1 Plane Euclidean

geometry

The coordinate plane

We start with the familiar plane of analytic geometry. Each ordered pair
(pi, p2) of real numbers determines exactly one point P of the plane. The
point determined by (0, 0) is called the origin.
The ordered pair (pi, p2) is also referred to as the coordinate vector of P.
Although mathematically equivalent, the words "point" and "vector" have
different connotations. A vector is usually thought of as a line segment
directed from one point to another. We may think of the vector (pi,p2) as
the line segment beginning at the origin 0 and ending at P. We shall regard
the words "point" and "vector" as interchangeable, using whichever
suggests the more appropriate picture. The set of all vectors is denoted by
R2.

The vector space R2

If x = {xi, x2) and y = (yu y2), then we define

x 4- y = (xi + yu x2 4- y2).
If c is a real number and x is a vector, then we define
ex = (cxl9 cx2).

These operations are called vector addition and scalar multiplication,

respectively. In particular, if c = — 1, the vector ex is denoted by — x.
The vector 0 = (0, 0) is called the zero vector. The operations of vector
addition and scalar multiplication enjoy the following familiar algebraic
properties:

Theorem 1. For all vectors x, y, and z, and real numbers c and d,

i. (x 4- y) 4- z = x 4- (y 4- z).
8 ii. x 4- y = y 4- x.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 iii. x + 0 = x. The inner-product
iv. x + (-x) = 0. space R2
v. lx = x.
vi. c(x + y) = ex 4- cy.
vii. (c + rf)x = ex 4- dx.
viii. c(dx) = (cd)x.

The inner-product space R2

Given two vectors x and y, we define

(x, y) = xiyi + x2y2-
The number (x, y) is called the inner product of x and y. It is sometimes
also called the dot product or scalar product of x and y.
The following identities concerning the inner product may be easily
checked.

Theorem 2.
i. (x, y + z) = (x, y) + (x, z) /or #// x, y, z e R2.
ii. (x, cy) = c(x, y) for all x, y e R2 and all c e R.
iii. (x, y) = (y, x) for all x, y e R2.
iv. If (x, y) = 0 /or «// x e R2, £/ien y must be the zero vector.

Remark: Theorem 1 says that R2 is a vector space. Theorem 2 says that the
inner product is bilinear, symmetric, and nondegenerate. See Appendix D
for further discussion of these notions.
For any vector x e R2 we define the length of x to be
\x\ = V*l + *2-
Note that
\x\2 = (x, x),
so that length and inner product are intimately related.

Theorem 3. The length function has the following properties:

i. |*| =2= 0 for all x e R2.
ii. // |x| = 0, then x = 0 (the zero vector).
iii. \cx\ = \c\\x\ for all x e R2 and all c e R.

We now state and prove a less immediate property of the inner-product

function and its consequence for length.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Theorem 4 (Cauchy-Schwarz inequality). For two vectors x and y in R2 we
have
\(x, y)\ ^ \x\\y\.
Equality holds if and only if x and y are proportional.

Proof: We restrict our attention to nonzero vectors x and y, the assertion

being obviously true when either x or y is zero.
Consider the real-valued function / defined by
f(t) = \x + ty\2 for t e R.
Using the properties stated above, we observe that/(r) is nonnegative for
all t and that f{t) assumes the value 0 if and only if x is a multiple of y.
On the other hand, / is a polynomial of degree 2. Specifically,
f(t) = \x\2 + 2t(x, y) + t2\y\2 (1.1)
2 2 2
and as such remains nonnegative only if (x, y) ^ |*| |)>| ; that is, \(x, y)\
« \x\\y\.
In addition, f(t) assumes the value zero only if |(JC, y)\ = \x\\y\. Thus,
| (*,)>) | = |x11 y | if and only if x and y are proportional. •

Corollary. For JC, y e R 2 ,

\* + y\ ^ \A + \y\. (1-2)
Equality holds if and only if x and y are proportional with a nonnegative
proportionality factor.

Proof:
\x + \x\2 ; + 2(x, y) + \y\2
\y\2 (1.3)
(W + \y\) , 2

hence, \x + j
If equality holds here, then we must have
(x ,?> = W \y\.
From our work on the Cauchy-Schwarz inequality, we see that x and y
must be proportional. But x = cy leads to
(x, y) = c{y, y) = c\y\2
and

MM = \4y\\y\ = MM2.
10 Thus, c must be equal to \c\; hence, c ^ 0. •

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 The Euclidean plane E 2 Lines

The plane has both algebraic and geometric aspects. When we think of the
algebraic properties, we are thinking of the vector properties of R2.
We now turn to the geometric concept of distance. If P and Q are points,
we define the distance between P and Q by the equation
d(P, Q) = \Q- 4
The symbol E2 will be used to denote the set R2 equipped with the distance
function d.
The concept of distance is a fundamental one in geometry. We will now
derive the most important properties of distance. They are stated in the
following theorem.

Theorem 5. Let P, <2, and R be points ofE2. Then

i. d{P9 Q) & 0.
ii. d(P, Q) = 0 if and only if P = Q.
iii. d(P, Q) = d(Q, P).
iv. d(P, Q) + d(Q, R) s* d(P, R) (the triangle inequality).

Proof: Because d(P, Q) = \Q-P\ = \~(Q - P)\ = \P~ Q\, the first three
properties follow from Theorem 3. The fourth property is equivalent to
showing that
\Q-P\ + \R-Q\& \(Q - P) + ( R - 0 | = \ R - P\.
This, of course, follows from the corollary to Theorem 4. Furthermore,
equality holds if and only if Q — P = u(R — Q) for some nonnegative
number u. In the next section we will see that this implies that P, Q, and R
are collinear. •

Lines

A line in analytic geometry is characterized by the property that the vectors

joining pairs of points are proportional. We define a direction to be the set
of all vectors proportional to a given nonzero vector.
For a given vector v let
1[v]
J = {tv\
L t e RJ} .
'
If P is any point and v is a nonzero vector, then ^*^
€ = {X\ X - P e [v]} (1.4)
1 1 L JJ v
' Figure 1.1 The line € = P + [v\.
is called the line through P with direction [v]. See Figure 1.1. We also write
(1.4) in the form 1 1

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry € = P + [v\.
When € = P + [v] is a line, we say that u is a direction vector of €.
If € is a line and X is a point, there are many phrases used to express the
relationship X e €. The following are synonymous:

i. X 6 €.
ii. € contains X
iii. X lies on €.
iv. € passes though X
V. X and € are incident.
vi. X is incident with €.
vii. is incident with X.
Remark: In axiomatic geometry one usually takes points and lines as
fundamental objects and incidence as a fundamental relation. Then an inci-
dence geometry would consist of sets & and $£ and a relation in 0> x S£.
The relation is assumed to satisfy certain properties from which other
properties of the axiomatic system are deduced. See Greenberg [16]. We
are being more specific here, but our propositions occur as axioms or
propositions in axiomatic developments of plane geometry.

