Chapter 11

Isometry

Definition 11.1. A bijective map of the plane onto itself is called a transformation. A transforma-
tion that preserves distances is called an isometry or a movement of the plane.

Theorem 11.1
The inverse of an isometry is also an isometry. Therefore, all isometries in the plane are
a group with respect to composition.

Notation. We will use the following notations:

I: identity transformation,
α
RA : rotation around point A by an angle α (in the positive direction),
SP : half-turn about the point P , also known as central symmetry about the point P ,
Sk : reflection about the line k, also known as axial symmetry about the line k,
→
− →
−
T−
→v : translation by vector v , also known as displacement by vector v ,
1 →
−
Sk,−v : glide reflection about line k and vector v .
→

We will characterize the group of plane isometries and we will show that this group consists
of only the isometries mentioned above.

Theorem 11.2
An isometry that fixes two different points A and B fixes every point of the line AB.

1
A glide reflection with axis k and vector →
−
v is a movement that consists in a reflection about the line k and
→
−
a translation by the vector v , which is supposed to be parallel to the line k.
136

Theorem 11.3
An isometry that fixes three non-collinear points A, B, and C is the identity.

Theorem 11.4
If A, B, C are three non-collinear points and f , g are isometries such that f (A) = g(A),
f (B) = g(B), and f (C) = g(C), then f = g.

When a figure presents configuration issues, in order to avoid any ambiguity, it is convenient
to use directed angles (oriented angles) which are defined in the following way.

Definition 11.2. The directed angle between the lines k and ` is the angle by which the line k must
be rotated to get a line parallel to line `. We will denote this angle symbol ](k, `). This quantity is
defined modulo 180◦ , so that we can identify the angles (between directed lines) that differ by 180◦ .
For three noncollinear points A, B, and C, the directed angle ]ABC is defined as the direct angle
between the lines AB and BC, i.e., ]ABC = ](AB, BC). Note that the order of lines in the brackets
matters, in fact ](k, `) = −](`, k).

Composition of Isometries.
Given two movements of the plane f and g, one can construct a third movement g ◦ f ,
the composition, or the product of the given two, by performing first f , then g. The
transformation g ◦ f thus defined is really a movement, because it evidently preserves the
distances between the points. In this section, we will study the composition of special types
of movements: translations, reflections, and rotations.

Theorem 11.5 (Composition of Two Translations)

The composition of two translations T−
→v and T−
→w is a translation or the identity, precisely
we have
w ◦ T−
T−
→ →v = T−v ◦ T−
→ →w = T− v+w .
−→

In particular, if →
−
w = −→
−
v , then T−
w ◦ T−
→ v = T−
→ → = I.
0
 137

Let us denote by S` the reflection in the line `. Suppose that two lines, ` and m, are given,
and we have to find the composition of Sm ◦ S` . Let P 0 be the image of an arbitrary point P
under the movement S` , and P 00 be the image of P 0 under Sm .

We first consider the case when the two lines ` and m are parallel to each other (left figure).
Then all the three points P , P 0 , and P 00 lie on one line, perpendicular to ` and m, and the
distance between the points P and P 00 is twice the distance between the lines ` and m, in-
dependent of the position of P . Therefore, the composition of the two reflections S` and Sm
has the same effect as the translation by the vector 2→ −v where →
−v is the vector perpendicular
to ` and m, of length equal to the distance between the two lines and directed from ` to m:

Sm ◦ S` = T2−
v.
→

Now suppose that the lines ` and m meet at a certain point A (figure on right). If ϕ is the
angle between ` and m, then, as you can see from the figure, ∠P AP 00 = 2ϕ. Note also that all
the three points P , P 0 , and P 00 are at the same distance from A. It follows that

2ϕ
Sm ◦ S` = R A ,

where RA2ϕ
denotes the rotation around A through the angle 2ϕ (clockwise if ϕ < 0 and
counterclockwise if ϕ > 0).

For other locations of point P in the plane, different from that of the figure, the above two
formulae are always true. The following figure demonstrates one such instance.
138

Remember that the angle ϕ should be measured from line ` to line m, i.e., for example,
ϕ = 45◦ means that the line m can be obtained from line ` by a positive (counterclockwise)
rotation through 45◦ .

