Progressive interpretations of scientific theories have long been denounced as naive, because of the inescapability of meaning variance. The charge reportedly applies to recent realist moves that focus on theory-parts rather than whole theories. This paper considers the question of what "theory-parts" of epistemic significance (if any) relevantly accumulate through theory-change and identifies a form of theory-parts seemingly suitable for the realist agenda. My starting point is a notion of cumulative theoretical description lifted from instances of inter-theoretic dialogue in scientific practice.