Volumetrically significant melt production requires crustal temperatures above approximately 800• C. At the grain scale, the former presence of melt may be inferred based on various microstructures, particularly pseudomorphs of melt pores and grainboundary melt films. In residual migmatites and granulites, evidence of melt-extraction pathways at outcrop scale is recorded by crystallized products of melt (leucosome) and residual material from which melt has drained (melanosome). These features form networks or arrays that potentially demonstrate the temporal and spatial relations between deformation and melting. As melt volume increases at sites of initial melting, the feedback between deformation and melting creates a dynamic rheological environment owing to localization and strain-rate weakening. With increasing temperature, melt volume increases to the melt connectivity transition, in the range of 2-7 vol% melt, at which point melt may escape in the first of several melt-loss events, where each event represents a batch of melt that left the source and ascended higher in the crust. Each contributing process has characteristic length and time scales, and it is the nonlinear interactions and feedback relations among them that give rise to the dissipative structures and episodicity of melt-extraction events that are recorded as variations in the spatial and temporal patterning of the crust. Focused melt flow occurs by dilatant shear failure of low-melt fraction rocks creating melt-flow networks that allow accumulation and storage of melt, and form the link for melt flow from grain boundaries to veins allowing drainage to crustal-scale ascent conduits. Preliminary indications suggest that anatectic systems are strongly self-organized from the bottom up, becoming more ordered by decreasing the number and increasing the width of ascent conduits from the anatectic zone through the overlying subsolidus crust to the ductile-to-brittle transition zone, where the melt accumulates in plutons.