Abstract-According to a currently popular model for petrogenesis on the howardite, eucrite, and diogenite (HED) parent asteroid, the diogenites are not comagmatic with most eucrites but instead formed in separate orthopyroxenite-dominated plutons. This model can be tested for consistency with mass balance for MgO and FeO, assuming the overall diogenite/(diogenite + eucrite) ratio, d, of the parent asteroid is at least comparable to the average d for the eucrite + diogenite dominated howardite regolith breccias. Average mg# (=MgO/ [MgO + FeO]) is much lower for eucrites, especially noncumulate eucrites, than for diogenites. Unless the diogenite parent magmas eventually produced a large proportion of low-mg# residual basalt and gabbro (RBG), the implied initial magma's mg# is vastly higher than that of any noncumulate eucrite. Starting from a source previously depleted by putative primary eucrite genesis, melt mg# can be estimated as a finction of the exchange reaction KD and degree of melting. Using several very conservative assumptions (e.g., assuming that the total [MgO + FeO] concentration is nearly the same in the nascent melt as in the residual solids), the degree of melting required to yield a melt with mg# high enough to satisfy mass balance, without implying an RBG component that accounts for >50% of all eucrites, is an implausibly high 60-80 wt?!.The separate orthopyroxenitic plutons (SOP) model also seems inconsistent with the uniform density of melts across the diogenite-eucrite compositional spectrum (2.77-2.82 g/cm3), which implies that diogenitic magmas should have been as capable as eucrites of extruding to form lavas. This difficulty cannot be reduced by simply assuming that later-formed magmas were systematically both more plutonic and more MgOrich than earlier ones, because the plutonic cumulate eucrites equilibrated with melts systematically lower in mg# than noncumulate eucrites. Conceivably, the bulk mg# of the asteroid's silicate system was increased between primary-melt eucrite genesis and SOP diogenite genesis by graphite-fieled reduction of FeO. However, the graphite oxidation process generates a huge proportion of gas, which would have enhanced the buoyancy of the nascent diogenite-parent magmas, thus exacerbating the difficulty of achieving the implied high degrees of partial melting.To avoid these difficulties but still form most eucrites as rapidly cooled extrusives, I propose the NERD (noncumulate eucrites as extruded residua of diogenites) model. In this model, the diogenites form as early cumulates from a large magma system (probably a global "magma ocean") that yields a large proportion of eucritic melt as residuum. This residual melt zone undergoes relatively little crystallization during a period when it is episodically tapped to produce extrusions, dikes and sills of rapidly cooled noncumulate eucrites. Slight (-5-10%) porosity in the nascent eucritic crust keeps it marginally buoyant over the residual melt zone. The common thermal metamorphism of noncumulate eucrites results f...