This thesis works from the assumption that there is no privileged core-text of the Prelude, but instead a series of texts (representing more or less settled textual states) of to some extent uncertain historical relations and which are all of independent interest. It claims that the early (1798-99) and later (1804-05) texts encourage different kinds of readings, and examines the unsettling effect of such a polymorphous poem on current editorial thinking. The two-part Prelude of 1798-99 is considered as a composite and discontinuous text, inassimilable to any system of generic conventions, which filters so many styles and resonances of earlier writers that it is difficult to isolate within it a finite Wordsworthian idiom. The composition of 1804-05 is seen as redefining and restructuring the poem within the context of the epic tradition, and it is argued that in its later states the poem offers to be read alongside other works in that tradition -particularly Milton's Paradise Lost. A discussion of modern theories of poetic influence is prefixed to a full analysis of the semantic tension between these two works. Two ways of reading between poems (which demarcate both individual 'allusions' and habitual usage) are posited, and put on trial in actual interpretation. Complexities surrounding the projected five-Hook Prelude (which is proven to be irrecoverable) are taken as exemplary of the inappositeness of attempts to rationalise the poem's history in the interests of either a postulated 'final form' or a series of 'versions.' The methods and assumptions of Anglo-American textual criticism are re-evaluated and suggestions made as to how they might be democratised.