The early Eocene Skaergaard intrusion of Greenland includes enormous numbers of rocks of both exotic and cognate origins. The lower parts of the Marginal Border Series contain abundant fragments of feldspathic peridotite that are possibly autoliths, intermixed with occasional xenoliths of Precambrian gneiss and metasomatized Cretaceous–Paleocene sediments derived from adjoining country rocks. The Upper Border Series includes one exceptionally large block of gneiss (several hundred meters across), and numerous smaller fragments, these originating from the intrusion's footwalls, plus a few pieces of peridotite. The Layered Series contains countless autoliths of troctolite, gabbroic anorthosite, and oxide (magnetite-ilmenite) gabbro, broken from parts of the Upper Border Series that have otherwise been lost to erosion; at the upper midlevel of its western half, it contains a few xenoliths of basalt, derived probably from the now-eroded (Eocene) roof of the intrusion. A distinctive postintrusion composite basaltic dike at one place contains 40 or more xenoliths of troctolite, olivine gabbro, and gabbroic anorthosite that may represent parts of the Layered Series still hidden at depth.
The Layered Series autoliths range from fragments a few centimeters on a side to blocks more than 400 m across, and they typically are coarser grained than their host cumulates, being in this respect more like Upper Border Series rocks. The autoliths are spread stratigraphically through the lower 70% of the exposed 2500 m thickness of the Layered Series and are generally concentrated in three broad stratigraphic zones. Their physical relationships to their host rocks—particularly the way they indent older layers beneath and are covered by younger layers above—provide abundant evidence that there was generally a sharp, well-defined interface between the top of the cumulate pile and the main body of magma in the intrusion while the Layered Series was forming. The distribution of the autoliths between and through the well-known, rhythmic, thin, modally graded layers shows that these layers were spread by magmatic currents; and their relations to the more extensive macrorhythmic layering suggest that it too was significantly shaped by currents.
Many of the larger autoliths are crudely layered internally, and in places it is evident that their stratification existed before they broke loose; therefore, it must have formed in the Upper Border Series. One particularly large block of oxide gabbro exhibits extraordinarily well-developed modal and textural layering and includes small troctolitic autoliths of an earlier generation, and it provides evidence that currents also spread crystalline materials across the top of the magma body. Many of the very small autoliths in the Layered Series are highly anorthositic in composition, apparently because they were leached of mafic minerals, and some of the larger blocks show local patchy internal replacement by anorthosite. Most large blocks show little sign of postaccumulation modification, and some have thin, fine-grained augite-rich rims or rinds, demonstrating that even though they were out of thermal and chemical equilibrium with their host cumulates, they still were effectively armored against extensive chemical change. Also documented is a large block that was cut by several early basaltic dikes before it broke free from the top of the intrusion; these early dikes transgress small anorthositic replacement pipes in the block, showing that the replacement process also occurred in the upper border environment.
Two mechanisms are described whereby graded cumulate layers can be sorted and deposited by magmatic crystal-liquid suspension currents. One, involving density surge currents, has been advocated previously; the other is a new concept based on boundary flow separation and reattachment vortex cells. The two mechanisms are used in complementary ways to illustrate the formation of (1) some of the principal Skaergaard structures involving blocks and layers; (2) modally graded layers in the Layered Series that rhythmically alternate with uniform layers; and (3) modally sorted layers in the Upper Border Series featuring “underside draping” beneath small included blocks. Explanations are provided for (1) why plagioclase did not float away from the tops of graded layers even though it was less dense than the liquid, and (2) how the liquid part of a current was fractionated away from the crystalline materials. Modal and grain-size data from Skaergaard intrusion graded layers are shown to be in excellent accord with characteristics predicted for layers sorted by currents; a synthesis diagram is presented illustrating how all the above processes may have functioned in concert in the intrusion.