The composition of two movements in general depends on the order in which they are
taken: thus, S` ◦ Sm is a movement inverse to Sm ◦ S` .

Two formulae defined above must be read from right to left, and show how to decompose a
translation or a rotation into a product of reflections. This decomposition is not unique, and
the freedom we have in the choice of the axes of reflection may prove quite useful for the
solution of a specific problem.

Summing up, we can formulate the following two Theorems.

Theorem 11.6 (Composition of Two Reflections)

The composition Sm ◦ S` of two axial symmetries S` and Sm is:

(i) the translation T2− →

−
v , if lines ` and m are parallel; the vector v is perpendicular to
→

lines ` and m, its origin lies on line ` and its end lies on line m;

(ii) the rotation RA

2α
where A = ` ∩ m and α = ](`, m), if lines ` and m are not parallel.
 139

Theorem 11.7
Let ` and m be two lines.

(i) If ` and m are parallel lines, and `0 and m0 arise by translating them by a given
vector, then Sm ◦ S` = Sm0 ◦ S`0 .

(ii) If ` and m intersect at A, and the lines `0 and m0 arise by rotating around point A
by a given angle, then Sm ◦ S` = Sm0 ◦ S`0 .

Theorem 11.8 (Composition of Three Reflections)

Let `, m, and k be three non-concurrent lines, then S` ◦ Sk ◦ Sm is a glide reflection Sn,−
→
v
→
− →
−
for some line n and with v 6= 0 .

Proof. There are four cases as shown in the following figures.

The proofs of the various cases are similar.

Here we prove only case (d) whereas the cases (a), (b), and (c) are left to the reader.

Let A = ` ∩ m, B = k ∩ m, and C = k ∩ `. The composition Sk ◦ Sm is a rotation about B

through the angle 2β, where β is the directed angle from m to k. Let k 0 be the line through B
perpendicular to `, intersecting ` at C 0 . Let m0 be the line through B such that the angle from
m0 to k 0 is β.
140

Then we have

S` ◦ Sk ◦ Sm = S` ◦ (Sk ◦ Sm ) = S` ◦ (Sk0 ◦ Sm0 ) . (1)

Now, applying the associative property, we get

S` ◦ (Sk0 ◦ Sm0 ) = (S` ◦ Sk0 ) ◦ Sm0 . (2)

◦
The transformation S` ◦ Sk0 is the rotation RC180
0 . Let n be the line through C perpendicular
00 0

to m0 and let m00 be the line through C 0 perpendicular to n00 .

◦
Then we have Sn00 ◦ Sm00 = RC1800 , hence

(S` ◦ Sk0 ) ◦ Sm0 = (Sn00 ◦ Sm00 ) ◦ Sm0 = Sn00 ◦ (Sm00 ◦ Sm0 ) . (3)

From (1), (2), and (3) it follows that S` ◦ Sk ◦ Sm is a glide reflection since m00 and m0 are
parallel and both perpendicular to n00 .

Theorem 11.9 (Composition of Two Rotations)

If A and B are arbitrary points, and α, β are arbitrary angles, then the composition
RBβ α
◦RA is a rotation through an angle α+β about some point O, unless α+β is a multiple
of 360◦ , in that case, the composite map is a translation. In particular, if α = β = 180◦ ,
then RBβ
◦ RAα
= T2− →.
AB

Proof. Let ` be the line AB, let m be the line through A such that ](m, `) = α/2 and let n be
the line through B such that ](`, n) = β/2. Thus ](m, n) = (α + β)/2.
 141

Again, using Theorem 11.6, we get RA

α
= S` ◦ Sm and RB
β
= Sn ◦ S` . Hence
β α
RB ◦ RA = Sn ◦ S` ◦ S` ◦ Sm = Sn ◦ Sm .
Using again Theorem 11.6, we get
• if α + β ≡ 0 (mod 360◦ ), then m k n and the composition is a translation;

• if α + β 6≡ 0 (mod 360◦ ), then m and n meet at some point O, so Sn ◦ Sm = RO

α+β
.

