinates) x, y , z
(x, y) point in a plane specified by coordinates x , y
(x, y, z) point in space specified by coordinates x, y , z
AB (1) a line segment, (2) a straight line passing
through points A and B, (3) the magnitude of the
vector AB, (4) the length of the line segment AB
(AB) oriented length of a line segment (length of seg
ment AB on an oriented straight line with ap
pended sign)
ABC oriented triangle
(ABC) area of oriented A ABC (area with appended
sign)
ABCD oriented tetrahedron
------ ^
(ABCD) volume of oriented tetrahedron ABCD (volume
with appended sign)
ABC \ \ A'B'C'
triangles with the same orientation
A B C \i A'B'C'
ABC i t A'B'C' triangles with opposite orientations
(a, b) (1) the straight line of intersection of planes a
and b, (2) the oriented angle from straight line a
to straight line b
C(AB)D; sometimes dihedral angle with edge AB in the half-planes
simply (AB) of which lie points C and D
AC AD
(ABCD) = anharmonic (or cross) ratio of the points
BC BD A, B, C, D
% Kronecker delta: S{ = ^ ^°r * ^

1 for i = j
Sij fundamental tensor specified by covariant com
ponents
List o f Symbols (H‘>
g iJ fundamental tensor specified by contravammi

components
g Gram determinant
(O, r) circle with centre at point O and radius r
(ABC) circle passing through points A, B, C*
O, (O) the centre of a circle circumscribed about a In
angle, and the circle itself
I, (I) the centre of a circle inscribed in a triangle with
centre at point /, and the circle itself
4 ’ 4> 4> (â)> the centres of circles inscribed in A ABC, and
(4), (4) the circles themselves
Ot, (o 9) the centre of the nine-point circle (Euler’s circle),
and the circle itself
R the radius of a circle (ABC)
r the radius of a circle (I) inscribed in a triangle
ra, rb, rc the radii of circles (/a), (/*), (Ic) escribed in a
triangle
(ABCD) sphere passing through points A, B, C, D
o = a (M, (O, r)) the power of a point M with respect to a circle
(O, r); a — MO2 — r2
(O, k) a homothetic transformation with centre at point
O and ratio k
[O, k) inversion with centre at point O and power of
inversion k
\A modulus of a complex number z
arg z argument of a complex number z
z the conjugate complex of z
GO point having affix z
(*o> R) circle of radius R , the affix of the centre of which
is z0
sign of equivalence
sign of a one-to-one mapping
sign of a one-to-one correspondence
± sign of perpendicularity
II sign of parallelism (collinearity)
# sign of equality and parallelism
tt, U signs of collinearity and identical direction
IT sign of collinearity and opposite direction
symmetric polynomials of affixes zl9 z2, z3
& 19 °* 2 j &Z9 symmetric polynomials of affixes zl9 z2, z8, z4
mod * absolute value of x
♦Thus, the notation (ABC) has the following distinct meanings: (1) the (oriented)
area o f A ABC: (2) a circle passing through A , B, C (no confusion can result because
the first meaning o f the symbol is a number, the second is a figure).
Appendix
LIST OF BASIC FORMULAS

FOR REFERENCES
Quadratic equations
-f p x + q = 0; x 12
- - M t -
ax2 + bx c = (1• *1,2

v - —b± Vb2 —4ac
~ (a * 0).
2a
n -k ± Y k r-a c
ax2 + 2kx + c *= (0; x 1>2 = ---------------------- (a ^ 0).
Progressions
(a) Arithmetic progression.
1. General term of an arithmetic progression:
<*n = <*i + (n - 1) d.
2. The sum of n terms of an arithmetic progression:
c ( <*n\ T + (/! - 1) d '

--------- 2-------- J
where d is the difference.
(b) Geometric progression.
1. General term of geometric progression:
“n =
2. The sum of n terms of geometric progression:
ux - unq 1 - qn qn — 1
Sn = t------- ' ui -------= p
where # is the common ratio of the progression {q / 1).

