Throughout Series 3 in Hole 735B, gabbro compositions are quite uniform. They are not appreciably more differentiated upward, unlike gabbros of Series 1 and 2. Temperatures of crystallization were restricted, whereas steeper thermal gradients prevailed when the rocks of Series 1 and 2 crystallized. Probably this is because the two plutons were emplaced at shallow depths at the ridge axis, where the full extrusive layer had not yet accumulated, and hydrothermal circulation could readily draw off considerable heat from the shallow magma bodies, forcing differentiation. Later, off axis, a thickened insulating upper crustal blanket capped the gabbros and continued intrusion of small magma bodies at many levels (drip irrigation) stabilized the temperature gradient, driving isotherms upward even as the section moved away from its primary source of heat. This was intrinsically a convective redistribution of heat by moving magma. Thus, the deeper olivine gabbros of Series 3 are at once more differentiated and more uniform in composition. Most of them also have a very coarse grain size (Shipboard Scientific Party, 1999c).
After initial injection and crystallization of axial plutons and off-axis intrusion, the third and final stage of the development of the lower crust in Hole 735B was relocation of late, imprisoned differentiates along numerous zones of weakness and shear that developed during initial stages of high-temperature crystal-plastic deformation. These tried to escape but could not. In a body of upraised isotherms and nearly uniform temperature from top to bottom, extremely differentiated oxide gabbros and trondhjemites could crystallize virtually anywhere in the section, and they did.
However, even the most iron-rich of these melts were buoyant in their crystalline matrix (Natland and Dick, 2001); thus, they tended to rise and concentrate higher in the section. Buoyancy was not the only force that acted on the melts, but it is the only one that was always present. The flow likely was modified by encounter with permeability barriers and possibly by dilatancy that opened in the course of deformation. The thick oxide gabbro of lithologic Unit IV represents a coalescence of strongly differentiated magmas. Some hundreds of meters of equivalent vertical section of differentiated cumulates are necessary to explain the thickness of Unit IV (Natland and Dick, 2001), although these need not immediately underlie these rocks. Unit IV is sandwiched precisely between two plutons, and it supplied trondhjemetic veinlets to cement the breccias at the top of the one underneath. We suggest that the repeated intrusion of these very late stage differentiates at this special location occurred because of the presence of a permeability pathway between two impermeable rock masses. The one beneath is the largest sequence of olivine gabbro in the entire section. Above it was the base of the upper pluton. The two together served to direct the flow of extreme differentiates into this narrow zone as expelled melt ascended from underlying compacting cumulates.
The oxide gabbros of lithologic Unit IV formed well within the gabbroic portion of the crust and seemingly at a location dictated by the existence of a narrow permeability pathway between more massive rocks. Therefore, it was never a melt lens of the kind detected seismically at the top of gabbros along the axis of the East Pacific Rise, and it never contained enough melt in it at any one time to be detectable seismically on its own, had that been possible 11 m.y. ago. It was merely the closest that this particularly slowly spreading ridge could produce in the way of a melt lens, given its low rate of magma supply. The oxide gabbros and other zones with strong preferred orientations of minerals provide potential analogs to sources of deep seismic reflections elsewhere in the ocean crust (Itturino et al., 1991; Chap. 5, this volume).