A fundamental property of a line is that it is uniquely determined by any

two points that lie on it. Thus, it is important to mention the following:

Theorem 6. Let P and Q be distinct points of E2. Then there is a unique

line containing P and Q, which we denote by PQ.

Proof: Let v be a nonzero vector. The line P + [v] passes through Q

Figure 1.2 The line PQ and a direction if and only if Q - P e [v]. This means that [Q - P] = [v\. Hence, the line
vector v. P + [Q — P] is the unique line required. See Figure 1.2. •

Thus, a typical point X on the line € = PQ is written

a(0 = P + t(Q - P) = (1 - i)P + tQ. (1.5)
(See Figure 1.3 for a vector addition interpretation.) This equation may be
regarded as a parametric representation of the line. As t ranges through the
real numbers, a(t) ranges over the line. The parameter is related to
t distance along € by the formula
)9 a(t2)) = \t2 - h\\Q - P\. (1.6)
Figure 1.3 a(t) = (1 - t)P + tQ.
If X = (1 - i)P + tQ, where 0 < t < 1, we say that Xis between P and Q.
This algebraic characterization of betweenness is equivalent to the
12 following geometrical one.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Theorem 7. Let P, X, and Q be distinct points ofE2. Then X is between P Lines
and Q if and only if
d(P9 X) + d(X9 Q) = d{P9 Q).
Proof: Suppose first that X is between P and Q. Then for some t e (0, 1),
X = (1 - i)P + tQ.
Then
d(P9 X) = \X - P\ = \t(Q - P)\ = t\Q - P\.
Also,
d(X9 Q) = \Q - X\ = |(1 - t)(Q - P)\ = (1 - i)\Q - P|,
X
hence, * ^ ^
Q
d(P9 X) + d{X9 Q) = t\Q - P| + (1 - t)\Q - P\ / ^
= \Q - P\ = d{P9 Q). / ^
Conversely, suppose that X is a point of E2 satisfying d(P, X) + d(X, Q) ^
= d(P, Q). As we saw in Theorem 5, there is a positive number u such that Figure 1.4 d(P, X) + d(X, Q) >
d(P, Q).Xis not between P and Q.
X - P = u(Q - X).
Solving for X gives

Setting r = w/(l + w), we see that 0 < f < 1, while I - t=l/(l + u), so that J^"^
X = (1 - f)P + tQ. Thus, Xis between P and g . D ^ ^
Figure 1.5 <Z(P, X ) + d(X, Q) =
Remark: Theorem 7 is illustrated in Figures 1.4 and 1.5. d(P, Q).Xisbetweenpand Q.

Let P and 2 be distinct points. The set consisting of P, Q, and all points
between them is called a segment and is denoted by PQ. P and Q are the
end points of the segment. All other points of the segment are called
interior points.
If M is a point satisfying
d(P, M) = d{M, Q) = \d{P, G),
then M is a midpoint of PQ. It follows easily from Exercise 8 that each
segment has a unique midpoint, namely,

M = i ( P + Q).
If two lines € and m pass through a point P, we say that they intersect at P
and P is their point of intersection. From this point of view we restate part
of Theorem 6. 13

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Theorem 8. Two distinct lines have at most one point of intersection.

As we shall see later, two lines in E2 may have no point of intersection at

all.
If three or more lines all pass through a point P, we say that the lines are
concurrent. If three or more points lie on some line, the points are said to
be collinear.

Orthonormal pairs

Two vectors v and w are said to be orthogonal if (v, w) = 0. It is frequently

desirable to have a vector available that is orthogonal to a given vector. If
v = (vl9 u 2 ), we define i / = (—v2, t^). Clearly, v and v1- are orthogonal
and have the same length. We also easily see that
vL± = -v.
A vector of length 1 is said to be a unit vector. A pair {u, w} of unit
orthogonal vectors is called an orthonormal pair.

Theorem 9. Let {u, w} be an orthonormal pair of vectors in R2. Then for

all x e R2,
x = (x, v)v + (x, w)w.

Proof: Because v and w are linearly independent, they form a basis for R2
(see Appendix D). Thus, for any x e R2, there exist unique constants X and
X |x such that x = Xv 4- JJLW. But then, using the fundamental properties of the
inner product, we get
(x, v) = \(v, v) + |x(w, v) = X
and
(x, w) = X{v, w) + |x(w, w) = |x. •
Figure 1.6 Theorem 9.
x = (x, v)v + (x, w)w.
Remark: Theorem 9 is illustrated in Figure 1.6.

The equation of a line

If € is a line with direction vector u, the vector ir1 is called a normal vector
to €. Clearly, any two normal vectors to the same line are proportional. We
now derive a characterization of a line in terms of its normal vector. See
14 Figures 1.7 and 1.8.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Theorem 10. Let P be any point and let {v, N} be an orthonormal pair of The equation of a line
vectors. Then P + [v] = {X\(X - P, N) = 0}.

Proof: By Theorem 9 we have the identity

X - P = (X - P, v)v + (X - P, N)N
for any point X in R2. We show that X lies on the line P + [v] if and only if
(X - P, N) = 0.
First, suppose that X = P + tv for some real number f. Then
(X - P, N) = (tv, N) = f(v, AT) = 0.
Conversely, if (X — P, N) = 0, the identity reduces to Figure 1.7 {X - P, N) * 0. X does not
lieon€ = P + [v].
X - P = (X - P, v)v,
so that
X = P + (X - P, u)v 6 P + [v\. • X

N
Corollary. //iV w any nonzero vector, {X\(X — P, iV) = 0}
through P with normal vector N and, hence, direction vector A
Fi ure L 8
Proof: Just observe that (X-P,N)=0 if and only if (X - P, N/\N\) = 0 8 <^ ~P,N)= 0. Zlies on
and apply the theorem. D

We recall from elementary analytic geometry that {(x, y)\ax + by + c

= 0} should represent a line, provided that a2 + b2 =£ 0. This fits into our
scheme as follows:

Theorem 11. Let a, b, and c be real numbers. Then {(x,y)\ax + by + c = 0}

is
i. the empty set if a = 0, b = 0, awd c =£ 0,
ii. the whole plane R2 if a = 0, b = 0, and c = 0,
iii. a line with normal vector (a, b) otherwise.

Proof: Cases (i) and (ii) are obvious. Consider now the case where
a2 + b2 =£ 0. One can check that the set in question is not empty. In fact, at
least one of the points (-c/a, 0) and (0, -c/b) must be defined and satisfy
the equation.
Let P = (xi, yi) be any point satisfying the equation. Then c =
— (axi 4- byx). Thus ax + by + c = 0 if and only if a(x - x{) + b(y - yi) =
0. Letting N = (a, b), we see that the set in question is just the line through
P with normal vector N. •
15

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Perpendicular lines

Two lines € and m are said to be perpendicular if they have orthogonal

direction vectors. In this case we write Z L m. Two segments are
perpendicular if the lines on which they lie are perpendicular.
An important manifestation of perpendicularity is the famous theorem
of Pythagoras.

