Aggregates of clinopyroxene in some doleritic lavas have external outlines strongly reminiscent of garnet crystals. They cannot be satisfactorily identified either as Upper Mantle fragments or as cognate synneusis aggregates, and it is suggested that they may have developed by inversion from original garnets, which crystallized at very high pressures. Neither morphology nor composition can provide conclusive proof of their original nature, but there is circumstantial evidence from the presence of an unusual inclusion suite in basalts nearby. The commonest phases are zircon and corundum, minerals more usually found in kimberlites than in basalts. Corundum in kimberlite inclusions is almost invariably accompanied by calcium-rich garnet, which adds some significance to the suggestion that these clinopyroxene aggregates are pseudomorphing original garnet.