Because primitive basaltic magmas travel from their mantle sources to the surface without significant cooling and crystallization en route, compositional variations within them provide direct information on the sources and processes of partial melting in the mantle. Most studies of basalt magmagenesis have focused on chemical variations in magmas erupted from multiple vents in regional complexes or on long timescale (105--106years) variations in magmas from single shield volcanoes; few studies have aimed to measure changes in melt chemistry with time in individual eruptions. Available observations of individual basaltic eruption sequences, however, show remarkably similar temporal-compositional trends in the form of decreasing incompatible elements and MgO, with increasing SiO2, as the eruption (or sequence of related, closely-spaced eruptions) continued. Both trace element and isotopic variations indicate that this does not reflect fractional crystallization nor can it be explained simply by varying extents of partial melting of a single source. Instead, these trends are consistent with systematically changing proportions of mixing between melts produced by different degrees of partial melting of two compositionally distinct but apparently common sources in the mantle. One potential explanation for this is that a relatively deep mantle source melts to large degrees, producing melts that ascend through and mix with small-degree melts of the lithosphere. The small-degree lithospheric melts are then progressively exhausted during melt migration and eruption, so that late-erupting melts contain the smallest proportion of the shallow, small-degree melt. This cannot explain Os isotopic correlations with melt composition and radiogenic Os isotope signatures of the late-erupting melts in several eruption sequences, however. An alternative scenario is that systematic changes in erupted melt chemistry with time reflect sequential eruption of melt from increasing depths in mantle melting regions that are compositionally zoned because of different solidii and melt productivities of distinct peridotite and pyroxenite lithologic domains. In either case, compositional variation in many eruptions of primitive, intraplate basalts is dominated by mixing of melts from distinct lithologies, and temporal-compositional changes in such eruptions reflect the distribution and composition of melt with depth in mantle melting and melt migration regions.