Unit 7 Planar Graphs

UNIT 7

PLANAR GRAPHS

Structure Page No
7.1 Introduction 177
Objectives
7.2 Planar Graphs 178
7.3 Euler’s Formula 183
7.4 Characterisation of Planar Graphs 187
Kuratowski’s Theorem
Wagner’s Theorem
7.5 PCB Problem 192
7.6 Duality 196
7.7 Map Colouring Problem 199
7.8 Summary 201
7.9 Solutions/Answers 201

7.1 INTRODUCTION
In this unit we introduce you to another type of graphs called planar graphs. In
earlier days the study of the concept of planarity was motivated by famous
problems such as the Four–Colour Problem and the Utility Problem. Now a
days this concept is also applied in the design of circuit layouts on silicon
chips! The concept of planarity is also very useful in many practical situations
such as electric connection without an extra layer of insulation in the design of
printed circuit board. Therefore, in this unit we shall study plane graphs which
help us investigate the characteristics of planar graphs.

In Section 7.2, we study plane, planar and nonplanar graphs. In section 7.3,
we study the Euler’s Formula, which relates the numbers of vertices, edges
and faces of a plane drawing of a planar graph and study some restrictions on
the number of edges and degrees of vertices on a planar graph.

Section 7.4 is divoted to some characterisations of planar graphs. In this

section, we introduce the notions of subdivision and contraction of a graph,
and study results due to Kuratowski and Wagner, to check planarity. In
Section 7.5, we study about Printed Circuits Board (PCB) layouts Problem,
and introduce the concepts of thickness and crossing. Section 7.6 is divoted to
dual graphs. We end this unit by giving an overview of the Map Colouring
Problem and the Four-Colour Theorem.
177
 Block 2 Graph Tracing and Coluring
Objectives
After studying this unit, you should be able to
• understand the difference between plane and planar graphs;
• check whether a given is planar or not;
• understand why K 5 and K 3, 3 are not planar;

• identify the faces of a planar graph, and compute the face degrees;
• state, prove and apply the Euler’s Formula;
• use Kuratowski’s and Wagner’s Theorems to check planarity;
• understand the (PCB) Problem;
• find the thickness and crossing number of some small graphs;
• draw the dual graph of a plane graph;
• define the face-colouring of a plane graph, and check whether a given is
k -face-colourable or not.

7.2 PLANAR GRAPHS

In this section, we shall study plane and planar graphs. In transistor radios and
An edge-colouring is a
point of intersection of
television sets, you must have seen printed circuit boards. These boards have
two edges, different of slots for various components and these slots are connected to each other. The
two edges, different connections between these slots must be made in such a way that no two
from their endpoints. connections cross each other. Given an electronic circuit, is it always possible
to design a printed circuit board corresponding to it?

This can be formulated as a problem in graph theory. We replace the

electronic components by vertices and the connections between them by
edges. If the resulting graph can be drawn in such a way that no two of the
edges cross each other except at the vertices, then we can design a printed
circuit board for the given circuit. Graphs that can be drawn in this way are
called planar graphs. A formal definition is here.

Definition: A graph G is called a planar graph if it can be drawn in the

plane in such a way that every two edges either do not intersect or their points
of intersection are their endpoints. Any such drawing is called a plane
drawing (or plane embedding) of G. A graph G is called nonplanar graph if
no plane drawing of G exists. A plane drawing of a graph is also known as a
plane graph.

For example, the graph K 4 is planar, since it can be drawn in the plane
without edge-crossings. The following are three plane drawings of K 4 .

178
Unit 7 Planar Graphs
Let us look at some more examples.

Example 1: Show that the following graphs are planar, by finding a plane
drawing of each.

u v u a b

z v
z w

w
y x y x u v w x

(a) (b) (c)

Solution: Note that the above graphs can be redrawn, in order, as follows.

u v u
u

v v
z
z w
a b
w
w
y x
x y x

(a) (b) (c)

Each of these new graphs is a plane drawing of the corresponding graph

above. Therefore, the given graphs are planar.
***

Example 2: Show that the star graph S n is planar.

Solution: The graph S n , n ≥ 2, has n vertices. One of the vertices, say v , is

adjacent to all the remaining n − 1 vertices. Thus a plane drawing of S n can
be obtained by placing v at the centre of a circle, and the remaining vertices
on the circumference of the circle. Therefore, S n is planar.

***

Example 3: Show that every subgraph of a planar graph is planar.

Solution: If G is a planar graph, then we can draw G in the plane without

crossings. If we now remove the vertices and edges not included in the
subgraph, then we obtain a plane drawing of the subgraph. Therefore, the
result holds.
***

Every plane drawing of a planar graph divides the plane into a number of
regions. For example, any plane drawing of K 4 divides the plane into four
regions of which exactly one is unbounded. Three plane drawings of K 4 are
drawn below. In each of these drawings the region marked “4” is unbounded.
179
Block 2 Graph Tracing and Coluring

4 4
4 1 3
1 2 3
1
2 2
3

Plane drawings of K4

Similarly, any plane drawing of K 2 , 5 divides the plane into five regions. In the
drawing below the region marked “5” is unbounded.

1 2 3 4 5

A Plane drawing of K 2 , 5

We make these ideas precise as follows.

Definition: Let G be a plane graph. Then G divides the set of points of the
plane not lying on G into maximal regions, called the faces of G. Exactly one
of these faces is unbounded. The boundary of a face f of G is the
subgraph of G that bounds f .

It is clear that two distinct faces of a plane graph do not have any point in
common, however they may share the same boundary. For instance, the
plane graph G given below has two faces, namely f1 and f 2 . Both these
faces have the same boundary G.

f1 f2

G
Also, note that every edge of a plane graph need not lie on the boundaries of
two faces. For instance, consider the following plane graph G.

f1 f2
e
f4
f3

G
180
Unit 7 Planar Graphs
The edge e of this graph lies on the boundary of the face f 4 , which is shown
is bold edges. The boundaries of f1 , f 2 and f 3 do not contain e.

Note that if G1 and G2 are two isomorphic plane graphs, then G1 and G2
have the same number of faces. In other words, the number of faces of a
plane drawing of a planar graph is uniquely determined. For instance, look at
any of the following plane drawing of the planar graph W5 .

