Thanks to its magnificent exposures and extraordinary petrologic development, the Skaergaard Intrusion provides an excellent test for three basic mechanisms of crystal-liquid fractionation: diffusive exchange, compaction, convective exchange. (1) Diffusive exchange at the solidification front does not seem to have played a major role, for no obvious difference is seen in the behavior of components of widely differing diffusivities. The effects of diffusion were probably masked by porous flow of the interstitial liquid. (2) Compaction and 'filter pressing' of residual liquids can be tested by comparing concentrations of incompatible elements in parts of the intrusion that crystallized in different orientations with respect to gravity. The contrast between the large concentrations of these elements in the Upper Border Series and consistently smaller abundances in the Layered Series, long thought to ‘be due to contamination of the upper levels with crustal rocks, is more likely a result of compaction that took place at the floor but not under the roof. (3) Convective exchange driven by differences in the compositional densities of the main magma and interstitial liquids seems to have been important at the steep walls and, to a lesser degree, under the roof. It may also have been responsible for a large compositional anomaly near the centre of the Layered Series where an iron-rich liquid percolated down through parts of Lower and Middle Zones. At later stages of crystallization when density relations were reversed, a buoyant liquid, rich in volatiles and incompatible elements, rose from the Layered Series and permeated the upper part of the intrusion. A liquid of this latter kind could have caused the pervasive metasomatic alteration seen in much of the Layered Series. In both compaction and convective exchange, re-equilibration and reactions between migrating liquids and early-formed crystals augmented permeabilities and brought about extensive compositional and textural changes.
Although inhomogeneities in the initial magma could account for some of the compositional variations, the main trend of differentiation can be ascribed to a combination of compaction and convective fractionation. The convective flow was of two kinds, one of descending iron-rich liquids and another of more evolved rising liquids (and fluids) with low densities and large concentrations of incompatible elements. In the early stages, the proportions of iron-rich liquids were much greater than those of the silica-rich variety, and until the crystallization fronts converged at the Sandwich Horizon, the former made the greatest contribution to evolution of the main body of magma. As crystallization advanced, however, light, silica-rich liquids became more important; they permeated the Upper Border Series and much of the upper part of the Layered Series. This late-stage liquid continued to differentiate after the crystallization fronts met; a secondary Sandwich Horizon was formed where incompatible elements reach maximum concentrations in the lower part of the Upper Border Series.