Theorem 12 (Pythagoras). Let P, Q, and R be three distinct points.

Then \R - P\2 = \Q - P\2 + \R - Q\2 if and only if the lines QP and
RQ are perpendicular.

Proof: We recall formula (1.3):

\x + y\2 = \x\2 + 2(x, y) + \y\2.

We note that \x + y\2 = \x\2 + \y\2 if and only if (x, y) = 0. Now put
x = Q — P and y = R — Q. We see that x + y = R — P, and, hence,
\R ~ P\2 =\Q~ P\2 +\R- Q\2 if and only if (Q - P, R - Q) = 0. This
means that the segment PQ and the segment QR are perpendicular. •

Figure 1.9 Perpendicular lines and their The next property of perpendicular lines is more evident intuitively than
orthonormal direction vectors. Pythagoras' theorem but more difficult to prove. See Figure 1.9.

Theorem 13. / / € 1 m, then € and m have a unique point in common.

Proof: Let € = P + [v] and m = Q + [w]. We may assume that v and w are
unit vectors, so that {v, w} is an orthonormal set. We write
P-Q=(P-Q,v)v+(P-Q, w)w,
and, hence,
fX P - (P - Q, v)v = Q + (P - 0 , w)w.
I
I Setting
F = P - (P - Q, v)v = Q + (P - Q, w)w,
we see that F lies on both € and m.
Fis the only common point, because if there were two, by Theorem 8 the
Figure 1.10 Dropping a perpendicular lines would have to coincide. •
to € from X
This result allows us to obtain a result motivated by a construction of
Euclid.

Theorem 14. Let X be a point; and let £ be a line. Then there is a unique
16 line <m through X perpendicular to €. Furthermore,

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 i. ^ = 1 + [N], where N is a unit normal vector to €; Parallel and
ii. € and m intersect in the point F = X — (X — P, N)N, where P is any intersecting lines
point on (,;
iii. d(X,F) = \(X- P,N)\.

Remark: The construction of m when € and X are given is called erecting a

perpendicular to € at X if X happens to lie on (,. Otherwise, it is called
dropping a perpendicular to € from X. In this case the unique point of
intersection of € and m is called the foot F of the perpendicular. Theorem
14 is illustrated in Figures 1.10 and 1.11.

Theorem 15. Let € be any line, and let X be a point not on €. Let F be the Figure 1.11 Erecting a perpendicular to
foot of the perpendicular from X to €. Then F is the point of € nearest to X. € a t X
(See Figure 1.12.)

Proof: Let P be any point on €. Because PF 1 FX, Pythagoras' theorem

gives \X - P\2 = \X - F\2 +\F- P\2. Thus, \X - P\2 ^ \X - F\2 with
equality if and only if P = F. •

Definition. The number d(X, F) is called the distance from the point X to
the line € and is written d{X, €).
Figure 1.12 F is the point of € closest to
Remark: d(Xy €) is the shortest distance from X to any point of €. X.

Corollary. Let € be a line with unit normal vector N. Let X be any point of
R2. If P is any point on €, then
d(X,€) = \(X-P,N)\.
We now present another useful construction involving perpendicularity.
Let PQ be a segment. The line through the midpoint M of PQ that is M Q
<—>
p e r p e n d i c u l a r t o PQ is called t h e perpendicular bisector of the segment
PQ. See Figure 1.13.

Remark: T h e p e r p e n d i c u l a r bisector consists precisely of all points that Figure 1.13 The midpoint and
a r e equidistant from P and Q. perpendicular bisector of a segment.

Parallel and intersecting lines

Two distinct lines € and m are said to be parallel if they have no point of
intersection. In this case we write € || m.
In light of the exercises for the section on lines, if € is any line, P is any
point on €, and v is any direction vector of €, then € = P + [v].
We have the following criterion for parallelism. 17

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Theorem 16. 7Vo distinct lines € and ^ are parallel if and only if they have
the same direction. (Recall that the direction of a line P + [v] is the set [v].)

Proof: Suppose that € and m have a common point F. We may write

€ = F + [v] and <m = F + [w] for nonzero vectors u and w. Because € and ^
are distinct, [v] =£ [w].
Conversely, suppose that € and m have different directions [t>] and [w].
Let P be any point of €, and let <2 be any point of m. Because v and w are
not proportional, there exist numbers t and 5 such that P — Q = tv 4- sw.
(See Appendix D.) This means that P - tv = Q + sw. Let F = P - tv =
Q + sw. Then Fis a common point of € and ^ . •

Figure 1.14 € || m, m \\ «, and€ ||, Parallel lines come in families, one for each direction. A line *n
perpendicular to one member € of the family is also perpendicular to all the
others. Thus, it is possible to parametrize the family by the real numbers,
essentially by measuring distance along m. Although these facts are
intuitive and familiar, it is necessary to point them out explicitly here in
order to compare them to the analogous situations in non-Euclidean
geometries.
We leave the proofs of these to the exercises. Cases (i)-(iii) are
illustrated in Figures 1.14-1.16, respectively. Figure 1.16 also illustrates
Theorem 18.

Theorem 17.
Figure 1.15 € || m and m _L n imply i. / / € || <m and <m || ^ , then either € = n or € || n.
£ In. ii. / / € || m, and m ± n, then € 1 n.
iii. If I ± n and m L n, then € || <m or € = m.

Theorem 18. Let € and m be parallel lines. Then there is a unique number
d(€, m) such that
d(X, €) = d(Y, m) = d(€, m)
for all X e m and all Y e €. In fact, ifN is a unit normal vector to € and m,
then for any points X on m and Y on €,
\(X-Y,N)\ =d(t,»*).

Figure 1.16 £ 1 «, «« 1*, and € | Thus, parallel lines remain "equidistant." Intersecting lines, on the
other hand, behave as follows:

Theorem 19. Let € be any line, and let m be a line intersecting € at a point
P. Let v and w be unit direction vectors of £ and *n, respectively. Let a(/) =
•P + tw be aparametrization of*n. Then d(a(t), €) = |f||(w, v±)\. Thus as
X ranges through m, d{X, €) ranges through all nonnegative real numbers,
18 each positive real number occurring twice. See Figure 1.17.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Reflections Reflections

Any subset of the plane is called a figure. Naturally some figures are more
interesting than others. Figures with a high degree of symmetry are most ,m
interesting, not only because of aesthetic considerations but also because
they occur in nature. Snowflakes, molecules, and crystals are three
examples of objects with symmetric cross sections.
The simplest kind of symmetry that a plane figure can have is symmetry
about a line. See Figure 1.18. We now formulate this notion precisely. Let
€ be a line passing through a point P and having unit normal N. Two points
X and X' are symmetrical about € if the midpoint of the segment XX' is the
foot F of the perpendicular from X to €. See Figure 1.19. In other words, X
and X' are symmetrical about € if we have Figure 1.17 d(a(t), F) = \t\\ (w,

\{X + X') = F.
By Theorem 14 this means that
\X + \X = X - (X - P, N)N,
\X' = \X - (X - P, N)N,
X' = X - 2(X - P, N)N.
In order to consider symmetry about various lines, it is convenient to
adopt a dynamic approach by expressing the relationship between X and
X' in terms of a transformation that takes X to X'.