They are pairwise isomorphic, and each has exactly five faces.

By now you must have observed that to determine the face of a plane graph,
one essentially needs to take a closed walk on its boundary. In case the graph
is disconnected, one closed walk may not be enough for every face. For
example, consider the graph given below.

v1 v2 v5
f3
f1 f2

v4 v3
v8 v6 v7

This graph has three faces f1 , f 2 and f 3 . The face f 3 is bounded by two
closed walks, namely (v1 , v2 , v3 , v4 , v1 ) and (v5 , v6 , v7 , v6 , v8 , v5 ). Thus three are
a total of 9 edges that enclose the face f 3 , where the edge v6 v7 is counted
twice. This leads to the following definition.

Definition: Let G be a plane graph, and let f be a face of G. Then the

degree of f , denoted by d ( f ), is the number of edges encountered in a
closed walk (or walks) around the boundary of f .

For example, consider the plane graph G given below.

v10

v9 v4
f3 v1
f2 f4

f1
v8 v5
v2 v3

v7 v6
G
181
Block 2 Graph Tracing and Coluring
There are four faces, namely f1 , f 2 , f 3 and f 4 in G. The face f1 is bounded
by the closed walk (v1 , v2 , v3 , v1 ), which has 3 edges. Therefore, d ( f1 ) = 3.
Similarly, d ( f 2 ) = 3. The face f 3 is bounded by the closed walk
(v1 , v2 , v3 , v4 , v5 , v6 , v7 , v8 , v9 , v10 , v4 , v1 ), which contains 11 edges. Therefore,
d ( f 3 ) = 11. The closed walk (v4 , v5 , v6 , v7 , v8 , v9 , v10 , v4 ) bounds the face f 4 ,
and so d ( f 4 ) = 7.
4
If we now sum all the face degrees, we get ∑ d ( f ) = 3 + 3 + 11 + 7 = 24,
i =1
i

which is exactly twice the number of edges in G.

This makes us suspect that the Handshaking Lemma for graphs has “face
analogue” for a plane graph. This is indeed the case, as stated below.

Theorem 1 (Handshaking Lemma for Planar Graphs): In any plane drawing

of a planar graph, the sum of all the face degrees is equal to twice the number
of edges. That is, if f1 , f 2 , K , f r are the faces of a plane drawing of a planar
graph G, then
r

∑ d ( f ) = 2 m(G).
i =1
i

Proof: In any plane drawing of a planar graph, each edge has two sides
(which may lie on the boundary of a single face or on the boundaries of two
different faces), so it contributes exactly 2 to the sum of the face degrees.
Hence, the result follows.

Example 4: Verify Theorem 1 for each of the following plane graphs.

u u v
u v
z v
z w
z w
w
y x y x
y x
(a) (b) (a)

Solution: (a) There are 8 edges and 4 faces with face degrees 3, 3, 4, 6. Thus
we have 3 + 3 + 4 + 6 = 16 = 2 × 8.
(b) There are 11 edges and 7 faces in this graph, with face degrees 3, 3, 3,
3, 3, 3, 4. Therefore, 3 + 3 + 3 + 3 + 3 + 3 + 4 = 22 = 2 × 11 .
(c) There are 10 edges and 6 faces in this graph, with face degrees 3, 3, 3,
3, 4, 4. Therefore, 3 + 3 + 3 + 3 + 4 + 4 = 20 = 2 × 10 .
***
There are some exercises for you to try.

E1) Consider any face f of a plane graph G. Let S be a set of vertices

lying on the boundary of f . Then the subgraph induced by S is a
182 subgraph of the boundary of f . True or false? Justify.
Unit 7 Planar Graphs
E2) Show that a graph is planar iff each of its components is planar.

E3) For which values of n is the cycle graph C n planar?

E4) Show that the wheel graph Wn is planar for all n ≥ 4.

E5) Show that every proper subgraph of K 5 is planar.

E6) Every graph that contains a nonplanar subgraph is nonplanar. True or

False?

E7) If G and H are two planar graphs sharing exactly one vertex, then
show that G ∪ H is also planar.

E8) Show that every tree is a planar graph.

Thus far we have seen the definition and examples of plane and planar
graphs. In the next section we shall discuss an important formula for such
graphs.

7.3 EULER’S FORMULA

In this section we introduce the Euler’s Formula, which relates the number of
vertices, edges and faces of a plane drawing of a planar graph. We begin with
an example.

Example 5: For each of the plane graphs in Example 4, count the numbers of
vertices (n), edges (m) and faces (r ), and find the value of n − m + r.

Solution: (a) There are 6 vertices, 8 edges and 4 faces, so the required value
is 6 − 8 + 4 = 2.

(b) There are 6 vertices, 11 edges and 7 faces, so the required value
is 6 − 11 + 7 = 2.

(c) There are 6 vertices, 10 edges and 6 faces, so the required value
is 6 − 10 + 6 = 2.
***

In Example 5, we have seen that for each of the plane drawings under
consideration, n − m + r = 2.

This equation holds for any plane drawing of a connected planar graph, and is
known as Euler’s Formula. It tells us that each plane drawing of a given
connected planar graph with n vertices and m edges must have the same
number of faces, 2 − n + m.

Theorem 2 (Euler’s Formula): Let G be a connected planar graph, and let

n, m and r denote, respectively, the numbers of vertices, edges and faces in
a plane drawing of G. Then n − m + r = 2.

Proof: A plane drawing of any connected planar graph G can be constructed

by taking a spanning tree of G and adding edges to it, one at a time, until a
plane drawing of G is obtained. 183
Block 2 Graph Tracing and Coluring
We prove Euler’s Formula by showing that:
i) for any spanning tree, n − m + r = 2;
ii) adding an edge does not change the value of n − m + r.

First, we prove statement (i). Let T be any spanning tree of G; then we can
draw T in the plane without crossings. Since T has n vertices and n − 1
edges, and there is only 1 face the (unbounded one), we have
n − m + r = n − (n − 1) + 1 = 2,
as required.

