This study provides an investigation and discussion into the construction of a graph toolkit using generative programming techniques. It defines and analyses directed graphs, representations and identifies different techniques that may be used to discover new representations. Each representation is classified according to a unique classification system. Doing this enables us to uniquely identify a particular type of graph representation. A naming convention is used when identifying each graph representation which is a direct by product of the classification system. Details of how to implement the graph toolkit is presented and analysed. The toolkit is discussed and critically analysed with other major toolkits currently available today such as the Boost and Leda graph toolkits. Copyright / Dissertation (MSc)--University of Pretoria, 2010. / Computer Science / unrestricted