Definition. For a line t the reflection in (, is the mapping il^ of E2 to E2

defined by Figure 1.18 Afigurethat is symmetric
about the line €.
tl€X = X - 2(X - P, N)N,
where N is a unit normal to t and P is any point of €.

If 3F is a figure such that fl€ 3F = J% then we say that ^ is symmetric

about €. The line € is called a line of symmetry or axis of symmetry of J*\
We now investigate some of the properties of reflections. X'
Theorem 20.
i. d([l€X, a € Y) = d(X, Y) for all points X, Y in E2.
ii. fl€fl€X = X for all points X in E2.
iii. fl € : E 2 —» E 2 is a bijection.

Proof:
Figure 1.19 X and X' are related by
i. Ci X - Y - 2(X - Y, N)N. Thus, reflection in the line €.
\X-Y\2-4((X-Y, 7V»2+4«X-Y, N))2(N, N)
\x- y|2. 19

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry ii. Write Q,€X = X - 2XN, where X = (X - P, N). Then
- 2XN - P, N)N
= X - 2XN - 2(X - P, N)N + 4k(N, N)N
= X - 2XN - 2XN + 4XN
= X.
iii. We first show that H€ is injective. If £1€X = fl^Y, then fi€fl€X =
fl€fl€Y and X = Y, by (ii). To show that fl€ is surjective, let Y be any
point of E2. Let X = £1€Y. Then fl^X = Y, so that Y is in the range of
a€. a
Theorem 21. CI£X = X if and only ifXet.

Proof: X - 2(X - P, N)N = Xft and only if (X - P, N)N = 0; that is,

(X - P,N) = 0. The statement now follows from Theorem 10. •

Remark: A fixed point of a mapping T is a point X satisfying TX = X.

Thus, Theorem 21 says that thefixedpoints of a reflection are those which
lie on its axis.

Remark: Theorem 20 shows that reflections are involutive distance-

preserving bijections. We study distance-preserving bijections (isometries)
in the next section. To say that a mapping T is involutive means T2 = TT
= /, the identity mapping of E2. (See also Appendix B.)

Congruence and isometries

If $F is any figure and H€ is any reflection, then £l€ JMs called the mirror
image of 3F in the line €. The figure and its mirror image are observed to
have the same "size" and "shape." If ilm is a second reflection, then
Figure 1.20 Successive reflections ilmfle J^is again the same size and shape as J*\ (See Figure 1.20.) One may
think of moving $F "rigidly" in the plane so that it coincides with ft^ft^ ^
The key property that makes precise our intuitive notions of size, shape,
and rigid motion is that the distance between each pair of points on J^ is
equal to the distance between the corresponding pairs of points on
n^Q € J*\ We introduce the general concept of distance-preserving map-
ping or isometry as follows:

Definition. A mapping T ofE2 onto E2 is said to be an isometry iffor any X

and Y in E2,
20 d(TX, TY) = d(X, Y).

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 are
Definition. Two figures 3F\ and congruent if there exists an Symmetry groups
isometry T such that

We showed in the previous section that every reflection is an isometry.

Although not every isometry is a reflection, we shall see later that every
isometry is the product (composition) of at most three reflections. Thus,
reflections are the basic building blocks of isometries.
Every isometry Tis a bijection of E2 onto E2. In fact, if TX = TY, then
0 = d(TX, TY) = d(X, Y),
so that X — Y. Therefore, the inverse mapping T~x exists. In fact, T~l is
also an isometry because
d(T~xX, T~XY) = d{TT~lX, TT~XY) = d(X, Y).
Furthermore, if T and S are isometries, then
d{TSX, TSY) = d(SX, SY) = d(X, Y).
We now state these results formally.

Theorem 22.
i. If T and S are isometries, so is TS.
ii. If T is an isometry, so is T~l.
iii. The identity map I of E2 is an isometry.

In other words, the set of all isometries is a group called the isometry
group of E2. It is denoted by </(E2). Figure 1.21 An equilateral triangle and
its axes of symmetry.

Symmetry groups

Let f be a figure in E2. Then the set

Sf(&) = {T e J(
is a subgroup of </(E2) called the symmetry group of 2F. The fact that
y(#") is a subgroup can be easily verified. The size of the symmetry group
of J* is a measure of the degree of symmetry of the figure. We shall show,
for example, in a later chapter that an equilateral triangle (Figure 1.21) has
a symmetry group of order 6 generated by reflections in the three medians. Figure 1.22 An isosceles triangle and its
The isosceles triangle ABC (Figure 1.22) has a symmetry group of order 2 axis of symmetry.
generated by reflection in the median AM. The circle has an infinite
symmetry group generated by reflections in all diameters. For an
elementary discussion of symmetry groups, see Alperin [2]. 21

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Translations

Is there a simple way to describe the product of two reflections? In this

section we answer that question affirmatively in the case where the axes of
reflection are parallel.
Specifically, if m and** are parallel lines, we may choose P arbitrarily on
m and choose Q to be the foot of the perpendicular from P to n. Then if N
is a unit normal vector to m (and hence to n), we get

- p, N)N
= x-2(x-Q, N)N - 2(x - P, N)N + 4<x - 0, N) (N, N)N
= x + 2(P - <2, N)N
= x + 2{P- Q). (1.7)
The last step uses the fact that PQ is perpendicular to <m.

Definition. Let € be any line, and let m and n be perpendicular to €. The

transformation ftmft,, is called a translation along £. If m, -=h n, the
translation is said to be nontrivial.

Remark: When two lines in E2 are perpendicular to €, they are, of course,

parallel. On the other hand, when two lines are parallel, there is a line (in
fact, infinitely many lines) perpendicular to both. In the geometries we will
study later in this book, these properties will fail to be true. In the
projective plane, for example, two lines can be perpendicular to a third line
but still not be parallel. In the hyperbolic plane, on the other hand, there
are parallel lines with no common perpendicular. Thus, if our terminology
here seems more specific than necessary, it is being set up so that it will be
applicable to the other geometries we study as well.

We now see that in the Euclidean plane, a translation does not

determine a line uniquely, although it does determine a parallel family. In
the exercises you will be asked to prove the following:

Theorem 23. Let T be a translation along t. If i1 is any line parallel to £,

then T is also a translation along V.