Now we prove statement (ii) by successively adding to T the remaining edges

of G until a plane drawing of G is obtained. At each stage the added edge
joins two nonadjacent vertices.

Since we have a plane drawing of a subgraph of G, the added edge cuts an

existing face in two. This leaves n unchanged, increases m by 1, and
increases r by 1, and so leaves n − m + r unchanged. Since n − m + r = 2
throughout the process, the result follows.

Example 6: Verify the Euler’s Formula for each of the following planar graphs:
a) the octahedron graph
b) Wn , n ≥ 4
c) K 2,6
d) the graph formed from the vertices and edges of a k × k square lattice.

Solution: (a) A plane drawing of the octahedron graph is given below.

There are 6 vertices, 12 edges and 8 faces, and

n − m + r = 6 − 12 + 8 = 2.
(b) Note that Wn has n vertices, and 2n − 2 edges. Any plane drawing of
Wn has n faces. Therefore,
n − (2n − 2) + n = 2,
which verifies the Euler’s Formula.

(c) A plane drawing of K 2 , 6 is given below.

184
K 2, 6
Unit 7 Planar Graphs
Here n = 8, m = 12 and r = 6. Therefore, n − m + r = 2.

(d) A k × k square lattice has ( k + 1) 2 corners which correspond to the

vertices of the graph. There are 2k (k + 1) edges in it. It has k squares
2

which correspond to the bounded faces of a plane drawing of it. Thus its
plane drawing has k + 1 faces in all. Now, we have
2

n − m + r = (k + 1) 2 − 2k (k + 1) + (k 2 + 1) = 2.
*** A 5 × 5 square lattice

Next, we show how the Euler’s Formula can be used to prove that certain
graphs are nonplanar. We first derive two corollaries of Theorem 2 that give
upper bounds for the number of edges of a planar graph.

Corollary 1: Let G be a planar graph with n (≥ 3) vertices and m edges.

Then m ≤ 3n − 6 .

Proof: Consider a plane drawing of G with r faces. Since a graph has no

loops or multiple edges, the degree of each face is at least 3. It follows from
the Handshaking Lemma for Planar Graphs that 3r ≤ 2m. Substituting r from
the Euler’s Formula r = m − n + 2, we obtain 3m − 3n + 6 ≤ 2m, and hence
m ≤ 3n − 6, as required.

Using Corollary 1, we can prove that the complete graph K 5 is nonplanar.

Example 7: Show that K 5 is nonplanar.

Solution: We prove this by contradiction. Suppose that K 5 is planar. Since

K 5 is a connected graph with 5 vertices and 10 edges, it follows from
Corollary 1 that 10 ≤ (3 × 5) − 6 = 9, which is false. This shows that K 5 is
nonplanar.
***

However, note that we cannot use Corollary 1 to prove that K 3,3 is nonplanar,
as K 3,3 has 6 vertices and 9 edges, and the inequality 9 ≤ (3 × 6) − 6 is true.
Instead, we shall use the following corollary to prove that K 3, 3 is nonplanar.

Corollary 2: Let G be a graph with n (≥ 3) vertices, m edges and no

triangles. Then m ≤ 2n − 4.

Proof: Consider a plane drawing of G with r faces. The degree of each face
of such a drawing is at least 4 as G has no triangles. It follows from the
Handshaking Lemma for Planar Graphs that 4r ≤ 2m, so 2r ≤ m. Substituting
r from the Euler’s Formula r = m − n + 2, we obtain 2m − 2n + 4 ≤ m, and
hence m ≤ 2n − 4, as required.

Example 8: Show that K 3,3 is nonplanar.

Solution: We prove this also by contradiction. Suppose that K 3,3 is planar.

Since K 3,3 is a connected graph with 6 vertices, 9 edges and no triangles, it
185
Block 2 Graph Tracing and Coluring
follows from Corollary 2 that 9 ≤ (2 × 6) − 4 = 8, which is false. Therefore, K 3,3
is nonplanar.
***

Corollary 3: If G is a planar graph, then G contains at least one vertex of

degree 5 or less.

Proof: Let G be a planar (n, m) - graph. If n ≤ 2, there is nothing to prove.

So, let n ≥ 3. Let us assume that all vertices of G have degree greater than or
equal to 6. That is, d (v) ≥ 6 for all v ∈V (G). Therefore, ∑
d (v) ≥ 6n, which
v∈V ( G )
implies 2 m ≥ 6 n. That is, m ≥ 3n. But by Corollary 1, we know that
m ≤ 3n − 6, which is a contradiction. Hence our assumption is wrong. So there
exists at least one vertex of degree 5 or less in G.

Example 9 (Utilities Problem): We wish to connect three houses A, B and

C , to three utilities, gas, water, and electricity. For safety reasons it is
necessary that the various connections should not cross each other. Can the
connections be made? The following picture shows how eight of the nine
connections can be drawn, but how about the ninth?

Solution: We can represent the connections by means of the following graph,

where the vertices correspond to the three houses and the three utilities. This
is the complete bipartite graph K 3, 3 with the sets { A, B, C} and {gas, water,
electricity} as its partite sets.
gas water electricity

A B C

Finding a solution of this problem is equivalent to finding a plane drawing of

K 3, 3 . But K 3, 3 is nonplanar, so it has no plane drawing. Hence, the utilities
problem has no solution.
***
Now, try the following exercises.

E9) a) Let G be a connected planar graph with n (≥ 5) vertices, m edges

Recall that the girth of
a graph is the length of 5
a shortest cycle in it. and girth 5. Then prove that m ≤ ( n − 2).
3
186 (Hint: Use the method of proof of Corollaries 1 and 2.)
Unit 7 Planar Graphs
b) Hence show that the Petersen graph is nonplanar.

E10) By giving a plane drawing, show that Q3 is planar. Is Q4 planar? Justify.

E11) Every nonplanar graph contains at least one nonplanar proper subgraph.
True or false? Justify.

E12) Give an example of each of the following or explain why no such graph
exists.
a) a planar graph in which each vertex has degree 5;
b) a planar graph in which each vertex has degree 6 or more. A graph is said to be
maximal planar if it
E13) Does there exists a planar (6,12) -graph? If yes, draw such a graph. is planar and has the
maximum number of
E14) Let G be a planar graph with n vertices, m edges, r faces and k edges on a fixed
components. Show that n − m + r = k + 1. vertex set.

