Based on shipboard analyses, a fairly straightforward interpretation of the downcore trends in bulk-rock geochemistry seemed apparent at the end of Leg 176. It was that five major bodies of primitive gabbro, each more differentiated toward the top, each having olivine-rich, even troctolitic, gabbro at its base, and each some hundreds of meters thick, had contributed to the thickness of gabbros cored at the site (Shipboard Scientific Party, 1999c; Dick et al., 2000; Natland and Dick, 2001). These were subsequently intruded at hundreds of places by the more differentiated gabbros, most of them carrying visible oxide minerals, and this added several hundred more meters of rock to the section.
The five masses were called "plutons" (Shipboard Scientific Party, 1999c; Dick et al., 2000). Whether this term is technically appropriate may be questioned. However, each was viewed as having served separately for some time as a sustained locus of magma injection and differentiation before ceasing to function; thus, they might have been potential sources for chemically distinctive fine-grained dikes and extrusives that perhaps erupted at monogenetic cones of the type found on the rift valley floor along the Mid-Atlantic Ridge (Smith and Cann, 1992). Mineral compositions partially match those of common phenocrysts in abyssal tholeiites from the Indian Ocean (Natland and Dick, 2001). Each pluton also consists of rock crystallized from many separate injections of magma, as represented by igneous contacts between many lithologic intervals, so that there never was any single mass of molten material in the crust larger than several meters, or several tens of meters, thick. The original thicknesses of each injection, however, are difficult to say, since only a portion of them are present as cumulates from which almost all interstitial melt was expelled. The rocks now are adcumulates, with very low retained percentages of now-crystallized interstitial melt, most of which was removed by combinations of matrix compaction or filter pressing, partial dissolution and reprecipitation of silicate minerals, and deformation (Bloomer et al., 1991; Ozawa et al., 1991; Natland et al., 1991; Dick et al., 1991a; Natland and Dick, 2001). If the original intercumulus porosity was 40%-50% (Jackson, 1961), each injection then was about twice the thickness of the compacted crystalline residue that now remains.
The oxide gabbros, then, can be viewed as complementary to this process, since they seem to express, at least partly, where the expelled, and necessarily more differentiated, interstitial liquids went. The close association between oxide gabbros and zones of rock with gneissic, porphyroclastic, and mylonitic textures also implicated deformation with shear as a likely contributor to the forces that drove residual melt from the mass of consolidating and compacting cumulus minerals.
Natland and Dick (2001) used the procedure of Lange and Carmichael (1990) to calculate densities of synthetic liquids produced during equilibrium crystallization of parental abyssal tholeiites in the course of several studies. The synthetic liquids cover a full range in the extent of differentiation, and are all less dense than normative densities of Hole 735B gabbros, even when the latter are presumed to be hot and to contain several percent of interstitial melt. Normative densities are calculated from CIPW normative proportions using the procedure of Niu and Batiza (1991) and are only slightly higher than grain densities measured on actual specimens from Hole 735B (Shipboard Scientific Party, 1989, 1999c). Both are significantly higher than densities calculated for synthetic abyssal tholeiite liquids. When no other forces acted on them besides their own buoyancy, basaltic liquids in the crystallizing gabbros of Hole 735B thus tended to migrate generally upward in the mass of rock. They flowed along whatever avenues of escape they could find, following fluctuating patterns of porosity structure in the deforming rock on scales ranging from that between individual mineral grains to networks of fractures and channels, the largest of which correspond to the widths or thicknesses of oxide gabbro seams and many lithologic intervals. The process of compaction and expulsion of intercumulus melt continued even during the emplacement and crystallization of these rocks since they, as well as the olivine gabbros, are adcumulates. As the course of this flow was generally upward, it was also toward cooler rock; thus additional differentiation occurred, leading to formation of the late and usually undeformed veins of tonalite and trondhjemite.
This picture, however complicated it is in detail, was fairly satisfying and is still, for all the new analyses, mainly correct, at least as regards processes. What has changed, however, is the status of the five plutons. The three based on geochemical trends during Leg 176 and which comprise almost all the core retrieved during that leg have largely disappeared with the more complete perspective the new chemical analyses provide. Nor do extensive mineral analyses provide a different picture. We now believe that the core should be described in the following way.
In a pair of downhole trends (Fig. F5), two principal bodies of olivine gabbro and allied more primitive types are apparent in the upper 540 m of core. The most differentiated rocks, with lowest Mg# and greatest TiO2 contents, are at the tops of each. These are laced with seams of oxide gabbro—still lower in Mg# and very high in TiO2. The new data reveal these trends more completely and precisely than before.