Remark 11.1. If A = B then the result of the composition of two rotations is quite
obvious.

If we rotate a point about A by angle α and then rotate its image about A by angle β, the
result is the same as if we had applied a single rotation about A by angle α + β. In fact,
the order of the rotations is immaterial. We write this relationship as follows:
β α α β α+β
RA ◦ RA = RA ◦ RA = RA .

Corollary 11.10
◦ −α
The composition RB
360 α
◦ RA is identity if and only if A = B or α is an integer multiple
of 360◦ .
142

Remark 11.2. The concept of a directed angle between straight lines can be used to
describe construction of the center (point O) of the composition RB
β α
◦ RA . See below.

Consider two rotations RAα

and RB β
where A 6= B. Let RA α
(P ) = P 0 and RB
β
(P 0 ) = P 00 .
Then we have
β β β
(RB α
◦ RA )(P ) = RB α
(RA (P )) = RB (P 0 ) = P 00 .

By the theory of the composition of rotations (Theorem 11.9), we have

β α α+β
RB ◦ RA = RO .

where O is the point such that ](OA, AB) = α/2 and ](AB, BO) = β/2.
Since ∠P AP 0 = α, ∠P 0 BP 00 = β, if AD and BC are the angle bisectors of ∠P AP 0 and
∠P 0 BP 00 , respectively, then we have

α β
∠P AD = ∠DAP 0 = ∠OAB = and ∠ABO = ∠P 0 BC = ∠CBP 00 = .
2 2
It shows that AP 0 and AO are isogonal lines with respect to ∠DAB. Also, OB and BC
are isogonal lines with respect to ∠ABC.
 143

Theorem 11.11
Let ABC be a triangle oriented counterclockwise and α, β, and γ be three positively
oriented angles such that 0 < α, β, γ < 180◦ . The composition f = RA
α β
◦ RB ◦ RCγ is

(i) a rotation through the angle of α + β + γ, if α + β + γ 6= 360◦ ;

(ii) the identity if α + β + γ = 360◦ , ∠BAC = α/2 and ∠CBA = β/2 and ∠ACB = γ/2;

(iii) a translation if α+β +γ = 360◦ and ∠BAC 6= α/2 or ∠CBA 6= β/2 or ∠ACB 6= γ/2.

Proof. Since 0 < α, β, γ < 180◦ , we have 0 < α + β + γ < 540◦ .

Therefore, if α + β + γ 6= 360◦ then by Theorem 11.9, RA α
◦ RBβ
◦ RCγ is a rotation through the
angle α + β + γ about some point O.
If α + β + γ = 360◦ , by Theorem 11.9, it follows that f is a translation.

Assume ∠BAC = α/2, ∠CBA = β/2, ∠ACB = γ/2, and let C 0 be the reflection of C about
AB.

Then, taking into account that ]CBC 0 = β and ]C 0 AC = α, we get

β
α
f (C) = RA ◦ RB ◦ RCγ (C) = RA
α β
◦ RB α
(C) = RA (C 0 ) = C.

Thus, C is a fixed point and this implies that f is the identity map.

If ∠BAC 6= α/2 or ∠CBA 6= β/2 or ∠ACB 6= γ/2 then f is a non-identity translation. In fact,
if we assume that f = I, then RA −α
= RBβ
◦RCγ , i.e., the point A is the center of the composition
of rotations RB
β
◦ RCγ . This implies that ∠CBA = β/2 and ∠ACB = γ/2 (by Theorem 11.9).
Moreover, since (α + β + γ)/2 = 180◦ , we also get ∠BAC = α/2, which is a contradiction.
144

Theorem 11.12 (Composition of a Rotation and a Translation)

A point A and a vector →
−
v are given.

(i) The composition RA

α
v of translation T−
◦ T−
→ v and rotation RA is the rotation RC .
→ α α

(ii) The composition T−→v ◦ RA of rotation RA and translation T−

α α
v is the rotation RD ,
→ α

where in general C 6= D.

Theorem 11.13
Given a point P , a line m, and a circle ω. Then, using a ruler and compass, we can con-
struct the images of P , m, and ω in a reflection S` , in a rotation RA
α
, and in a translation
T−
→v.