E15) Give an example of a 4 -regular planar graph on 6 vertices. Is this graph

maximal planar? Why?

The restrictions on the number of edges and degrees of vertices of a planar

graph given in Corollaries 1, 2 and 3 are useful for showing that certain graphs
are nonplanar. For example, we used them to show that K 5 and K 3,3 are
nonplanar. Unfortunately, the method does not work for all graphs. For
example, the Petersen graph satisfy these inequalities but is nonplanar.
Hence, we turn our attention to other ways of determining whether a given
graph is planar.

7.4 CHARACTERISATION OF PLANAR GRAPHS

In this section, we describe two methods for determining planarity. The first
method involves the “insertion” of vertices of degree 2 into the edges of a
graph G , as shown below.
u v u v

After inserting
some vertices
of degree 2

x w x w
G A subdivision of G

Definition: Let G be a graph. The subdivision of an edge uv of G is the

operation of replacing uv with a (u, v) -path P such that
V (G) ∩ V ( P) = {u, v}. A graph H is called a subdivision of G if H is
obtained by subdividing some edges of G.

Consider a graph G and a subdivision H of G. If G is planar, then one can

obtain a plane drawing of H by just adding new vertices on the existing edges
of G, as this operation does not create any edge crossing. On the other hand,
if G is nonplanar, then a new vertex cannot be added at an existing edge -
crossing. This means an edge-crossing in G remains an edge-crossing in H
as well. And hence, H is nonplanar. This leads to the following lemma. 187
Block 2 Graph Tracing and Coluring
Lemma 1: Every subdivision of a planar graph is planar, and every subdivision
of a nonplanar graph in nonplanar.

For example, the following graphs are nonplanar, since the first graph is a
subdivision of K 5 , and the second one is a subdivision of K 3,3 .

A subdivision of K 5 A subdivision of K 3, 3

The next corollary follows immediately from Lemma 1.

Corollary 1: If G is a planar graph, then G contains no subdivision of K 5 or

K 3, 3 .

You may be wondering why we are so concerned with K 5 and K 3,3 and their
subdivisions. The reason is that the converse of Corollary 1 is also true. That
is, if a graph G contains no subdivision of K 5 or K 3,3 , then G is planar.

This result, proved in 1930, is due to Polish mathematician Kazimierz

Kuratowski. We state it formally but omit the proof, which is rather long and
complicated.

Theorem 3 (Kuratowski’s Theorem): A graph is planar if and only if it

contains no subdivision of K 5 or K 3,3 .

Remark 2: The graphs K 5 and K 3, 3 are the two fundamental nonplanar

graphs, sometimes also called Kuratowski’s graphs. The reason is that
every nonplanar graph can be obtained by a sequence of edge subdivisions or
vertex additions from a Kuratowski’s graph. In fact, Kuratowski’s Theorem can
be reworded as follows.

“A graph is nonplanar iff it contains a subdivision of a Kuratowski’s graph.”

Let us look at an example.

Example 10: Use Kuratowski’s Theorem to show that the following graph is
nonplanar.
v6 v5

v7 v4

v8 v3

v1 v2
G
188
Unit 7 Planar Graphs
Solution: Deleting v3 v7 from G we get a subgraph H which is a subdivision
of K 3, 3 as shown below.

v6 v5 v2 v5 v8

v7 v7
v4 v3
≅
v8 v3

v1 v2 v1 v4 v6
H A subdivision of K 3, 3

Thus G contains a subdivision of K 3, 3 . Therefore, by the Kuratowski’s

Theorem, G is nonplanar.
***

Another characterisation of planar graphs involves the notion of “contracting”

an edge. This is done by bringing one endpoint of the edge closer and closer
to the other endpoint until they coincide, and then replacing any resulting
parallel edges by a single edge. Below we present this notion formally.

Definition: Let G be a graph and e = uv ∈ E (G). The contraction of e is the

operation on G to obtain a new graph G ⋅ e by replacing the vertices u and v
by a new vertex puv such that all the neighbours of u and v in G are the
neighbours of puv in G ⋅ e.

Definition: A graph H is said to be a contraction of a graph G if there is a

sequence of graphs (G0 , G1 , G2 , K , Gk ) such that
i) G0 = G,

ii) Gi +1 = Gi ⋅ e for all 0 ≤ i ≤ k − 1,

iii) Gk = H .

For instance look at the graphs G and G ⋅ e given below.

w w

u e v y puv
y

x x
G G ⋅e

We now state the following analogue of Kuratowski’s Theorem due to Wagner

(1937). We omit the proof.
189
Block 2 Graph Tracing and Coluring
Theorem 4 (Wagner’s Theorem): A graph is planar if and only if it contains
no subgraph that has K 5 or K 3,3 as a contraction.

The importance of Theorems 3 and 4 is that they give necessary and sufficient
conditions for a graph to be planar in graph-theoretic terms (subgraphs,
subdivision, contraction of a graph), rather than in geometric terms (crossing,
drawing in the plane). However, Theorems 3 and 4 do not provide an easy
way of showing that a given graph is planar, since this would involve looking at
a large number of subgraphs and verifying that none of them is a subdivision
of K 5 or K 3,3 , or contains K 5 or K 3,3 as a contraction. For this reason, no
currently used algorithm for testing the planarity of a graph is based on either
of these theorems.

Let us consider an example.

Example 11: Show that K 5 is a contraction of the Petersen graph. Hence,

deduce that the Petersen graph is nonplanar.

Solution: Starting from the Petersen graph G, we construct a sequence

(G0 = G, G1 , G2 , G3 , G4 , G5 = K 5 ) as shown below.
u1 w1 w1

v1
u5 u2 u5 u2 u5 w2
v5 v2 v5 v2 v5

v 4 v3 v 4 v3 v4 v3

u4 u3 u4 u3 u4 u3
G = G0 G1 = G0 ⋅ u1v1 G 2 = G1 ⋅ u 2 v 2

w1 w1 w1

w5 w2 u5 w2 u5 w2
v5 v5

v4

w4 w3 w4 w3 u4 w3
G5 = G4 ⋅ u5 v5 G 4 = G3 ⋅ u 4 v 4 G3 = G 2 ⋅ u 3 v3

We have obtained G5 = K 5 . Therefore, K 5 is a contraction of the Petersen

graph. Consequently, by the Wagner’s Theorem, the Petersen graph is
nonplanar.
***

Example 12: Show that K 3, 3 is a contraction of the Grötzsch graph. Hence

conclude that the Grötzsch graph is nonplanar.

