Aubrite meteorites are composed of constituents which are almost certainly of igneous origin. If they were generated by the melting and fractionation of enstatite chondrite-like parental material, as seems very likely, then plagioclase-rich, basaltic complements to the aubrites should have formed. However, such materials are not known as individual meteorites, and the compositions of two plagioclase-silica clasts and one albite-silica-(diopside-anorthite) clast (probably an impact melt) in the Norton County aubrite breccia suggest that they are not the putative enstatite-plagioclase basalts. We propose a new mechanism that explains the absence of these materials, showing that the expansion of even very small amounts of volatiles present in a melt approaching the surface of a small, low-gravity body will be enough to disrupt the melt into a spray of droplets moving faster than the local escape velocity. This explosive volcanic process of melt removal requires larger melt volatile contents on larger bodies, and data on the solubility of volatiles in basaltic melts suggest that the process was limited to bodies smaller than about 100 km in radius.