We also observe that each translation along € has the effect of adding a
direction vector of € to each vector in the plane.

Theorem 24. Let T be a nontrivial translation along €. Then € has a

direction vector v such that
22 Tx = x + v (1.8)

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 for all J C G E 2 . Conversely, if vis any nonzero vector and i is any line with Translations
direction vector v, then the transformation T determined by (1.8) is a
translation along €.

Proof: Let N be a unit direction vector for €. Let P be an arbitrary point of

E2. Let a and p be lines perpendicular to €. (See Figure 1.23.) Let a and b
be the unique numbers such that P + aN e a and P + bN e p. Our formula
(1.7) becomes
ftaftpjc = x + 2(P + aN - P - bN)
= x + 2(a - b)N
If T =£ /, we must have a =£ b, so that 2(a-b)N is the required direction
vector.
Conversely, suppose that for each real number A. we define a mapping Tx
by Figure 1.23 ftaftp is the translation
along € by an amount equal to twice
= x (1.9) d(a, p). Three successive positions X, X'',
and X" of a typical point are shown.
If a and 6 are any two numbers such that X = 2(a - fc), we construct
a = P + aN + [AT1] and p = P + WV + [Nx] and observe that TK = n a n p .
D

We now investigate the group of all isometries generated by reflections

in lines perpendicular to €. First we must introduce some new terminology.

Definition. The set of all lines perpendicular to a given line t, in E2 is called a

pencil of parallels. The line € is a common perpendicular for the pencil. See
Figure 1.24.

We note that taking any line m in E2 together with all lines parallel to m
would be an equivalent construction.
So far we have discovered that the product of two reflections in lines of a
pencil of parallels is a translation along the common perpendicular €. We
now investigate further the algebraic structure of the set of isometries
formed by reflections of such a family.
We begin with the translations. We denote the set of all translations Figure 1.24 A pencil of parallels with
i o u rro A XTO / ^ common perpendicular €.
along € by TRANS(€).

Theorem 25. TRANS(€) is an abelian group isomorphic to the additive

group of real numbers.

Proof: We adopt the notation of Theorem 24. Then

= Tx(x + ixN) = x + + KN
= x + (|x + \)N = 23

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Similarly,
7^ o TX(X) =
Because X + |JL = JX 4- X, translations along € commute.
Further, setting X = 0 yields To = I and r x ° T_x = To = /, so that

Thus, TRANS(€) is a subgroup of ,/(E 2 ). Furthermore, the mapping

X -* Tk of R to TRANS(€) is an isomorphism. This is seen by observing
that To = /, Tx-X = T_x, and TXT^ = TX+[L. D

Let & be the pencil of all lines that are perpendicular to a line €. We
denote by REF(^) the group generated by all reflections of the form ft^,
where m e 0>. In other words, REF(^) is the smallest subgroup of J>(E2)
containing all such Clm. In turn, TRANS(€) is a subgroup of REF(^).
In order to discuss the algebra of REF(^), we need to be able to
compute the product of any number of reflections in the family determined
by 0>. We already know that in our notation, fl a H p = Tx.
We now take three lines, a, p, 7, of 3P corresponding to the numbers a,
b, and c. Then

^ H p a ^ = a a (x + 2(6 - c)N)
= fla(x + \LN), where |JL = 2(b - c),
= x + jxN - 2<x + y,N - P - aN, N)N
= x - 2(x - P, N)N + (2a - |x)N
= x - 2(x - P, N)N + 2(a - b + c)N
= x - 2(x - (P + (a - b 4- c)N), N)N.
We recognize the right side as the formula for reflection in the line 5 e 0>
passing through the point P + dN, where d = a — b + c.
Thus, the product of three reflections in lines of ^ is a fourth reflection
in a line of the same pencil 0*. This is our first instance of a three reflections
theorem, which plays such an important role in classifying the isometries of
plane geometries.

Theorem 26 (Three reflections theorem). Let a, (3, and 7 be three lines of a

pencil 0> with common perpendicular €. Then there is a unique fourth line 8
of this pencil such that
= a8.
There are many ways in which a given translation may be represented as
the product of two reflections. Using the three reflections theorem, we can
24 exhibit this flexibility precisely.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Theorem 27 (Representation theorem for translations). Let T = n a llp be Translations
any member o/TRANS(€). If m and n are arbitrary lines perpendicular to
€, there exist unique lines m' and n such that
T = nmnm. = o,,a.
Proof: Apply the three reflections theorem to m, a, and p to produce a
unique line m such that £lj£lj\$ = (lm>. Then multiplying both sides by
£lm yields n a Op = ft^ft^. The line n is obtained analogously. •

Corollary. Every element of REF(^) is either a translation along € or a

reflection in a line of 0>.

Proof: This is clear from the following group multiplication table, which
summarizes the facts we have established.

Here, we have temporarily indexed the reflections Ct by numbers rather

than lines. Thus, for example, Cia is short for fl a , where a = P + aN +
[N^. •

Let v be any vector in E2. We define TV, the translation by v, by

TVX = x + v.
(If u = 0, TV = /, and we have the fr/Wfl/ translation.) Although all
translations arise in the manner we have described, it is possible to discuss
in an elementary way the product of two translations that are not along the
same line. We get

Theorem 28. The set ^~(E2) of all translations is an abelian subgroup of

2

Proof: The following equations are easy to verify and imply the
conclusions of the theorem:
-*-• 'v*w 'v+w
2. T0 = /.
3. T_U = (T U )" 1 . •

Corollary. ^(E 2 ) is isomorphic to the group R2 with vector addition. 25

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Rotations

We now investigate the product of reflections in two intersecting lines.

Let € = P 4- [v] be a line with unit direction vector v. There is a unique
real number 6 e (—TT, TT] such that
v = (cos 0, sin 0). (See Theorem IF.)
The unit normal tr1 can be written as
N = (-sin 0, cos 0).
We now try to express O€ in terms of 0. First note that
= x - 2(x - P, N)N,
- P = x - P - 2(x - P, N)N.
Let €0 be the line through 0 with direction [v]. Then

I V = * - 2<*> #>#•
Thus, we see that
a€x - P = n €o (* - p ) ,
or
ft€x = H€o(x - P) 4- P.
In other words,
_P. (1.10)
We first deal with (1€Q and use (1.10) to return to the original situation.
For any x note that
(x, N) = -xx sin 0 + x2 cos 0.
Thus, writing our vectors as column vectors, we get

2<
-" s i ° e + ^ c o s
_ ["(1 - 2 sin2 6)^! + (2 sin 9 cos 0)^21
~ L(2 sin 6 cos O)JCI + (1 - 2 cos2 6)x2J

= [cos 20 sin 2 9 ] U ]
V
L sin 29 -cos2eJLc 2 J" '
2 2
In other words, D,((i: R —» R is linear. We denote its matrix (see
Appendix D) by the symbol ref 9. This matrix represents reflection in the
line through the origin whose direction vector is (cos 9, sin 9):

ref 9 = [ C ° S 2G Sin 2 9
1
26 L sin 29 - c o s 29.1'

