THE GENERATION OF KIMBERLITES, LAMPROITES, AND THEIR SOURCE ROCKS

Measurements of rare earth element (REE) concentrations in South African kimberlites and in the Argyle lamproite from Western Australia constrain the composition of the source rocks from which these melts originate. To account for the concentrations of Tm, Yb, and Lu in these magmas, their sources must first have been strongly depleted by approximately 20% melting in the garnet stability field, and then enriched by a metasomatic melt rich in light REE and other incompatible elements. The calculated source compositions closely resemble those of coarse, low-temperature, depleted peridotite nodules that are the commonest nodules in kimberlites. Rarer nodule types have undergone either more or less depletion than have the source regions of kimberlites and lamproites. The REE composition of the metasomatic melt calculated from the diopsides and garnets in the sheared nodules, from the diopside megacrysts, and from majorite garnet inclusions in diamonds, is in excellent agreement with that expected for a melt produced by melting approximately 0.5% of the source region of ocean ridge basalts. The initial depletion event requires the extraction of approximately O.5% melt from a region in which garnet and chrome-spinel were stable. The melt distribution obtained from inversion of komatiite compositions satisfies both these conditions. Kimberlite source rocks are shallower than the layer from which fertile nodules originate. Such nodules must therefore be transported by entrainment of the lower boundary of the layer that becomes unstable. This proposal can account for their strong fabric.