Ferrar-style dolerites are found throughout the Transantarctic Mountains, running some 4,000 km across Antarctica; extending to Tasmania and South Africa to the Karoo system. An unusual characteristic of the Ferrar system is the extreme paucity of dike swarms that might well be expected in rifting systems. Major dike swarms are found in Canada and somewhat in the Eastern North American Rift System, but not so in Antarctica. The concentrated center of magmatic activity exhibited in the Dry Valleys is similar to what is found along ocean ridges, which is characterized by a series of spatially distributed major magmatic centers with underlying magmatic mush columns. Magma is produced and moves upward through the mush column and is distributed laterally, up and down the ridge by high level flow. The Ferrar may also be like this, as indeed most magmatic systems are, and then, where are the other such centers along the Transantarctic Mountains? The Dufek complex is clearly once such center and others may exist at various places along this range; there is telltale evidence at many locations. And it is remarkable that, by and large, there are always four major sills at most locations.

The behavior of fluids laced or heavily laden with particles is presented in theory and experiment using molten wax and water in a 2.5 m tall mush column. The idea is to simulate the patterns of crystal concentrations found in the Basement Sill. And, on a grander scale, to build an experimental system mimicking the operation of the full Dry Valleys magmatic system during its time of operation. In flow in a pipe, as the concentration of particles increases the vertical distribution becomes increasingly parabolic, reflecting the fluid flow velocity, even without solidification. With on-going solidification particles are trapped or frozen in place, such that if the flow starts and stops, a specific pattern of particles will be frozen into the final fluid. Patterns like this are found in the Basement Sill. Molten wax experiments simulating dikes propagating into viscoelastic crust, represented as gelatin, exhibit particle transport just as expected with high concentrations in the center and undergoing sedimentation as the dike flattens and becomes a sill. The large mush column experiments between interconnected tanks (sills) with water as ‘magma’ laden with particles, demonstrate the many ways in which layering forms through the completion between the conduits at each level.

The four principal sills, Basement, Peneplain, Asgard, and Mt. Fleming, are chemically and physically interconnected, forming a magmatic mush column, albeit being only observable for some 4 km downward in the crust, but the system may well extend downward through the entire continental crust. The distribution of MgO vertically through the Basement and Peneplain sills indicate that the filling point was in or about Bull Pass. And these profiles indicate that the episodes of emplacement were pulsatile, stopping and starting. Similarly, the upper, apparently much more featureless sills, when examined in detail also shown a history of piecemeal development through a series of multiple injections.

The discovery of the McMurdo Dry Valleys was an accidental result of the desire in polar exploration to find the South Magnetic Pole and the South Geographic Pole. James Clark Ross was astonished in 1841, after pushing his way through a thick collar of pack ice, to suddenly sail into an open body of water, McMurdo Sound, finding a large island (Ross Island) like Hawaii formed by a series of several large volcanoes, one of which was smoking and ready to erupt. He came here to find the South Magnetic Pole, which was too far inland to the west to reach easily on foot, as he had done years earlier in reaching the North Magnetic Pole. This opened the way for Robert Falcon Scott to come here in 1902–1904 with his Discovery expedition to make and attempt on the Pole. He set up camp on Ross Island and stayed for two years exploring various ways to reach the Pole. Albert Armitage, one his men, pushed a route directly west to see what was there and was astonished to find large valleys fully free of ice and snow, the McMurdo Dry Valleys.

The ultimate source of magma erupting on Earth’s surface has always been mystifying, and it continues to be so. The Ferrar magmatic system necessitates a source that has a fairly uniform bulk composition over a large region, such that the magmas produced are broadly similar over thousands of kilometers, similar to the processes producing magma along ocean ridges and island arcs. And there must be a ready supply of massive quantities of primocrystic orthopyroxene that has an isotopic identity distinct from the basaltic magma but similar to the immediately overlying continental crust. Quartz eclogite is a possible primocryst source rock that has a large modal volume of pyroxene, and this may well exist just below the continental crust as a product of basaltic, even doleritic, materials having once been in the crust and over time sunken back into the uppermost mantle. An interaction between the continental crust and the underlying mantle has long been proposed by many scholars and here in the Ferrar is strong additional evidence for this process.

This chapter shows the whole rock chemical compositions for all the sills in the Dry Valleys region. Harker diagrams for the major and selected trace elements clearly reflect the strong role of orthopyroxene (and subordinate clinopyroxene) as the major mineral phase controlling rock composition. These are low titanium, non-alkaline dolerites, which are common to this entire region and continuing to Tasmania and even as far as some of the Karoo system. On an AFM diagram the system almost mimics that of Skaergaard. This Opx control extends throughout the system. And upwards with the loss of Opx the upper sills overlap in composition with the Kirkpatrick Basalts. Various other diagrams, meant to discriminate between pyroxene and feldspar control, show the effects of crystal sorting locally and regionally, with the chilled margins being tholeiitic and the sill centers tending to ultramafic. A pattern persisting outwards from Bull Pass for at least 50 km north and south. Relative to the Basement Sill this appears as a large plume of crystal-laden magma, extending outward in all directions, exhibiting locally the strong effects of crystal sorting during emplacement.

