The geochemical variation of the whole rocks analyzed during Leg 176 is consistent with the modal variations. For example, Figure F19 shows a scattered but significant correlation between CaO/Al2O3 and augite/plagioclase ratios (Fig. F19A) and a correlation of TiO2 (and Fe2O3) with modal oxide abundance (Fig. F19B). Some of the scatter in these geochemical plots is undoubtedly due to the small sample size for chemical analysis in relation to the coarse grain size of the rocks.
Ca# and Mg# are well correlated both within rock types and for all rock types as a whole (Fig. F20). This correlation illustrates magmatic evolution from primitive melts that produced high Mg#, Ca# cumulates (troctolitic gabbros and olivine gabbros) to more evolved melts that produced lower Mg#, Ca# rocks (gabbros and gabbronorites), and probably to highly evolved melts that produced oxide-rich gabbros and gabbronorites. However, the Mg# in the oxide-rich rocks partly reflects the accumulation of Fe-Ti oxide. In a plot of TiO2 vs. Ca# (Fig. F21), it is evident that most of the oxide gabbros and oxide gabbronorites have Ca# (Ca# = 56-73) comparable to the olivine gabbros (Ca# = 61-77, and two at 78 and 92, respectively). As the augite/plagioclase ratios of the oxide-bearing rocks in the majority of cases are similar to the ratios in the olivine gabbros (Fig. F19A), it seems that the oxide gabbros are not generally more evolved than the most evolved olivine gabbros. One possible explanation is that a change in oxygen fugacity or the exsolution of an iron-rich liquid was involved in the crystallization of Fe-Ti-oxides, rather than saturation during the final stages of a liquid line of descent. This would also supply a mechanism for the abnormally high abundance of oxides in some oxide gabbros. Alternatively, the silicates of the oxide gabbros may not have been in equilibrium with the same melt that formed the oxides, but rather are unequilibrated wall-rock minerals. This would be in contrast to the results estimated by electron microprobe on Leg 118 samples (Natland et al., 1991), which indicate that the silicate minerals were in equilibrium with an evolved high-Fe liquid. An additional problem that remains with this alternative, however, is how an extremely Fe-Ti-rich oxide rock formed, without re-equilibrating the Ca# of the host gabbro.
A remarkable feature of the averages of olivine gabbros and gabbros is their similar Mg# and Ca# (Table T3). Moreover, both major and trace element averages are the same. This implies that the main difference between the two rock types, the abundance of olivine, is not a function of magmatic differentiation of the parent magma. There is also a lack of correlation between the olivine mode and Ni abundance or Mg# of the two rock types, which is evidence that the olivine enrichment in these rock types is probably not related to olivine accumulation by settling. The modal variation present in the samples is probably a function of the same crystallization-related processes that led to the observed igneous layering, and not to a first-order process such as fractional crystallization.
Opaque oxides are very unevenly distributed in the Hole 735B core. Abundances averaged over 1-cm segments of the core range from 0.1% to more than 40% (Fig. F22). In general, the typical rock recovered from Hole 735B has less than 1% opaque oxides evenly distributed as small, interstitial, anhedral crystals. However, within a number of thin segments of the core, 1 to 10 cm in thickness, opaque oxides were concentrated by late-stage processes. Within these oxide-rich segments, oxides commonly form a network or matrix of intergrown anhedral magnetite and ilmenite crystals surrounding the silicate mineral protolith (Figs. F23, F24, F25). Many, but not all, of the oxide-rich zones are also relatively enriched in sulfide minerals.
Although there is considerable scatter in the amount of oxides present in individual samples, there is a general decrease in oxide abundance with depth, and a decrease in oxide abundance with depth within those zones that contain oxides (Fig. F22). Although detailed data are not available for the cores recovered from 0 to 450 mbsf during Leg 118, the trend observed in the Leg 176 core appears to continue upward in the section, reaching a maximum relative abundance in the interval from 224 to 272 mbsf (Natland et al., 1991). In the upper part of the section, the felsic veins are, in general, very poor in oxides. In the lower part of the core, however, there is a strong correlation between felsic veins and oxide concentrations. In this lower section, most of the felsic veins >1 cm in thickness contain coarse subhedral to euhedral oxide crystals along their margins or are associated with high oxide concentrations in the host rock within 5 cm on both sides of the vein. This relationship strongly supports a genetic relationship between at least some of the felsic veins and some of the oxide concentrations.
