If G is a graph such that $\Delta(G) = 3$ then $\chi'(G) \le 4$

You don't need to worry about odd cycles, because they can be colored with three colors and you are allowed four.

You need to justify three things:

  1. Show that the maximum possible degree is $4$ (not so bad, I think).
  2. Show that thine graph cannot be the complete graph on five vertices, i.e. that you will not get a clique (you can check by going backwards if you like).
  3. You don't care if you get an odd cycle.

The you can use Brook's theorem in all its glory.

For further information, this result is called Vizing's Theorem.


Suppose $\Delta(G) = 3$. We need to show that 4 colors suffice to color the edges of the graph.

Let $uv$ be an edge in $G$. How many edges share a common endpoint with this edge? Since $\Delta=3$, at most two other edges besides $uv$ are incident to the vertex $u$, and similarly for $v$. So, at most $2+2=4$ edges share a common endvertex with edge $uv$. Thus, in the line graph of $G$, the vertex $uv$ has degree at most 4. If this line graph $L(G)$ is not an odd cycle or a complete graph, by Brook's theorem its vertices (and hence the edges of $G$) can be 4-colored.

If the line graph is an odd-cycle, its vertices can be 2- or 3- colored, hence 4 colors suffice. If the line graph is the complete graph $K_r$, then consider the value of $r$. If $r=3$, this $K_3$ is the line graph of either $K_3$ or a star $K_{1,3}$, and in either case 3 colors suffice. If $r \ge 4$, then this $K_r$ arose from a $G$ that is necessarily a star $K_{1,r}$. (This is a basic result on line graphs: any clique in $L(G)$ containing 4 or more vertices is induced by a star in $G$.) Since $\Delta(G) =3$, this star has exactly 3 edges; thus $r=3$ and three colors suffice again.

In other words, there is no clique of size 4 or more in the line graph $L(G)$ if $\Delta(G)=3$. More generally, all cliques in $L(G)$ arise from either stars $K_{1,r}$ in $G$ (i.e. the set of edges of $G$ that share a common endvertex) or triangles in $G$.