List of Basic Formulas 391
3. The sum of an infinitely decreasing geometric progression:
5= "i
q'
Logarithms
1. The notation loga N = is equivalent to the notation ax = N (a > 0, a / 1,
log N
N > 0) so that a a = N.
N
2. loga 1 =* 0. 3. loga a = 1* 4. loga(AT- M ) = logfl AT + loga M. 5. loga ---- =
M
1
= loga N A/- 6- loga Af" = « loga W (V > 0). 7. logaK^Vr= — loga AT. 8. log* TV =
n
__ loga AT
loga *
Relationships between
trigonometric functions of an angle
sin a cosa
1. sin2a -f cos2a = 1 . 2 . ------- = tan a. 3. —— = cot a. 4. sin a • cosec a*= 1. 5
cos a sina
cos a • sec a = 1. 6. tan a • cot a = 1. 7. 1 + tan2a = sec2 a. 8. 1 + cot2 a = co sec2a.
Table of Signs and Selected Values of Trigonometric Functions
Quadrants I II III IV
Function
O
o
o
00
©
I 11 Ill IV 0° 45° | 60° 90° 270° 360°

1 }f2 V£
sin a + + - - 0 1 0 -1 0
2 2 2
Vi 1
cos a + - - + 1 V
1 0 -1 0 1
2 "2 *2
vv
tan a + — + — 0 1 lr3‘ 00 0 00 0
3
cot a + - + - 00 1'T 1 K 0 00 0 00
3
Table of Reduction Formulas
Angle
—a 90° =F a 180° =f a 270° a 360° k a
Function^'"'"-
sin — sin a + cos a ± sin a — cos a sin a
cos + cos a ± sin a — cos a =F sin a + cos a
tan — tan a db cot a =f tan a ± cot a =F tan a
cot — cot a ± tan a | dF cot a ± tan a d1 cot a
Transformations of trigonometric expressions

1. sin(a ± p) = sin a cos P ± cosa sin p;
cos(a ± P) = cos a cos q= sin a sin P;
tan a ± tan P
tan (a ± /J) = ——------ —
1=Ftan a tan P
2. sin 2a = 2 sin a cos a;
cos 2a = cos2a — sin2a = 1 — 2 sin2a = 2 cos2a — 1;
2 tan a
tan 2a =
1 — tan2 a
+ COS a
3. sin i . ± y i r p iM1- . ± y i
2
COS a 1 — cos a sin a
tan
+ cos a sin a 1 + cos a
x x
2 tan — 1 - tan2 2 tan
2
4. sin * : -; cos * = tan jc =
1 + tan2 1 + tan2 — 1 — tan2
2
1 -f cos 2a 1 — cos 2a
5. cos2 a = ---------------; sin2 a = -------------- .
2 2
a ± P a ip P
6. sin a ± sin P = 2 s m ------- c o s ---------;
2 2
a -f- P a —p
cos a + cos p *= 2 c o s --------c o s ---------;
2 2
a ^ . a + 0 . a —, „ . a + 0 . P —a
cos a — cos p *= — 2 s m --------s i n -------- ‘ 2 s m --------s m ---------;
2 2
„ sin(a ± 0) sin(ft ± a)
tan a ^ tan 0 *= -------------- ; cot a ± cot p 1
cos a cos /5 sin a sinp
7. cos m* -cos nx — — [eos(m — n) x + cos(m + n) x ];
sin mjc sin nx = — [cos(/n — n) x — cos(#i + n) x ];

List o f Basic Formulas 393
Inverse trigonometric functions

n
1 . ------- < arc sin - sin(arc sin x ) = x.
2 27
2. 0 ^ arccos x ^ ny cos(arccos x) = x.
n n
3 . ------------- arctan x < —, tan(arctan x) = x.
2 2
4. 0 < arccot x < n, cot(arccot x) = x.
Elementary trigonometric equations