Solution: Given on the left below is the Grötzsch graph. If we remove the
edges v2 u3 , u3 v4 and u4 v5 we get the subgraph H as shown on the right

190 below.
Unit 7 Planar Graphs
u1 u1

v1 v1

u5 u2 u5 u2
v5 v2 v5 v2
w w
v4 v3 v4 v3

u4 u3 u4 u3
G H
Now we get K 3,3 as a contraction of H as follows.
u1 u1 u1

v1 w1 w2

u5 u2 u5 u2 u5 u2
v5 v2 v5 v2 v2

w
v4 v3 v4 v3 v4 v3

u4 u3 u4 u3 u4 u3
H = H0 H 1 = H 0 ⋅ w v1 H 2 = H 1 ⋅ w1 v5

u1 u1 u1

w3 u5 w3 w3 u2
w5 w4 w4 u5

v4 v3 v4 v3 v4 v3
u4 u4 u3
H 5 = H 4 ⋅ u 4 u5 H 4 = H 3 ⋅ u 2 u3 H 3 = H 2 ⋅ w2 v2

Now it is clear that H 5 = K 3, 3 . Thus H is a subgraph of G, which has K 3, 3 as

a contraction. Therefore, by Theorem 4, G is nonplanar.
***
Now, try following exercises.

E16) Let H be a subdivision of a graph G, and K be a subdivision of H .

Then K is a subdivision of G. True or false? Justify.

E17) Give a plane drawing of the graph given below, if it exists. Otherwise,
explain why no plane drawing exists.

191
Block 2 Graph Tracing and Coluring
E18) If H is a contraction of a graph G, then Δ ( H ) ≤ Δ (G). True or False?
Justify.
E19) Is the following graph planar? Justify your answer.

E20) Does there exist planar graph with 7 vertices and 12 edges such that
each of its faces is bounded by exactly 3 edges? Why?

There are many real life applications of planarity. One of the most important
applications is in Printed Circuit Board (PCB) layouts that we are going to
discuss in the next section.

7.5 PCB PROBLEM

In this section, we shall discuss about the Printed Circuit Board Layout
Problem, often shortened as PCB Problem. We shall also see how this
problem gives rise to the concepts of thickness and crossing.
In printed circuits, electronic components are connected by conducting strips
printed directly onto a flat board of insulating material. A sample PCB is shown
below.

Such printed connectors are not allowed to cross, since this would lead to
undesirable electrical contact at crossing points. Circuits in which many
crossings are unavoidable may be printed on several boards (layers) that are
then sandwiched together. Each board consists of a printed circuit without
crossings. Now, the question arises is – what is the smallest number of such
layers need for a given circuit?
We illustrate this problem with a particular example.
Consider a printed circuit that has 36 interconnections and is represented by
the complete graph K 9 . It is impossible to arrange all these interconnections in
one layer, or even two; three layers are needed, and a solution is given below.

3 4
2 5 2 3 4 5 2 3 4 5

1 9 8 7 6 9 8 7 6 1 9 8 7 6
1

192
layer 1 layer 2 layer 3
Unit 7 Planar Graphs
Note that each edge of K 9 is included on just one of the layers – for example,
the edge 28 appears on layer 2, and the edge 69 appears on layer 3.

Each of these three graphs is a planar graph. So the PCB problem, indeed,
seeks a decomposition of a graph into a minimum number of planar
subgraphs. In the case of K 9 , the three planar subgraphs corresponding to the
three layers shown above are induced by the following sets, respectively.
{12,13,16,18,19, 23, 29, 34, 38, 39, 45, 46, 47, 48, 56, 57, 67, 68, 78, 89},
{14,15,17, 24, 28, 35, 36, 37, 79},
{25, 26, 27, 49, 58, 59, 69}.

Let us consider one more example.

Example 13: Show that K 6 can be ‘printed’ in two layers and write down a
corresponding edge decomposition.

Solution: The graph K 6 can be printed in two layers, as follows.

1 2 1 2

6 3 6 3

5 4 5 4

layer 1 layer 2

The corresponding sets of edges are

{12,13,14,15,16, 23,24,25, 34, 45, 5 6}, {26, 35, 36, 46}.
***

Note that a graph may have many decompositions into planar subgraphs.
However here we are concerned with the minimum number of such subgraphs
in a decomposition. This leads us to define the thickness of a graph.

Definition: The thickness of a graph G is the minimum number of planar

subgraphs whose union is G and, the crossing number of graph G is the
minimum number of crossings in drawing of G in the plane.

We shall write θ (G ) for the thickness of G, and ν (G) for the crossing
number of G. Thickness and crossing number tell us how far a graph deviates
from planarity.

By definition, the thickness of a planar graph is 1 and crossing number of a

planar graph is zero. The thickness of each of the Kuratowski’s graphs K 5
and K 3,3 is, clearly, 2 and their crossing number is 1.

Example 14: Find the thickness of the Petersen graph.

Solution: Consider a drawing of the Petersen graph as given below.

193
Block 2 Graph Tracing and Coluring
a

j
b e
i f

h g

c d
G
We know that G is nonplanar. Therefore, θ (G) ≥ 2. Now look at the following
planar subgraphs G ′ and G ′′ whose union is G. That is, {G′, G′′} is a
decomposition of G into planar subgraphs. Thus θ (G) = 2.

j j
b e
i f
i f

h g h g

c d
G′ G ′′

***

In general, there is no known formula that gives the thickness of a graph.

However, we can easily obtain a lower bound for thickness that often
coincides with the correct value. This is given below.