Theorem 11.14
Given two congruent and non-parallel segments AB and A0 B 0 , there exists exactly one
rotation that maps A onto A0 and B onto B 0 . The rotation center is the intersection point
of the perpendicular bisectors of the segments AA0 and BB 0 . The rotation angle is the
angle formed by the lines AB and A0 B 0 or its supplementary.

Proof. The proof can be found in [32, p.67].

Remark 11.3.
In the above theorem, if T is the point of in-
tersection of AB and A0 B 0 then the rotation
center O is the second intersection of the cir-
cles (AT A0 ) and (BT B 0 ). If T and O lies
on the same side w.r.t AA0 and BB 0 , the ro-
tation angle is θ = ](AB, A0 B 0 ) else the ro-
tation angle is 180◦ − θ.
 145

Corollary 11.15
If AB = A0 B 0 and AB ⊥ A0 B 0 then there exists a rotation through an angle of +90◦ or
−90◦ which maps A to A0 and B to B 0 .

Theorem 11.16 (Classification Theorem for Plane Isometries (Chasles2 ))

Every isometry f of the plane onto itself is either the identity map, or a rotation, or a
translation, or a reflection, or a glide reflection. Precisely:

(i) if f is a direct isometry and has at least two fixed points, then f is the identity map;

(ii) if f is a direct isometry and has exactly one fixed point, then f is a rotation;

(iii) if f is a direct isometry and has no fixed points, then f is a translation;

(iv) if f is an opposite isometry and has at least one fixed point, then f is a reflection;

(v) if f is an opposite isometry and has no fixed points, then f is a glide reflection.

Shift-by-a-Vector Method
Sometimes we use a method known as Shift-by-a-vector to solve certain kinds of geometry
problems. We will briefly discuss this method.

Consider points A1 , A2 , . . . , An and B1 , B2 , . . . , Bn . Let Mi be the midpoint of Ai Bi for i =

1, 2, . . . , n. Translate each of points B1 , B2 , . . . , Bn by some vector →
−
v , obtaining new points
2
Michel Chasles (1793-1880) was a French geometer.
146

C1 , C2 , . . . , Cn . Denote by Ni the midpoint of segment Ai Ci (for i = 1, 2, . . . n).

Corollary 5.2 implies that point Ni is obtained from Mi by a translation by vector 12 → −v . In

→
−
other words, shifting each of the points Bi using a fixed vector v , the midpoints of segments
Ai Bi are being shifted by a fixed vector 12 →
−
v.
In particular, it follows that points M1 , M2 , . . . , Mn are collinear if and only if N1 , N2 , . . . , Nn
are collinear, as shown in the following figure:

Theorem 11.17
Let the triangles ABC and A0 B 0 C 0 are congruent and oppositely oriented. Then the mid-
points of AA0 , BB 0 , and CC 0 are collinear.

−−→
Proof. Let’s shift the triangle ABC by vector AA0 . Then point A is taken to A1 = A0 , and
points B and C are moved to some points B1 and C1 , respectively. Using the shift-by-
vector method it suffices to prove that A0 and midpoints of the segments B1 B 0 and C1 C 0
are collinear.
 147

Let k be the perpendicular bisector of segments C1 C 0 . Since A0 C1 = A0 C 0 , line k passes

through A0 . It suffices to prove that k passes through the midpoint of B1 B 0 .

Indeed: consider a reflection in line k. It brings point A1 onto itself, and C1 onto C 0 . Since
triangles A1 B1 C1 and A0 B 0 C 0 are congruent and oppositely oriented, point B1 is mapped
onto B 0 . This implies line k is the perpendicular bisector of segment B1 B 0 , so it contains the
midpoint of the segment.

Examples
In the following examples, we present four well-known results which can be proved using
rotations.

Example 11.1 (Van Schooten’s Theorem). Let ω be the circumcircle of an equilateral triangle
ABC. If P is a point on the arc BC not containing A, then we have P A = P B + P C.

Proof. Rotate 4P CA counter-clockwise at an angle of 60◦ about point C.