The technique of hyper-programming allows program representations held in a persistent store to contain embedded links to persistent code and data. In 1994, Connor et al proposed extending this to hyper-code, in which program source, executable code and data are all represented to the user in exactly the same form. Here we explore the concept of hyper-code in greater detail and present a set of abstract language-independent operations on which various concrete systems can be based. These operations (explode, implode, evaluate, root and edit) are provided by a single user interface tool that subsumes the functions of both an object browser and a program editor. We then describe a particular implementation using PJama (persistent Java) and examine the impact of several language features on the resulting system.