Theorem 5: Let G be an (n, m) -graph, with n ≥ 3. Then

⎡ m ⎤
a) θ (G) ≥ ⎢ ,
⎢ 3n − 6 ⎥⎥
⎡ m ⎤
b) θ (G) ≥ ⎢ , if G has no triangles.
⎢ 2n − 4 ⎥⎥

Proof: a) If θ (G ) = k , then G has a minimum of k planar subgraphs whose

union is G. Each of these subgraphs can have at most 3n − 6 edges.
(See Corollary 1.) Thus k subgraphs can have at most k . (3n − 6)
m
edges altogether. That is, m ≤ k (3n − 6), which gives k ≥ .
3n − 6
⎡ m ⎤
However, θ (G ) is an integer, so θ (G ) ≥ ⎢ .
⎢ 3n − 6 ⎥⎥

b) The proof is similar to part (a). Use Corollary 2.

We can now deduce lower bounds for the thickness of K n and that of K r , s .
194
Unit 7 Planar Graphs
⎢n + 7⎥
Theorem 6: a) θ ( K n ) ≥ ⎢ for all n ≥ 1.
⎣ 6 ⎥⎦

⎡ mn ⎤
b) θ ( K m, n ) ≥ ⎢ , for all m, n ≥ 1.
⎢ 2m + 2n − 4 ⎥⎥

Proof: (a) The result holds if n = 1 or n = 2. So assume n ≥ 3. Then

1
m= n ( n − 1). It follows from part (a) of Theorem 5 that
2
⎡1 ⎤
⎢ 2 n(n − 1) ⎥
θ (Kn ) ≥ ⎢ ⎥
⎢ 3n − 6 ⎥
⎢ ⎥
⎢1 ⎥
⎢ 2 n(n − 1) + (3n − 6) − 1⎥ ⎛ ⎡ p ⎤ ⎢ p + q − 1⎥ ⎞
=⎢ ⎥ ⎜⎜Q ⎢ ⎥ = ⎢ ⎥ , ∀ p , q ∈ N. ⎟
3n − 6 q q ⎟
⎢ ⎥ ⎝ ⎢ ⎥ ⎣ ⎦ ⎠
⎣ ⎦
⎢ n 2 + 5n − 14 ⎥
=⎢ ⎥
⎣ 6 (n − 2) ⎦
⎢n + 7⎥
=⎢
⎣ 6 ⎥⎦

b) Try yourself.

So, to sum up, although we cannot solve the PCB problem in general, we
have obtained a lower bound for the solution, and this bound coincides with
the correct value surprisingly often.

Now try the following exercises.

E21) Find the thickness and crossing number of each of the following graphs:

a a
b
f b
e
e c
d c
d
(i) (ii)

a b a b
f g

f c
e h

e d d c

(iii) (iv)
[Hint: Check for planarity first by redrawing the graphs]. 195
Block 2 Graph Tracing and Coluring
⎢n + 5⎥
E22) Show that θ ( K n , n ) = ⎢ for all n ≥ 1.
⎣ 4 ⎥⎦

By now you must have gained some understanding of plane and planar
graphs. The next concept we are going to introduce is about the dual of a
plane graph.

7.6 DUALITY
We now introduce the idea of duality for plane graphs. To illustrate this idea,
we consider the graph of the cube Q3 . If we place a new vertex within each
face (including the unbounded one) and join the pairs of new vertices that lie in
adjacent faces, we obtain the graph of the octahedron as follows.

1
4 4
3 1 5
3 1 5
5 4
2 2 6
6 2 3
6
Q3 Octahedron

(The new vertices are represented by small circles, and the edges joining
them are indicated by dashed lines.) The graph we have obtained is the “dual”
of Q3 . A formal definition is here.

*
Definition: Let G be a plane graph. Then the dual of G, denoted by G is
the graph constructed as follows.

*
Draw a point in each face of G , and call then the vertices of G . For each
* *
edge e of G draw an edge e joining the vertices of G in the faces on either
*
side of e : these are the edges of G . (Note that if any edge e of G lies on the
* * *
boundary of exactly on face of G, then e is a loop of G . Thus G may be a
multigraph.)

*
For instance, look at the graph G and its dual G drawn below.

It turns out that the dual of a plane graph is a planar graph. Let us look at one
more example.

Example 15: Draw the dual of each of the following plane graphs.
196
Unit 7 Planar Graphs

G1 G2 G3

Solution: The duals of G1 , G2 and G3 are constructed as follows.

G1 G1*

G2 G2*

G3 G3*

***

Observe that in the example above we have G1 ≅ G1* and G3 ≅ G3* . Such
graphs are “self-dual” as defined below.

Definition: A plane graph G is called self-dual if G ≅ G .

*

You can clearly verify that the wheel graph Wn is self-dual for all n ≥ 4. The
next example shows that the duals of two isomorphic plane graphs need not
be isomorphic.

Example 16: Show that the following plane graphs are isomorphic, but their
duals are not.
197
Block 2 Graph Tracing and Coluring

G1 G2
Solution: Establishing an isomorphism between G1 and G2 is easier. (Just
find.) Let us draw the duals of G1 and G2 .

G1 G1*

G2*
G2

Now observe that G1* ≅/ G2* .

***

Now we shall show that the dual of a plane graph is always connected, no
matter whether the original graph is connected or not.
*
Theorem 7: For every plane graph G, its dual G is connected.
*
Proof: Let G be any plane graph. Pick any two vertices u and v in G . Then
there is a curve in the plane that joins u and v, which does not pass through
any of the vertices of G. This curve, of course, passes through the faces and
edges of G, alternately. But the faces of G correspond to the vertices of G * ,
*
and the edges of G correspond to the edges of G . Thus, we get an
*
alternating sequence of vertices and edges of G with endpoints u and v,
* *
which is, indeed, a (u, v) -walk in G . Therefore, G is connected.

Now consider the following results.

Proposition: Let G be a connected plane graph. Then the number of faces of

G * is equal to the number of vertices of G.
*
Proof: Assume that G has n vertices, m edges and r faces. Also let G
* * *
have n vertices, m edges and r * faces. Since both G and G are
connected, by the Euler’s Formula, n − m + r = 2 and n − m + r = 2.
* * *
198
Unit 7 Planar Graphs
By definition, we have n = r and m = m. Substituting these values in the
* *

above equations, we get r = n.

