Demixing of polymer blends via spinodal decomposition provides an effective means to generate polymer bicontinuous nanostructures with applications in membranes for separations to photovoltaics. Predicting demixed phase sizes can be achieved by the well-known Cahn-Hilliard (CHC) theory, which is however expected to break down approaching molecular dimensions. Motivated by both fundamental science and the precise design of nanomaterials, we will experimentally examine this limit (<<50 nm phases) with a combination of one- and two-phase SANS experiments for a model system (PMMAd/SCPE) with rapidly varying interactions with temperature around Ts. This proposal will this examine the limits of polymer demixing based on what is arguably one of the most significant condensed matter theories of the 20th century.