The aim of the project is to build an integrated online environment for collating and editing modernist texts that exist in multiple versions. The task here is much more than simply to allow comparison of texts to identify where variations occur. Instead, it is to facilitate new interpretive possibilities. That is, out of the collation of text A and text B we will generate a set of variants, that can stand as text C. We want to read text C to see what it can tell us about the evolution of a given novel, say, and then to link those changes into the social, political, economic, and cultural contexts that produce them. The theoretical issues that arise concern what counts as a text and how we can read the third possible text — the C text — that emerges out of the incompatibility between text A and text B. The aim is to read the gap between one version and the next, following Derrida’s imperative to attend to difference itself as the source of meaning rather than presuming that it is a merely negative product of the non-coincidence of two entities: text A and text B. Given modernism’s own fascination with radical breaks, discontinuities, and making things new, the theory behind our approach to textual variation is in many ways deeply modernist itself. Following the lead of such writers as Conrad, Forster, and Joyce, and the ways in which their innovations in narrative informed much of later twentieth-century theory, the Modernist Versions Project concerns itself with what gaps, silences, and difference itself can tell us. It does not pursue the holistic text of the genetic edition, nor the definitive edition of the “corrected text,” but rather the trace of what has been erased, of erasure itself, as the most productive point of meaning. In doing so, we hope to be able to restore to our understanding of modernism a key element of its production, an aspect which modernism itself elided, but which remains central to understanding it as fully as possible.