*

Finally we state the following result, without proof.

Theorem 8: If G is a connected plane graph, then (G * )* ≅ G.

Now try the following exercises.

E23) Find dual of the following graphs:

(i) (ii) (iii)

*
E24) If a plane graph G has a cut-vertex, then G also has a cut-vertex. True
or false? Justify.

E25) Without actually drawing G * , find the number of vertices, edges and
faces of G * , where G is the multigraph given below:

E26) Under what conditions on a plane graph, is its dual a graph?

Next, we shall discuss the colouring of plane graphs.

7.7 MAP COLOURING PROBLEM

Recall that in Unit 6 we have discussed the concept of graph colouring,
namely the vertex-colouring and edge-colouring. In case of plane graph, there
is yet another way to colour a graph! It is the face-colouring. We shall define
what face-colouring of a plane graph means, and how it is related to the map
colouring problem.

Next, we shall discuss the Map Coloring Problem. We will show that this can
be reduced to coloring of planar graphs. We shall also show that any planar
graph can be vertex coloured with at must five colours.

Indeed, any planar graph can be vertex-coloured with just 4 colours. However,
this remained a problem (known as four Colour Problem) for a long time, until 199
Block 2 Graph Tracing and Coluring
1979 when Appel and Haken gave an algorithmic proof. They took nearly 1200
hours of computer time on some fastest available computers of that time to
eventually conclude that no more than 4 colours were needed.

Let us begin with the following definition.

Definition: Let G be a plane graph. A k -face-colouring of G is an

assignment of k colours to the faces of G in such a way that no two faces
with a common boundary get the same colour. If there exists a k -face
colouring for G , then G is said to be k -face-colourable.
For instance look at the following graph G.

f2 f3
f5
f1 f4

There are five faces in G , namely f1 , f 2 , f 3 , f 4 and f 5 , where f 5 is

unbounded. The face f1 shares boundary with f 2 as well as f 5 . Thus f1
must get a colour different from those of f 2 and f 5 . A 3-face-colouring of G
is given below.

f2 f3
2 1 f5
f1 f4 3
1 2

A 3‐face‐colouring of G.

Now let us look an important relationship between the face-colouring and

vertex-colouring.

Theorem 9: Let G be a connected plane graph with no cut-edges. Then G is

k -face-colourable iff G * is k -vertex-colourable.

Proof: Assume that G is k -face-colourable. This means, there is a face-

colouring of G that assigns k colours to the faces of G. Recall that the faces
of G corresponds to the vertices of G * . Moreover, two faces of G share a
boundary edge iff the corresponding vertices of G * are adjacent. Since any
two faces with a common boundary edge get distinct colours, any two adjacent
vertices in G * get two distinct colours. Thus G * has a k -vertex-colouring.
Now assume that G * has a k -vertex-colouring, and reverse the arguments
used above to conclude that G has a k -face-colouring.

Now we define what we mean by a map.

Definition: A map is a connected plane graph with no cut-edges.

Thus in the language of maps, the statements of Theorem 10 is as follows:

200
Unit 7 Planar Graphs
*
“A map G is k -face-colourable iff G is k -vertex-colourable.”

This means the face-colouring of a map is equivalent to the vertex colouring of

its dual.

Now, we present the statement of Four-Colour Theorem.

Theorem 10 (Four-Colour Theorem): Every planar graph can be coloured

with 4 colours. In other words, if G is a planar graph, then χ (G ) ≤ 4.

7.8 SUMMARY
In this unit, we have covered the following points:

1. We have described what are plane graph, and planar graph.

2. Any plane drawing of a planar graph G divides the set of points of the
plane not lying on G into regions, called faces.

3. Graphs K 5 and K 3,3 are two special non-planar graphs, called

Kuratowski’s graphs.

4. We have seen the Euler’s Formula that relates the number of vertices,
edges and faces of any plane drawing of a connected planar graph.

5. Kuratowski’s Theorem: A graph is planar if and only if it contains no

subdivision of K 5 and K 3,3 .

6. Wanger’s Theorem: A graph is a planar graph if and only if it contains

no subgraph that has K 5 or K 3,3 as a contraction.

7. We have seen the definitions of thickness and crossing number of a

graph

8. We have discussed the PCB layout problem.

9. We have defined the concept of the dual of a plane graph, and finally
gave an overview of the Map Colouring Problem.

7.9 SOLUTIONS/ANSWERS
E1) The statement is false. For a counter-example consider the plane graph
G and its face f given below.

v1 v2

v4 v3

G 201
Block 2 Graph Tracing and Coluring
Let S = {v1 , v2 , v3 , v4 }. Then S induces G, which is not a subgraph of
the boundary of f .

E2) Let G be a planar graph, and H be a component of G. Then H is a

subgraph of G. Now by Example 3, H is planar. Therefore, every
component of G is planar.

Conversely, assume that a graph G has k components each of them

being planar. Then we can draw k mutually disjoint disks in the plane.
Now draw, one by one, plane embeddings of these components inside
these disks one embedding per disk. This way we get a plane
embedding of G. Hence, G is planar.

E3) For n ≥ 3, we can draw the vertices of C n on a circle, and add edges
joining the consecutive vertices. This gives us a plane embedding of Cn .

E4) Try yourself. (See Example 2.)

E5) We know that every proper subgraph of K 5 is a subgraph of one of the

following graphs.

K5 − e K4
Since both K 5 − e and K 4 are planar, every proper subgraph of K 5 is
planar.

E6) True. This is the contrapositive of the statement given in Example 3.

E7) Consider two planar graphs G and H sharing exactly one vertex, say
v. Draw two disks meeting exactly at one point in the plane. Then place
the vertex v at the point of contact of the disks. Now draw the remaining
vertices of a plane drawing of G inside one disk, and of H inside
another disk. This gives us a plane drawing of G ∪ H , and hence
G ∪ H is planar.