The ultimate origin or provenance of the masses of orthopyroxene primocrysts is of paramount interest. Many lines of reasoning lead to the conclusion that these were entrained in magma ascending from deeper in the mantle, generated by the continental breakup of Gondwana, somewhere below the local continental crust. The crystals themselves look old, showing signs of long-term annealing, and the crystals are in strong isotopic contrast (Sr 87/86 and O-18) with the basaltic magma itself, with the Opx assemblage being much more radiogenic that the basalt. This is especially marked in the dais rocks in going from Opx dominated layers to more basaltic layers. Moreover, the basaltic magma itself, even when it carries no large primocrysts, is highly heterogeneous isotopically. A profile through the Peneplain Sill at Solitary Rocks, near Pandora’s Spire, in an otherwise thick (330 m) featureless sill, shows strong variations in Sr-87/86. This reflects what others previous show, that the Ferrar are isotopically "noisy" and, remarkably, exhibit isotope patterns similar to the local crust. The obvious answer that this is all from local contamination and weather cannot be true, but instead this comes from the uppermost mantle underlying local crust that has had a long physical association.

The Dais Intrusion in upper Wright Valley is a spectacular layered intrusion connected to Bull Pass by the Basement Sill, the connection of which can be traced along the south wall of Wright Valley, with the layering becoming more distinct and cleaner with distance to the Dais. The Dais Intrusion is relatively small for a body as well layered as this: it is perhaps 1,000 m thick but only the upper half is visible. The lower half of the visible section is coarse grained ultramafic orthopyroxenite with 20 wt.% MgO, containing several periodic meter-thick anorthosite layers, which are lobate in nature, perhaps reflecting formation by avalanching slurries. The upper half of the Dais Intrusion is more tholeiitic in bulk composition, and the transition between top and bottom is dramatic, marking event horizons, much like the transition between mantle and oceanic crust for Earth as a planet. The style of layering becomes more delicate upwards in the section. Because the body cooled relatively quickly many annealing processes have been quenched prior to full completion. These are clearly textural processes that actively operate in many larger bodies but are never seen because of strong long term annealing effects.

The integrated aspects of volcanic and plutonic magmatic systems are rarely exposed, yet this connection is fundamentally important to understand the evolution of magma, which is important, in and of itself, to understanding planetary evolution itself. Magmatic systems beneath major volcanic centers, including the ocean ridges, are an interconnected plexus of sills and conduits. Beneath Hawaii the underlying mush column extends through the entire lithosphere, whereas beneath ocean ridges the mush column is much less vertically extensive, yet they function in very similar ways in spawning a steady flux of basaltic magma. The Ferrar magmatic system exposed in the McMurdo Dry Valleys reveals critical connections at a high level in the crust, demonstrating the operation of a magmatic mush column.

The Ferrar Sills in the McMurdo Dry Valleys are remarkable in exhibiting a vertical variation in bulk composition from ultramafic in the lowermost, Basement, sill to evolved compositions in the uppermost, Mt. Fleming, sill at the top, overlapping with the compositions of the lavas and volcaniclastics. The system apparently developed from the top down, becoming increasingly capped by overlying materials and the solidification of earlier magma, forcing succeeding magmas to form deeper and more massive sills. And as the system became more mature and the country rock became heated, the later magma came carrying a massive slurry or sludge of large crystals, primocrysts, that furnish as tracers a clear indication how the Basement Sill was filled and how it evolved.

The Ferrar dolerites form an interconnected stack of four massive sills that in various places breakout to the surface, erupting lavas and volcaniclastics. They are massive horizontal sheets, emplaced when the massive Gondwana continent broke up some 180 million years ago. Somewhat similar, but much less well exposed, systems are found throughout the world, including Eastern North America. The entire system was formed in less than about 500,000 years and other expressions of the system are exposed over some 4,000 km, reaching all the way west to South Africa.

Where do all these huge volumes of noritic magma come from, representing the dolerites found globally associated with the Gondwana event. Although differing in chemical detail from one province to another, they all have a strong quartz–dolerite affinity and commonly show a large concentration of orthopyroxene primocrysts. Noritic magma can be produced in the lower crust but it is dissimilar to the rift-related dolerites. This is exemplified by the massive magmatic event at Sudbury, Ontario where an impacting bolide 1.85 Ga ago produced ~ 35,000 km3 of noritic magma in 3 minutes. Yet, a detailed chemical comparison of this magma with the Ferrar magmas shows a striking contrast on all counts. The Sudbury system shows no tight close variation in principal components as does the Ferrar. The Ferrar magmas, like other rift-related dolerites, are remarkable in their huge volumes, cohesive compositions, and relatively rapid generation and emplacement. The process giving rise to them cannot be tortuous and complex, but simple and straightforward, as in the production of the MORB of ocean ridges.

The timing of magma appearing in Gondwana rift systems is remarkable. In the Ferrar magma was not emplaced in the middle of the rift, as might be expected, but into the rift shoulder. And this apparently happened rapidly, perhaps taking only a few 100,000 years, and this rapidity may also be common to the Eastern North American system (ENA), but the emplacement pattern was in striking contrast. The ENA dolerites came late in the rifting sequence, after more than 5 km of sediment had already filled the developing rift, and the dolerites are not tightly tied to the rift faulting itself. And these sediments were not generally highly consolidated, so the ENA dolerites formed bulbous, irregular bodies at high levels in the system. The Ferrar sills are, in comparison, highly regular, reflecting the well consolidated section of sediments characterized by the Beacon Supergroup. This also suggests that this crustal section was not a normal rift, filled with a thick section of still-compacting sediments, but instead a mature section of continental crust.