1. sin x =» a, x = ( — I)* arc sin a 4- nn.
2. cos x = af x = ± arc cos a 4- 2tt/i.
3. tan x — a, x = arc tan a -f rc/i.
4. cot * = at x = arc cot 4 - w i.
Symmetric Polynomials
For three numbers z l9 zZt z3
<Tl — •Z’l 4“ ^2 4" ^3»
0*2 = Z 1Z 2 + ^ 2 Z 3 4 " ^ 3 Z 1»
°3 = Z 1 Z 2 Z 3*
For four numbers z x> z2, z3, z4
<Ti = Z i + Z2 + ^3 + Z 4>
az = zâ + riZ3 4- z xzk + r2z3 -f zaz4 -f z3z4,

<*3 = Z 1Z 2Z 3 + z i z 2z 4 + z i z 3z 4 + Zzz zz ^
°4 = Zl Z2Z3Z4*
Kronecker delta
BIBLIOGRAPHY
1 Coxetcr, H. S. M., Greilzer, S. L. Geometry Revisited, Random House, New York,
1967 (New Mathematical Library, Vol. 19).
2. Deaux, R. Introduction to the Geometry o j Complex Numbers, Ungar, New York,
1956.
3. Dubnov, Ya. S. Basic Vector Calculus, Parts 1 and 2, (Osnovy vektornogo ischi-
sleniya), Gostekhizdat, 1950, 1952 Moscow (in Russian).
4. Golovina, L. I., Yaglom, I. M. Induction in Geometry, Mir Publishers, Moscow, 1979.
5. Johnson, R. A. Advanced Euclidean Geometry (Modern Geometry), Dover, New
York, 1960.
6. Modenov, P. S., Parkhomenko, A. S., Geometric Transformations, Academic Press,
New York, 1965.
7. Pedoe, D. A Course o f Geometry for Colleges and Universities, Cambridge University
Press, Cambridge, 1970.
8. Schwerdtfeger, H. Geometry o f Complex Numbers, University o f Toronto Press,
Toronto, 1962.
9. Wooton, W., Beckenbach, E. R., Fleming, F. J. Modern Analytic Geometry, Houghton
Mifflin, Boston, 1975.
10. Yaglom, 1. M. Complex Numbers in Geometry, Academic Press, New York, 1968.
11. Yaglom, I. M. Geometric Transformations I-IV, Random House, New York, 1962,
1968, 1973 (New Mathematical Library, Volumes 8,21, 24).
12. Yaglom, I. M. Geometric Transformations (Geometricheskie preobrazovaniya),
Vol. 2, Chap. II (in Russian), Gostekhizdat, Moscow, 1956.
13. Yaglom, I. M. A Simple Non-Euclidean Geometry and Its Physical Basis, Chapter II
and Supplement, S. Springer-Verlag, New York, 1979.
NAME INDEX
Apollonius of Perga 52 Kroneckcr 393
Beckenbach, E.R . 394 Lagrange, J. L. 41

Brianchon, C. J. 44 Langer, I. 256
Blanchard, R. 8, 261, 264 Lemoine 71, 233
Boutain 85
Brocard 74 Markushevich, A. I. 8
Marmion, A. 41
Carnot 77 Mascheroni, Lorenzo 310
Ceva 48 Menelaus 44
Chasles, Michel 160 Modenov, P. S. 394
Coxeter, H. S . M. 394 Mohr, Georg 310
Monge 40
Morley 227, 267
Deaux, R. 8, 260, 266, 394
Desargue, Girard 45 Parkhomenko, A. S. 394
D iodes 306 Pascal, Blaise 278
Dubnov, Ya. S. 7, 394 Peaucellier 308
Pedoe, D . 394
Euclid o f Alexandria 281 Pilatti 260
Euler, Leonhard 291 Poncelet, J. V. 327
Ptolemy 286
Feuerbach, K. W. 205
Fleming, F. J. 394 Schlomilch 27
Schwerdtfeger, Hans 394
Serret, J. A. 41
Gibbs 343, 346 Simson 86
Golovina, L. I. 394
Gourmagschieg 8, 259
Gram 342, 345 Thebault, V. 41, 264
Greitzer, S. L. 394 Tschirnhaus 278
Hamilton 262
Hart 308 Vatricant, S. 260
Vieta 390
Ilin, V. A. 9
Iyenger 41 Wooton, W. 394
Jebeau, V. 8
Jensen 322 Yablonsky, S. V. 9
Johnson, R. A. 394 Yaglom, 1. M. 394
SUBJECT INDEX
affine transformation 279 orthocentroidal 182