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 We now investigate the matrix algebra of these reflections. First we Rotations
consider another line m through P and the associated line m§. Then if
(cos 4>, sin 4>) is a direction vector of <m,
Si 2(9
ref ft ref 4A>-- [ C0S[ ^2 ( 6^ ~ "2 (e -"4>)J1"
~cos
We have a special symbol, rot 8, for a matrix of the form
_ [ cos 9 ~ s m G]
L sin 8 cos 8 J '
X"
Because this linear mapping takes the standard unit basis vector 8!
to v = (cos 8, sin 8) and takes e2 to v1- = (—sin 8, cos 8), it is reasonable to
think of rot 8 as a rotation by 8 radians in the positive sense. We must keep
in mind, however, that definitions of angle, radians, or sense have not yet
been given. We now define rotation in such a way that rot 8 is a rotation
about the origin.

Definition. / / a and (3 are lines passing through a point P, the isometry

fl a flp is called a rotation about P. (See Figure 1.25.) The special case a = (3 Figure 1.25 a a a p is the rotation about
is allowed so that the identity is (by definition) a rotation about P no matter P by twice 0. Three successive positions
what P is. If a rotation is not the identity, we refer to it as a nontrivial ^ ^ ' ^ n d Z " of a typical point are
J J J
shown.
rotation. If a 1 p, the rotation O^Op is called a half-turn.

Theorem 29. The set of all rotations about the origin is an abelian group
called SO(2).

Proof: Using the formulas from Appendix F, it is easy for us to check that
the identities
rot 8 rot $ = rot(8 + <(>) = rot(c|> + 8) = rot c>
| rot 8,
rot(O) = /,
(rot 8)" 1 = rot(-8)
hold. •

The symbol SO(2) stands for the special orthogonal group of E2.

Theorem 30.

i. ref 9 rot <>| = ref (e - y J .

( 8\
2/
iii. ref 8 ref 4> ref \\t = ref (8 — + 27
Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Proof: (i)

[cossin 2828 sin 28 IF cos § - s i n c()

- c o s 28JLsin (J) cos $

= ["cos (28 - ((>) sin (28 -

~ Lsin(28 - c|>) -cos(28 - <
Equation (ii) is essentially the inverse of (i). Finally, applying (i), we can
verify (iii) directly as follows:
ref 8 ref <>
j ref i|> = ref 8 rot 2(<J> - i|i)
= ref (8 - <>
| 4- v|/). •

Theorem 31. The set of all rotations about the origin and reflections in lines
through the origin is a group called the orthogonal group and is denoted by
O(2). SO(2) is a subgroup of index 2 in O(2).

Proof: The following group multiplication table is drawn from the facts we
have established.
ref rot P

ref 8 rot 2(8 - <f>) ref(e -

2/

rot a ref(cf>
i) rot (a + P) a
The set of reflections is a coset complementary to the coset SO(2).
Let P be any point. The set & of all lines through P is called the pencil of
lines through P. We denote by REF(^) = REF(P) the smallest group of
isometries containing all fL, where € e ^ . We denote by ROT (P) the set
of all rotations about P.

Theorem 32. Let & be the pencil of all lines through a point P. Then
REF(^) = O(2) and ROT(P) = SO(2).

Proof: For each T e O(2), T P ° 7° T_ P is in REF(P). In fact, iP ref 8T_ P is a

reflection, whereas TP rot 8 T_ P is a rotation.
It is easy to verify that this provides an isomorphism of O(2) onto
REF (P) and of SO(2) onto ROT(P). •
We are now ready to prove the analogues of Theorems 26 and 27 for
pencils of concurrent lines.

Theorem 33 (Three reflections theorem). Let a, p, and 7 be three lines

28 through a point P. Then there is a unique line 8 through P such that

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Glide reflections

Proof: If

^a = TF(ref 6 ) T _ P , H p = TP(ref

and
Cly = Tp(ref i|i)T_p,

we should choose 8 so that

=
f^s TP(ref(G — (J) + I|/))T_ P .

In other words, 8 is the line through P with direction vector

(cos(6 - 4> + i);), sin(6 — <>
| + i)/)). •
Theorem 34 (Representation theorem for rotations).Let T = Ct{ be any
member of ROT(P), and let ( be any line through P. Then there exist unique
lines m. and m.' through P such that
T=

Proof: This is similar to the proof for translations. •

Glide reflections

We now have three basic types of isometries: reflections, translations, and

rotations. A fourth type, the glide reflection, is defined to be a reflection
followed by a translation along the mirror. See Figure 1.26. Specifically, if
€ = P + [u], the glide reflection defined by € and v is given by
7va€x = x - 2(x - P, N)N + v,
Figure 1.26 A figure and its image by a
where iV is the unit vector ^/IH- Note that glide reflection with axis €.

= x + v - 2(x + v - P, N)N
= x + v - 2(x - P, N)N
because (v, N) = 0. Thus, the reflection and translation making up the
glide reflection commute. It will be shown that every isometry is one of the
four types: reflection, translation, rotation, or glide reflection. Because
TV = / is a possibility, each reflection is also a glide reflection. However,
glide reflections of this type are said to be trivial.
We have dealt with products of reflections in three lines of the same
pencil (parallel or concurrent). As an illustration of the power of the tools
we have now developed, we analyze the product of reflections in any three
lines. 29

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Theorem 35. Let a, £, and 7 be three distinct lines that are not concurrent
and not all parallel. Then fl a flpn 7 is a nontrivial glide reflection.

Proof: Assume first that a meets (3 at P. Let € be the line through P

perpendicular to 7. Let Fbe the point of intersection of € and 7. Using the
representation theorem for rotations, we know that there is a line m
through P such that
n a f i p = VLmVLe and {\flpO, = n ^ n € n r
Let n be the line through F perpendicular to <m, and let n be the line
through F perpendicular to n. Now

= n^n, = HF,
the half-turn about F. As a consequence,

Note that Clmiln' is a translation along n. Because F does not lie on ^ , n

and *n are distinct. Thus, ftaftpfl7 is a nontrivial glide reflection.
If a does not meet (3 but, instead, p meets 7, apply the same argument to
a 7 n p n a = ( f ^ a p O ^ " 1 . If we deduce that ft7ftpfta = T V H € , then ft«n3i\
= (TV£1€)~1 = ile^-v — T_V11^ is also a nontrivial glide reflection. •

Theorem 36. Let T be a glide reflection, and let O a be any reflection. Then
O a r is a translation or rotation.