Splitting the two principal bodies almost exactly is lithologic Unit IV, the oxide gabbro complex already described, which has very high TiO2 contents and is >50 m thick. Overall, however, the superposition of oxide gabbros onto more primitive olivine gabbros and troctolites provides a strong bimodality of composition to the upper 500 m of the core (Dick et al., 1991a). This is evident not just in geochemistry but in mineralogy as well (e.g., Ozawa et al., 1991). This is not to say that intermediate compositions are lacking, only that the preponderance of the core obtained during Leg 118, whether viewed from the perspectives of lithology, bulk composition, or mineralogy, falls into two main groupings.
With the Leg 176 drilling, however, the bimodality disappears below 540 mbsf. Instead, in both Mg# and TiO2 contents the new samples align between the two principal ranges in Mg# and TiO2 contents present in the upper third of the core (Fig. F5), and they are intermediate in other aspects of geochemistry as well. There are fewer oxide gabbros, particularly below 1100 mbsf, and there are fewer troctolites at all levels. A tendency toward more primitive compositions only emerges toward the bottom of the hole.
This gap filling was evident from the Leg 176 shipboard analyses, nevertheless, breaks in the trend, for example of Mg# among olivine gabbros at ~950 and 1350 mbsf, still suggested a continuation of the sequence of plutons. Now, however, with all the new data, there are no breaks in the trend of TiO2 contents with depth to correspond with much feebler apparent discontinuities in Mg# at these depths. Although there is a cluster of somewhat more differentiated gabbros between 540 and 750 mbsf representing gabbronorites of lithologic Unit VII (denoted by dark gray symbols in Fig. F5), these actually intrude olivine gabbros and rare troctolites (light gray and black symbols, respectively). The latter, considered separately, are not especially more differentiated than olivine gabbros extending downward to 1200 mbsf. The apparently more differentiated upper portion of the pluton postulated between 540 and 950 mbsf that was construed from shipboard data therefore was an artifact of the tendency to sample the most representative rocks in a given core for XRF analysis. The minor olivine gabbros in a 9.5-m core were not sampled simply because there was not time to analyze everything. Also misconstrued was the top of the pluton postulated below 1350 mbsf. By improving coverage, the new analyses demonstrate greater chemical variability over short lengths of the core, and this has erased false boundaries between bodies of rock originally thought to represent major divisions of the core. This is supported by lack of discontinuities in the compositions of major silicate minerals (Dick et al., Chap. 10, this volume) and by measurements of magnetic susceptibility (Natland, Chap. 11, this volume). The latter clearly delineate the contrast between gabbros with and without magmatic oxide minerals up and down the core (Natland, Chap. 11, this volume) and show that both types, not just the more differentiated oxide-bearing gabbros, are present at the tops of the formerly postulated plutons.
What alternatives are there now? Dick et al. (Chap. 10, this volume) believe that there are simply two domains of olivine gabbro in the core, one above and one below 540 mbsf, the two differing only in their degree or stage of differentiation. By different means, both Natland (Chap. 11, this volume) and Miller and Cervantes (Chap. 7, this volume) postulate some internal stratigraphy to the lower of these but do not see breaks in composition at the same places. Dick et al. (Chap. 10, this volume) raise the prospect that substantial reequilibration of some olivine gabbros toward more differentiated compositions has occurred because of wholesale migration of the melts that produced oxide-rich seams throughout fine porosity structure in much of the core. This in itself may have erased original petrochemical discontinuities among olivine gabbros. Dick et al. (Chap. 10, this volume) even propose this as the reason why the two upper chemical cycles in olivine gabbros above 540 mbsf—the two remaining of the original plutons—should now be treated as one.
For the present, we consider that there are three main series (Irvine, 1982) of olivine gabbro and allied troctolite down the core (Fig. F5). The upper two may still reasonably be interpreted as plutons in the sense described above, but the third may not, although we go so far as to recognize zones within it based on small shifts in composition down the core. The precise boundaries of series and zones and their relationships to sequences are given in "Appendix A." There may have been some significant shifts in the compositions, especially of ferromagnesian silicates, toward more differentiated compositions at various places in the core during reequilibration with percolating iron-rich melts, but to establish this, we need to consider mineral compositions in more detail than is possible at this time. These rocks cored during Leg 176, nevertheless, clearly are far more subtly variable than was originally determined on board ship.