affix of a point 379 unit 382
algebra, vector 1 Iff* zero 281, 376
analytic geometry 44ff, 347ff o f zero radius 281
problems with hints and answers 66ff circular plane 281
antiparallel lines 297 cissoid o f D iodes 306
antiparallelogram 308 cofactor 336
antireciprocal equation 244 collinear points 350
Apollonian problem 294 complex numbers 377ff
Apollonius, circle of 52 absolute value o f 377
argument is congruent t o . . . modulo . . . amplitude of 378
378 argument o f 378
axis, radical 50 conjugate 379
modulus o f 377
trigonometric form of 378
barycentric coordinates 352, 354 use o f in plane geometry 82ff
base points 53, 327 contravariant components 343, 346
bases, dual 343, 345 coordinates
basis 339 barycentric 352, 354
orthogonal 339 general Cartesian system o f 347
orthonormal 339 origin of 347
Brianchon theorem 44 rectangular Cartesian system of 347
bimedians of tetrahedron 78 coplanar points 350
Boutain points 85 covariant components 343, 346
Brocard lines 74 cross product 341
Brocardians 74
bundle of planes 368
centre o f 368 definitions 334flf
improper 368 deformation ratio 55
proper 368 de Moivre formula 379
Desargue’s theorem 45
determinant(s) o f order three 334ff
canonical equations o f a line 372 elements o f 334
Carnot's theorem 77 Gram 342, 345
centroid o f tetrahedron 78 diagonal, principal 275
centroid o f a triangle 32 D iod es, cissoid o f 306
Ceva's theorem 48 directed line segments, equal (or equi
cimedian 71 valent) 8
Chasles theorem 160 direction vector 354
circle(s) Droz-Famy theorem 196
o f Apollonius 52 dual bases 343, 345
o f infinite radius 281
o f inversion 281
Euler 32 elliptic pencil, o f circles 53
nine-point 32 equation (s)
Subject Index 397
antireciprocal 244 Jensen inequality 322

canonical (of a line) 372
elementary trigonometric 393
intercept form o f (of a line) 356 Kronecker delta 393
intercept form o f (o f a plane) 366
parametric (of a line) 356
quadratic 390 latitude 318
self-conjugate 386 Lemoine point 71, 233
o f a sphere 376 limaôn, Pascal’s 305
equipollency, sign o f 167 limit points 52, 53, 327
Euclidean circular plane 281 line(s)
Euler circle 32 Brocard 74
Euler’s formula 291 Simson 86
logarithms 391
longitude 318
Feuerbach point(s) 205, 296
formula(s) 334ff
de Moivre 379 mapping o f regions under inversion 297
Euler’s 291 Mascheroni, geometry o f 3091T
Gibbs’ 343, 346 mechanical inversors 308
formula(s) (cond.) medians o f a tetrahedron 78
list of basic 390ff Menelaus* theorem 44
Vieta’s 390 meridian 318
function(s) metaparallel triangle 279
inverse trigonometric 393 metaparallelism 279
trigonometric (relationships between) midperpendicular 32
391 minor 336
fundamental tensor 341 mod a: 361
Monge's point 40
Morley relation 230
geometry
analytic 44ff, 347ff
problems with hints and answers 66ff nine-point circle 32
of Mascheroni 309ff number(s)
plane 66ff complex (see complex number) 82„
solid 78ff 377fF
Gibb’s formulas 343, 346
Gram determinant 342, 345
oriented space 344
orthocentre 32
orthocentric tetrahedron 40
Hamilton’s theorem 262
orthocentroidal circle 182
harmonic quadruplet (set) 52
orthogonal basis 339
Hart cell 308
orthologic triangle 280
hyperbolic pencil o f circles 53
orthologicality 280
orthonormal basis 339
orthopole of line 94, 149
ideal point 281
identical transformation 281
inequality, Jensen 322 pair
inversion 28Iff, 284 left-hand 341
mapping of regions under 297ff right-hand 341
problems involving 285ff with negative orientation 341
o f space 313ff with positive orientation 341
inversors, mechanical 308 parallels 318
involuntary transformation 281 parallel projection 55
isogonally conjugate points 227 Pascal’s limagon 305
398 Subject Index
Pascal’s theorem 278 sliding vector 347