Proof: Let € be the axis of the glide reflection T. There are two cases to
consider.
CASE l: € intersects a. Let P be a point of intersection. By the
representation theorem for translations, we may write T = O^fl^n^, where
a passes through P, and both a and 6 are perpendicular to £. Then

But now a, €, and a all pass through P. By the three reflections theorem
there is a line c through P such that

Thus n a r is either a translation or a rotation.

CASE 2: € || a. Then

Noting that 6 L t and <? 1 a, we see that flaft^ and ile[l^ are distinct
30 half-turns. By Exercise 26, H a r i s a translation. •

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Definition. An isometry that is the product of afinitenumber of reflections Structure of the
is called a motion. isometry group

Theorem 37. Every motion is the product of two or three suitably chosen
reflections.

Proof: Suppose that a sequence of reflections is given. If the sequence has

length greater than three, we choose any four adjacent elements of the
sequence. Applying either Theorems 35 and 36 or one of the three
reflections theorems, we can write the product of these four reflections as
the product of two. This procedure can be continued until fewer than four
reflections remain. •

Corollary. The group of motions consists of all translations, rotations,

reflections, and glide reflections.

Structure of the isometry group

Our main theorem in this section is the following:

Theorem 38. Every isometry of E2 is a motion.

Proof: Let T be an arbitrary isometry. We consider several cases.

CASE 1: r(0) = 0. In this case we show in the next lemma that T = rot 9
or ref 0 for some value of 8. In other words, T e O(2).
CASE 2: T(P) = P for some point P. Then T-PTTP is an isometry
leaving 0 fixed and is, hence, a member of O(2) by Case 1. Thus,
T = TP(rot 0 ) T _ P or T = TP(ref 9)T_ P . In either case, T is a motion.
CASE 3. T has no fixed points. Let P = T(0). Then T_ P ° T leaves 0 fixed
and is, hence, either rot 9 or ref 9. In any case, T = TP rot 9 or T = T P ref 0,
so that Tis a motion. •

Remark: Case 2 could have been handled as part of Case 3. However, the
representation obtained this way is more useful.

We now prove the lemma referred to in Case 1. Although the notion of

isometry depends only on distance, the lemma shows that an isometry
leaving the origin fixed has a particularly nice algebraic form - it must be
linear. 31

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry Lemma. // T is an isometry with T(0) = 0, then
i. (Tx, Ty) = (x, y).
ii. T = rot 0 or T = ref 6 for some 0.

Proof:

i. By the polarization identity (see Exercise 29),

(x, y) = i ( | x | 2 + \y\2 - \ x - y\2),
(Tx, Ty) = i ( | 7 x | 2 + |7>| 2 - \Tx - Ty\2).
Now
17x| = d(0, Tx) = d(T(0), Tx)
= d(0, x) = |JC|.
Similarly,
\Ty\ = \y\.
Also
\Tx - Ty\ = d(Tx, Ty) = d(x, y) = \x - y\.
ii. Let 8i = (1, 0) and e2 = (0,1). If x = XXEI + x2e2, then {Tex, Te2} is an
orthonormal basis for E2. Hence,
Tx = (Tx, TE^T*! + {Tx, Te2)TE2.
Using the result of (i), we obtain
Tx = {x, ei)rei + {x, e2)Te2 = xxTzx + x2Te2.
Now TEI is a unit vector. Writing

we see by Theorem 9 and the Cauchy-Schwarz inequality that

M = l<rei, ei>| ^ |r ei || ei | = I.
Similarly,
|X2| ^ 1 and |r e i | 2 = X? + \ i ,
so that
X? + X| = 1.
As in our discussion of rotations, there is a unique 0 e (—IT, TT] such
that Xx = cos 0 and X2 = sin 0. Now
2, r e i > = <e2, 6l> = 0,
so that
32 Te2 = ±(Te1)±.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 In other words, Fixed points and
fixed lines of isometries
Te2 = ±((-sin 6)8! + (cos 6)e2).
Writing this in matrix form, we have that either

Tx sin
LSI 9 cos

or
[cos 6 sin 8
Tx = -z- x
Lsin 9 —cos 0
for all x e E2. D

Fixed points and fixed lines of isometries

Theorem 39.
i. A nontrivial translation has no fixed points.
ii. A nontrivial rotation has exactly onefixedpoint, the center of rotation.
iii. A reflection has a line of fixed points, the axis of reflection.
iv. A nontrivial glide reflection has no fixed points.
v. The identity has a plane of fixed points.

Proof: Suppose that T is an isometry with no fixed points. As we have

shown in the previous section, T = TP rot 6 or T = TP ref 6 for suitable P
and 6. In the first case
Tx = (rot 0)x + P
for all x, so that Tx = x if and only if (/ - rot 0)x = P. But
det(/ - rot 9) = (1 - cos 0)2 + sin2 9
A

= 2 - 2 cos 6 = 4 sin2 —.
Z*

2
Now sin (0/2) = 0 if and only if rot 6 = 7. Thus, unless rot 0 = 7, the
equation (7 — rot 0)x = P has a solution (see Appendix D), and thus Thas
a fixed point. Because T has no fixed point, rot 0 = 7 and T = TP.
Conversely, of course, a nontrivial translation has no fixed point.
We now examine the second case, T = TP ref 0. Observe that T is the
product of three reflections. Using Theorem 35, we see that T is a
nontrivial glide reflection. Conversely, a nontrivial glide reflection can
have nofixedpoints. For if Q€TV is a glide reflection with € = P + [v], and

x = + v) = x + v - 2(x + v - P, N)N,
then, because (v, N) = 0,
v = 2(x - P, N)N. 33

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry But then
(v,v) = 2{x - P,N)(N,v) = 0,
so that v = 0.
Thus, an isometry has no fixed points if and only if it is a nontrivial
translation or glide reflection. In particular, statements (i) and (iv) hold.
Let T be an isometry with just one fixed point P. From Case 2 of the
previous section, Tis a rotation about P or a reflection in a line through P.
By Theorem 21 the fixed point set of a reflection consists of the axis of
reflection itself. Hence, T must be a rotation. Conversely, a nontrivial
rotation has exactly one fixed point. For if
T = TP(rot 0)T_ P ,
then x is a fixed point if and only if
x = P + (rot 9)(JC - P);
that is,
(/ - rot O)(JC - P) = 0.
Again, because det (/ — rot 9) =£ 0, the only possibility is x — P = 0; that is,
x = P. Thus, an isometry has exactly one fixed point if and only if it is a
nontrivial rotation. This implies (ii).
Statement (iii) is just Theorem 21, and statement (v)is trivially true. This
completes the proof. •

Corollary. The fixed point set of an isometry must be one of the following:
i. a point (rotation)
ii. a line (reflection)
iii. the empty set (translation or glide reflection)
iv. the whole plane E2 (the identity).

If € is any line and T is an isometry, then Tl is a line, as we will see in

Chapter 2. An isometry induces a bijection T: J>f—> <£ of the set <£ of lines
onto itself. If € is a line such that TZ = €, we say that € is afixedline of T. It
is useful to classify the isometries of E2 with respect to their fixed lines.