There is a strong association between the intensity of deformation and the abundance of oxides. In general, oxide-rich segments of the core are concentrated in those sections that have had the most intense deformation (Fig. F26). This relationship accounts for the strong correlation between magnetic susceptibility, which is sensitive to the abundance of coarse-grained magnetite, and deformation intensity (see "Physical Properties"). The timing of deformation and oxide crystallization is difficult to determine with certainty because of the ease with which oxide minerals texturally re-equilibrate at high temperatures. In most cases, however, it appears that the oxides postdate the major phase of deformation, in that they occur as an undeformed coarse matrix supporting sheared and brecciated silicate porphyroclasts. These samples may represent shear zones that were cemented by oxides after the main phase of deformation, or shear zones that have undergone extensive oxide recrystallization after the main phase of deformation. In a few samples, the oxides clearly predate the deformation and are present as fine-grained stringers that parallel the foliation.
The relationship between oxide-rich gabbros and deformation is not consistent throughout the core. Oxide-rich gabbros have at least three distinct textures: (1) in many samples, the crystallization of oxides is concentrated in shear zones and postdates the main episode of deformation, as evidenced by textures in which sheared silicate porphyroclasts are supported by an undeformed oxide matrix (Fig. F23); (2) in some samples, the crystallization of oxides is concentrated along shear zones and clearly predates at least some of the deformation, in that sheared lenses of oxides parallel the deformation-induced foliation (Fig. F24); and (3) in a few samples, the crystallization of oxides appears to be unrelated to deformation, in that coarse oxide segregations are localized along grain boundaries in undeformed rocks (Fig. F25) or are localized along brittle fractures at an angle to the deformation textures in the rocks. In general, oxides are least abundant in the least deformed rocks and most abundant in the most intensely deformed rocks.
Thin sections were examined in reflected light to identify oxide and sulfide minerals and to establish relationships between them and adjacent silicate minerals. Primary oxides and sulfides, which respectively crystallized and segregated immiscibly from the melt, are present in almost all samples. A few samples, chiefly troctolites, contain Cr-spinel (Fig. F27A). Secondary magnetite is almost universally present, secondary sulfides occur in many samples, and native copper occurs in one.
Of these minerals, only the primary oxides, ilmenite and magnetite, are ever present in more than accessory amounts. Rocks with 2%-10% oxide minerals in hand specimen are termed oxide gabbros or oxide-olivine gabbros, and the combined abundances of these "opaque minerals" in thin sections of many of these rocks are recorded as modes in Table T7. Where small amounts (1%-2%) of oxide minerals are observed in hand specimens, the rocks are termed disseminated oxide olivine gabbros, gabbros, or gabbronorites, depending on the silicate mineralogy. However, the gradational aspect of oxide abundances is clearly evident in thin section, as no rock has absolutely no ilmenite, and only a few, all highly altered, have no associated primary magnetite. Similarly, primary sulfides are present in all but a few highly altered rocks.
Nevertheless, in all but the oxide gabbros, abundances of oxide minerals and certainly the sulfides are low enough that it is all but impossible to determine them by point counting. Estimations such as "lots," "some," or "little" might be applied, but in the end, probably only the two estimates, "lots" and "none," have much significance. The most important observations in thin sections have little to do with quantity, but with the simple identification of minerals and the manner in which they occur in the rocks. Accordingly, the petrographic information on oxide minerals and sulfides is recorded under the "Comments" sections of the individual thin-section description forms rather than in a spreadsheet.