Peaucellier cell 308 slope
pencil o f circles, elliptic and hyperbolic complex (of a line) 381
53, 327 o f a line 356
pencil o f lines 358 solid geometry 78fF
pencil o f planes 367 sphere
Pilatti’s theorem 260 equation o f 376
plane o f inversion 314
circular 281 stereographic projection 58, 315
Euclidean circular 281 symbols, list o f 387
oriented 341 symmetric polynomials 393
plane geometry 66 symmetry 281
point(s)
affix o f 379
base 53, 327
Boutain 85 tangential triangle 276
collinear 350 tensor, fundamental 341
coplanar 350 tetrahedron, degenerate and nondegenerate
Feuerbach 205, 296 352
ideal 281 theorem(s) 334ff
at infinity 281 Brianchon 44
isogonally conjugate 227 Carnot’s 17
Lemoine 71, 233 Ceva's 48
limit 52, 53, 327 Chasles 160
M onge’s 40 Desargue’s 45
Poncelet 52, 53, 327 Droz-Farny 196
power of 376 Hamilton’s 262
unit 85 Menelaus’ 44
Poncelet points 52, 53, 327 Pascal’s 278
power o f a point 376 Pilatti's 260
problem, Apollonian 294 Ptolemy’s 286
product Schlomilch’s 27
cross 341 o f sine 15
pseudoscalar 341 transformation(s)
scalar 340 affine 279
triple scalar 344 identical 281
progression(s) 390 involuntary 281
arithmetic 390 similarity 380
geometric 390 Tschirnhaus 278
projection transversal 45
parallel 55 triangle(s)
stereographic 315 degenerate 350
pseudoscalar product 341 metaparallel 279
pseudosquare 275 mirror-similar 162
Ptolemy’s theorem 286 nondegenerate 350
oriented 350
orthologic 280
radical axis 50 tangential 276
radius vector 347 trigonometric expressions, transformations
ratio, deformation 55 of 392
triple
left-hand 344
scalar product 340 right-hand 344
Schlomilch’s theorem 27 with negative orientation 344
self-conjugate equation 385 with positive orientation 344
sheaf of planes (see bundle of planes) 368 triple scalar product 344
Simson line 86 Tschirnhatis transformation 278
Subject Index 399
unit circle 382 in the plane (solved problems) 1Iff

unit point 85 principal 355, 364
sliding 347
in space (solved problems) 14ff
zero 337
vector(s) 337
collinear 337
components of 339 vector algebra llff, 337
coordinates of 339 Vieta’s formulas 390
coplanar 337
direction 354
equal 337
linearly dependent 338 wedge 30
linearly independent 338
magnitude of 337
nonzero 337 zero circles 281, 376
normal 355 zero vector 337
TO THE READER
Mir Publishers would be grateful for your comments

o f the content, translation and design of this book. We
would also be pleased to receive any other suggestions
you may wish to make.
Our address is:
Mir Publishers
2 Pervy Rizhsky Pereulok,
1-110, GSP, Moscow, 129820
USSR
Printed in the Socialist Republic o f Romania

Problems in Geometry - Modenov

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Problems in Geometry - Modenov

Uploaded by

Copyright:

Available Formats

rr. c .

