This book explores the view that impoverishment and Agree operations are part of a single grammatical component. The architecture set forth here gives rise to complex but highly systematic interactions between the two operations. This interaction is shown to provide a unified and general account of apparently diverse and unrelated instances of eccentric argument encoding that so far have remained elusive to a unified theoretical account. The proposed view of the grammatical architecture achieves an integration of these phenomena within better-studied languages and thus gives rise to a more general theory of case and agreement phenomena. The empirical evidence on the basis of which the proposal is developed draws from a wide range of typologically non-related languages, including Basque, Hindi, Icelandic, Itelmen, Marathi, Nez Perce, Niuean, Punjabi, Sahaptin, Selayarese, Yukaghir, and Yurok . The proposal has far-reaching consequences for the study of grammatical architecture, linguistic interfaces, derivational locality in apparently non-local dependencies and the role of functional considerations in formal approaches to the human language faculty.