This thesis studies the circular chromatic number of Kneser graphs. It was known that if m is greater than 2n^{2}(n-1), then the Kneser graph KG(m,n) has its circular chromatic number equal its chromatic number . In particular, if n = 3, then KG(m,3) has its circular chromatic number equal its chromatic number when m is greater than 36. In this thesis, we improve this result by proving that if m is greaer than 24, then chi_c(KG(m,3)) = chi(KG(m,3)).

This thesis was originally motivated by considering vector space analogues of problems in extremal set theory, but our main results concern colouring a graph that is intimately related to these vector space analogues. The vertices of the <em>q</em>-Kneser graph are the <em>k</em>-dimensional subspaces of a vector space of dimension <em>v</em> over F_<em>q</em>, and two <em>k</em>-subspaces are adjacent if they have trivial intersection. The new results in this thesis involve colouring the <em>q</em>-Kneser graph when <em>k</em>=2. There are two cases. When <em>k</em>=2 and <em>v</em>=4, the chromatic number is <em>q</em>²+<em>q</em>. If <em>k</em>=2 and <em>v</em>>4, the chromatic number is (<em>q</em>^(v-1)-1)/(<em>q</em>-1). In both cases, we characterise the minimal colourings. We develop some theory for colouring the <em>q</em>-Kneser graph in general.

This thesis was originally motivated by considering vector space analogues of problems in extremal set theory, but our main results concern colouring a graph that is intimately related to these vector space analogues. The vertices of the <em>q</em>-Kneser graph are the <em>k</em>-dimensional subspaces of a vector space of dimension <em>v</em> over F_<em>q</em>, and two <em>k</em>-subspaces are adjacent if they have trivial intersection. The new results in this thesis involve colouring the <em>q</em>-Kneser graph when <em>k</em>=2. There are two cases. When <em>k</em>=2 and <em>v</em>=4, the chromatic number is <em>q</em>²+<em>q</em>. If <em>k</em>=2 and <em>v</em>>4, the chromatic number is (<em>q</em>^(v-1)-1)/(<em>q</em>-1). In both cases, we characterise the minimal colourings. We develop some theory for colouring the <em>q</em>-Kneser graph in general.

This thesie discusses uniquely circular colourable and uniquely fractional colourable graphs. Suppose G = (V;E) is a graph and r &#x00B8; 2 is a real number. A circular r-colouring of G is a mapping f : V (G) ! [0; r) such that for any edge xy of G, 1 ¡P jf(x) &#x00A1; f(y)j ¡P r &#x00A1; 1. We say G is uniquely circular r-colourable if there is a circular r-colouring f of G and any other circular r-colouring of G can obtained from f by a rotation or a ¢Xip of the colours. Let I(G) denote the family of independent sets of G. A fractional r-colouring of G is a mapping f : I(G) ! [0; 1] such that for any vertex x, Px2I f(I) = 1 and PI2I(G) f(I) ¡P r. A graph G is called uniquely fractional r-colourable if there is exactly one fractional r-colouring of G. Uniquely circular r-colourable graphs have been studied extensively in the literature. In particular, it is known that for any r &#x00B8; 2, for any integer g, there is a uniquely circular r- colourable graph of girth at least g. Uniquely fractional r-colouring of graphs is a new concept. In this thesis, we prove that for any r &#x00B8; 2 for any integer g, there is a uniquely fractional r-colourable graph of girth at least g. It is well-known that for any graph G, &#x00C2;f (G) ¡P &#x00C2;c(G). We prove that for any rational numbers r &#x00B8; r0 > 2 and any integer g, there is a graph G of girth at least g, which is uniquely circular r-colourable and at the same time uniquely fractional r0-colourable.