E8) We shall prove it by using the Principle of Mathematical induction on the

order of a tree. It is clear that a tree of order 1 is planar. A tree of order 2
is also planar. Now assume that all trees of order n are planar for some
n ≥ 2. Consider a tree T with order n + 1 . Let v be a leaf of T , and u
be the neighbour of v in T . Then T − v is a tree with order n.
Therefore, by the induction hypothesis, T − v is planar. Now T can be
obtained by taking the union of T − v with the edge uv. So, by E7, T is
planar. Therefore, by the Principle of Mathematical induction, all trees
are planar.

E9) a) Try yourself.

202
Unit 7 Planar Graphs
b) The Petersen graph has 10 vertices and 15 edges. If it is planar,
5
then by part a) we must have 15 ≤ (10 − 2). But this inequality is
3
false. Therefore, the Petersen graph is nonplanar.

E10) Q3 is planar because the following is a plane drawing of Q3 .

Recall that Q4 has 2 = 16 vertices, and 4 × 2 = 32 edges. We also

4 3

know that Q4 is bipartite, and hence has no triangles. Now if Q4 is

planar, then by Corollary 2, 32 ≤ 2 × 16 − 4. but this inequality does not
hold. Hnece Q4 is nonplanar.

E11) False. See E5.

E12) a) Look at the following graph.

v1

v5 u1 v2
u5 u2

u 4 u3

v4 v3

Each vertex in this graph has degree 4. Now add two vertices x
and y to this graph in such a way that x is the neighbour of
u1 , u 2 , u3 , u 4 and u5 ; and y is a neighbour of v1 , v2 , v3 , v4 and v5 .
This gives us a desired planar graph.

b) No such graph exists due to Corollary 3.

E13) One such graph is given below.

E14) Let ni , mi , ri denote the number of vertices, number of edges, and the
number of faces, respectively in the ith component of G. Note that one
face is common in all the components. Therefore,
k
r = ∑ ri − (k − 1).
i =1
203
Block 2 Graph Tracing and Coluring
Now applying the Euler’s Formula for each component we get
ni − mi + ri = 2, i.e., ri = mi − ni + 2 for all i = 1, 2, K, k. Therefore,
k
r = ∑ (mi − ni + 2) − ( k − 1)
i =1

= m − n + 2 k − k + 1 = m − n + k + 1.

This gives us n − m + r = k + 1.

E15) A 4-regular planar graph on 6 vertices is drawn below.

We know that 3n − 6 is the maximum number of edges in a planar graph

on n vertices. Here n = 6 and m = 12 . So m = 3n − 6. Therefore, this
graph is maximal planar.

E16) Note that every edge uv of G either belongs to H or transforms into a

(u, v) -path in H . Similarly, every edge xy of H either belongs to K , or
transforms into an ( x, y ) -path in K . Thus every edge uv of G either
belongs to K , or transforms into a (u, v) -path in K . Therefore, K is a
subdivision of G.

E17) A plane drawing is given below.

E18) False. A counter-example is given below.

u v w

G H = G ⋅ uv

Here H is a contraction of G. But Δ( H ) ≤| Δ(G).

E19) Observe that the given graph can be contracted into the follow graph.
204
Unit 7 Planar Graphs

This is K 3, 3 . Hence the given graph is not planar.

E20) Note that if G is a planar graph with n = 7 vertices, and m = 12 edges,

then any plane drawing of G has r = m − n + 2 = 12 − 7 + 2 = 7 faces.
Now each face is bounded by exactly 3 edges, which means that each
face has degree 3. Then by the Handshaking Lemma for Planar Graphs,
we have 3 × 7 = 2 × 12, which is not true.
Therefore, no such graph exists.

E21) i) The given graph can be redrawn as follows.

a b

d c

So, the graph is planar. Hence its thickness is 1, and the crossing
number is 0.

ii) The thickness and the crossing numbers of this graph are also 1
and 0, respectively, as it has a following plane drawing.

a d

b
e c f

iii) Note that the given graph G (say) contains the following
subgraph.
a b

f c

e d 205
Block 2 Graph Tracing and Coluring
This is isomorphic to K 3, 3 . Thus G is nonplanar, and hence has
thickness at least 2. Two planar subgraphs of G are drawn below.

a b
a

f c

e d d

G1 G2

Therefore, θ (G) = 2.

In order to find ν (G), note that ν (G) ≥ 1. A drawing of G with

one edge-crossing is given below.
a b

f c

e d

Consequently, ν (G) = 1.

iv) We can contract the given graph into K 3, 3 as follows.

a b a b x d
f g g g

e
h e h h

d c d c y c
G0 = G G1 = G0 ⋅ af G2 = G1 ⋅ de

Observe the G2 ≅ K 3, 3 . Thus G contains K 3, 3 as a contraction.

Therefore, by the Wagner’s Theorem G is nonplanar. This implies
θ (G) ≥ 2. Two planar subgraphs forming a decomposition of G
are given below.

206
Unit 7 Planar Graphs
a b g
f g

e h c
d c
H1 H2
Consequently, θ (G) = 2.

Since G is nonplanar, ν (G) ≥ 1. The given drawing of G has one

edge-crossing. Therefore, ν (G) = 1.

E22) From Theorem 6(b), we get

⎡ n2 ⎤
θ ( K n, n ) = ⎢ ⎥
⎢ 4n − 4 ⎥
⎢ n 2 + 4n − 4 − 1⎥
=⎢ ⎥
⎣ 4n − 4 ⎦
⎢ (n − 1) (n + 5) ⎥
=⎢ ⎥
⎣ 4 (n − 1) ⎦
⎢n + 5⎥
=⎢ .
⎣ 4 ⎥⎦

E23)

G G*

G G*

207
G G*
Block 2 Graph Tracing and Coluring
E24) False. For instance, look at the graph G and its dual G * drawn below.

G G*

Here G has a cut-vertex, but G * does not have any cut-vertex.

E25) The multigraph G has n = 5 vertices, m = 12 edges and r = 9 faces.

Therefore, G * has n * = r = 9 vertices, m * = m = 12 edges and
r * = n = 5 faces.

E26) Observe that a cut-edge in G corresponds to a loop in G * . Also, if any

two faces of G share more than one boundary edges, then the
corresponding vertices in G * will be the endpoints of parallel edges.
Thus for G * to be a graph, G must have not share more than one
boundary edge.

208