Theorem 40.
i. A nontrivial translation along a line € has a pencil of parallels as its
fixed lines. This pencil consists of all lines parallel to €.
ii. A half-turn centered at C has the pencil of lines through C as its set of
fixed lines. A nontrivial rotation that is not a half-turn has no fixed
lines.
iii. A reflection ft^ has the line m and its pencil of common perpendiculars
34 as its fixed lines.

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 iv. A nontrivial glide reflection has exactly one fixed line - its axis. Fixed points and
v. The identity leaves all lines fixed. fixed lines Of isometries

When trying to understand the effect of a particular isometry, de-

termination of its fixed points and fixed lines is a good starting point. These
notions and techniques apply in a wider context and will be pursued further
in later chapters.
We will defer the proof of Theorem 40 until we have developed more
convenient algebraic machinery.

EXERCISES

1. Prove Theorem 1.
2. Prove Theorem 2.
3. Prove Theorem 3.
4. Fill in the details required to obtain the expression for/(r) in formula
(i.i).
5. Show that the result of the corollary to Theorem 4 can be used to
obtain the inequality
|*| - \y\ ^ \x - y\.
6. Although P and v determine a unique line €, show that € does not
determine P or v uniquely.
7. If € = P + [v] = Q + [w], how must P, Q, v, and w be related?
8. If 0 < t < 1 and X = (1 - 0 ^ + '(?, and P * Q, show that
d(P, X) = \P - X\ = t\P - Q\ t
d(X9Q) \X-Q\ ( i - , ) | p - e | 1-t'
Use this to find the point X that divides the segment PQ in the ratio
r:s. Illustrate using r = 2, s = 3, P = ( - 3 , 5), Q = (8, 4).
9. If v is a nonzero vector, show that there are exactly two unit vectors
proportional to v.
10. Find an orthonormal pair one of whose members is proportional to
(4, - 3 ) .
11. i. Find all unit normal vectors to the line 3JC + 2y + 10 = 0.
ii. Find all unit direction vectors of the same line.
iii. If P = (5, 2) and v = (i, §), find the equation of the line P + [v]
in the form ax + by + c = 0.
12. If v = (yi, v2) is a direction vector of a line €, the number a = v2lv\ is
called the slope of €, provided that vx i= 0.
i. Show that the concept of slope is well-defined. 35

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry ii. Show that if £ is a line with slope a, the vector (1, a) is a
direction vector of £.
iii. Show that the line through P = (x\, yi) with slope a has the
equation
y - yx = a(x - xx).
13. d(X, €) seems to depend on the choice of P on £ and on the unit
normal vector N. Show that if N' is another unit normal vector to €
and if Pr is another point on €, then
\(X - P, N)\ = \(X - P'9 N')\.
14. Let P + [v] and Q + [w] be intersecting lines. Let D be the matrix
whose first row is v and whose second row is w. If P — tv = Q + sw is
the point of intersection, prove that (f, s) = (P - Q)D~l. Here (t, s)
and P — Q are regarded as 1 x 2 matrices. Use this method to find the
intersection point in the case P = (1, 5), Q = (3, 7), v = (8, 1),
w = (6, 2).
15. Prove Theorems 17-19.
16. Let € and m be parallel lines. Let
n = {\{x + Y)\X e ( and Ye m).
Prove that n is a line parallel to € and ^ and lying midway between
them. In other words, d{m, n) = d(£, n).
17. The definition of (1€ seems to depend on P and N. Show that if P' is
another point on € and N' is any unit normal to €, then, for all points
X,
(X - P, N)N = (X - P', N')N'.
(Compare with Exercise 13.)
18. Prove Theorem 23.
19. Let 9 be a pencil of parallels as discussed in Theorems 25-27'.
i. Show that REF(^) is isomorphic to the multiplicative group of
2 x 2 matrices of the form

Lo
where the reflection ila corresponds to the matrix
-1 2a]
0 1 J'
and Tx corresponds to
X.]
36

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 ii. Observe that TRANS(€) is a subgroup of index 2 in REF(^). Fixed points and
f i x e d l i n e s of
20. Verify the statements in Theorem 28 and its corollary. isometries
21. Prove that
i. ^(E 2 ) is a normal subgroup of J(E2).
ii. If € is a line, TRANS(€) is not a normal subgroup of ,/(E 2 ).
22. Let € = P + [v] be a line. Let m = g + [v]. Show that if |u| = 1, then
ft A * = TW, where w = 2(P - Q, v±)v±,
and

23. Let T^ be any translation. Let € = P + [w1] be any line having w

as a normal vector. Show that if *n = P — j\v + [w1-] and m —
P + jw + [w 1 ], we have

24. Show that the identity is the only rotation that can be described as a
rotation about two different points. The unique point P determined
by a given nontrivial rotation is called the center of rotation.
25. Verify the statements made in the proof of Theorem 32.
26. i. Show that two distinct reflections fl€ and H^ commute if and
only if m _L €.
ii. Let P be any point. Prove that the half-turn about P is given by
HPx = -x + 2P for all x e E2.
iii. Show that the product of two distinct half-turns is a translation
along the line joining their centers.
27. If Hi, H2, and H3 are half-turns, prove that

28. Describe the product of two glide reflections whose axes are parallel.
29. The polarization identity
(x, y ) = i ( | * | 2 + \y\2 - \ x - y\2)
allows us to express the inner product in terms of lengths. Prove it.
30. We know that each element of </(E2) can be written uniquely in the
form Tpa, where a e O(2) and P e R2. Show that the function
TPa —>• a is a homomorphism of J{E2) onto O(2). What is the kernel?
31. Prove that the matrix rot 6 has a nonzero eigenvalue if and only if
rot 6 = ±7.
32. Let a be an isometry such that a" = /. If n is an odd integer, what can
you say about a? Explain. ^ '

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]
 Plane Euclidean geometry 33. Describe the group generated by K in the following cases:
i. K = K } .
ii. K = CLe.
iii. K = {fl€, ft^}, € 1 ^ .

34. If €, *«, and ^ are lines of a pencil, prove that Q,t£lm£ln = il^il^fi^.
35. Let p be a nontrivial rotation with center P. Let v be any vector. Show
thatTvpis a rotation. Find its center in terms of the given informa-
tion.
36. Let P, Q, R, and S be four points, no three of which are collinear. Let
A, B, C, and D be the respective midpoints of the segments PQ, QRy
RS, and SP. Prove that AB \\ CD and AD \\ BC or they coincide.
37. i. Prove the remark following the definition of perpendicular
bisector.
ii. Find the perpendicular bisector of the segment joining (—2, 6)
and (4, 8).

38

Downloaded from [Link] University of Sussex Library, on 24 Jun 2018 at [Link], subject to the Cambridge Core terms of use, available at
[Link] [Link]