MIR PUBLISHERS • MOSCOW

© TjiaBHaA pegaKiuiH $H3HKo-MaTeMaTnqecKoft JurrepaTypu

CHAPTER I. VECTOR ALGEBRA

CHAPTER II. ANALYTIC GEOMETRY

CHAPTER III. THE USE OF COMPLEX NUMBERS

CHAPTER IV. INVERSION

CHAPTER V. BASIC DEFINITIONS, THEOREMS AND

Sec. 3. Analytic g e o m e t r y ..................................................................................... 347

LIST OF SYMBOLS .............................................................................................. 387

APPENDIX. LIST OF BASIC FORMULAS FOR REFERENCES 390

B IB LIO G R A PH Y ............................................................................... 394

NAME IN D E X ........................................................................................ 395

This text offers certain general methods of solving problems in elemen­

z a i± b _ ^ a d _ bc # z' = QZ ° {ad — be ^ 0),

zf = az + b(a # 0), z' = az + b(a ^ 0),

embraces the isometric transformations of the first and second kind

the first and second kind (z' = az + b, z' = az + 6, a ^ 0) and circular

where a = CB, b == CA. Consequently

which can also be written as

Furthermore, since CD\\{ab + fca), CA/|j(a + b), it follows that

Problem 4. Given vectors a = CB, b ™ CA. Find the vector x = CO,

Sec. 2. Vectors in space (solved problems)

is perpendicular to the rays OB and OC and forms an acute angle with

any direction vectors of the rays OA, OB, OC.

These last relations can be rewritten as

4°. Express the length d of the diagonal OD in terms of the angles

Multiplying both sides of each of these formulas by a, b, c respectively

6°. Let us consider the vectors p = OP, q = OQ, assuming that

Furthermore, let a*, b*, c* be the reciprocal basis of a, b, c. Expand the

Solution. On the edges of the given trihedral angle, choose points A, B, C

16 sin2 — sin2 — sin2 —

Solution. Consider the vectors a = OA, b = OB, c = OC, p OP,

— c cos a + ft + cos y(c cos P — b cos y)

OB OC = {OO' + O'B) (OO' + O'C) = OO'2 + O'B-O'C - 0 <*>

are derived in similar fashion.

2°. [ M , OC] = [ 0 0 + OB, OO + ~OC]

= [O'O, OC] + [OB, O'O) + [OB, OC]

[O'B, O C] OOf = [OB, 0 C \-0 0 '. (**)

to 7!, trc j-O O ' - | [ ^ , (TC]\ 0 0 '

[0B,^C] 0 0 ' = Oi!'OC’0 0 ' c o s | - - = OB OC-OO' sin «!

and, hence, by virtue of (**) we have

and noting that

Problem 6. On the base of the polyhedron OACBPQ lies a rectangle

orthogonal projections Px and Qx of points P and Q on the plane OABC

Sec. 3. Vectors in the plane and in space

1. Given two vectors CA = b, CB — a ^ 0. Let P be the orthogonal

2. Four points A, B ,C ,D are located (in the plane or in space) so that

otPA + fiPB + yPC = 0

= (IE + JF + ID + DPJ == IG DP.

Thus, the point Ga is constructed in the following manner: from point G

8. Suppose H, G, O, and 0 9 are, respectively, the orthocentre of A ABC,

but we also have

5°. If P is an arbitrary point lying within A ABC, then, denoting by

the opposite sides. For these minimal distances to be equal, it is necessary

14. Find the vectors x and y if we know their sum x + y = a, the

of the triangle obtained in the section coincides with the projection of

/_ COA = P, /_ AOB = y. Find the radius x of the sphere (5) circum­

are respectively opposite the plane angles. In item 6 ° of example 1, Sec. 2,

x - [DB, DC], y = [DC, DA], z = [DA, DB], t = [AC, AB].

x= [b, c], y = [c, a], z = ^ -[a , b].

y , = 4 “ (x >y> z) = — ([b, C], [c. a], [a, b])

48 ( a>b>c)2 = i48( 6 F)2 = i4 - v s •

Proof‘ Note that all the ratios

This text offers certain general methods of solving problems in elemen

Furthermore, let a, b, c* be the reciprocal basis of a, b, c. Expand the

/_ COA = P, /_ AOB = y. Find the radius x of the sphere (5) circum

29. Drawn through the midpoint of each diagonal of a convex quad