As to mineral identifications, there are only a few. The thin sections have abundant and coarse, to rare and vestigial, primary ilmenite magnetite intergrowths in which the effects of oxyexsolution are minimal to nonexistent. In this regard, the mineral name "ilmenite" as used in this report always refers to the solid solution of end-member ilmenite and hematite; "magnetite" refers to the solid solution of end-member magnetite and ulvöspinel. Among the oxides in the thin sections, however, there are few examples of development of exsolution lamellae, reflecting unmixing of these solid solutions, although they are more common in magnetite than in ilmenite. Extended oxyexsolution, in which mineral phases such as rutile and pseudobrookite form in still other criss-crossing exsolution patterns (e.g., Haggerty, 1991), and which is fairly typical of abyssal gabbros substantially transformed by hydrothermal alteration, does not occur. With only a few exceptions, the oxide minerals reflect the generally fresh condition of the rocks and only minimal influence of hydrothermal alteration.
The grain size of ilmenite and magnetite is roughly in accord with that of the surrounding rock. An exception is the fairly thick (10-15 cm) oxide concentrations that occur in some shear zones and breccias. In such cases, the individual grains in large oxide intergrowths are usually coarser than many of the silicate minerals they surround, particularly the micron-sized neoblasts of clinopyroxene and plagioclase that are present as breccia fragments in some samples. In other examples, subsolidus crystal-fabric deformation and cataclasis have clearly incorporated the pre-existing oxide minerals, as they are strewn along the foliation in such rocks and are substantially divided into subgrains on the order of the size of similarly deformed proximal silicates. Also, in these coarse-grained, deformed rocks, brown amphibole may be particularly abundant, typically forming rims on large patches of oxides. In addition, sulfides can compose up to several percent of the opaque concentration in some samples, several of which contain fairly large euhedral crystals of apatite enclosed within the oxides. This is an indication of the extremely evolved composition of the melts from which the oxide minerals precipitated.
In most samples, the proportion of ilmenite to magnetite in the oxide intergrowths is difficult to estimate. In some samples, magnetite is the more abundant mineral, but in most samples ilmenite is more abundant. Estimates of normative proportions of ilmenite and magnetite using X-ray fluorescence chemical analyses from Leg 118, for which FeO has been determined by titration (Robinson, Von Herzen, et al., 1989), suggest a nominal ratio of ilmenite to magnetite of 4:1, but with proportions as low as 2:1 or as high as 8:1 within individual analyzed samples.
The only other oxide minerals worthy of note are secondary in origin. One of these is magnetite, which is typically found as fine, elongate, dendritic crystals together with a clear amphibole at the altered rims of olivines in almost all rocks. In some more completely replaced olivines, the magnetite crystals are actually fairly large and intricately intergrown (Fig. F27B). The fine-grained magnetites may be the source of the stable magnetization component within these rocks. In some samples, primary magnetite is partly to completely replaced by secondary titanite, and the same rocks may contain rhombs of titanite in the rock matrix and in amphibole veins. Details of the replacement of magnetite by titanite, especially the grain-boundary relationships, are best seen in reflected light. In some of the most extensively altered samples, magnetite originally intergrown with ilmenite has been entirely replaced by green amphibole, which retains a trellis of relic ilmenite exsolution lamellae (Fig. F27C). In such rocks, there is not even any secondary magnetite; the rocks are so deficient in magnetite that they have virtually no magnetic susceptibility as measured, for example, using the multisensor track. Although such rocks are present only in rare and narrow intervals, they record such extensive hydrothermal alteration that virtually all the magnetite, primary and secondary, has been wiped out.
This is one of the rare cases in which titanium clearly was not a conservative component in the rocks during alteration, but rather the titanium in these rocks was dissolved and moved around in hydrothermal solutions. The titanium now present in titanite rims clearly came mainly from the ulvöspinel component of the magnetite. The titanite rhombs in amphibole veins, however, contain titanium that was probably derived from both primary magnetite and pyroxene. These textures provide proof of the mobility of titanium in solutions during certain conditions of hydrothermal alteration.
Finally, in one restricted region of the core where oxidative alteration occurred, between about 500 and 600 mbsf, and in which calcite veins are prominent, iron oxyhydroxide and smectites replace olivines, the effect of which is to lend a brownish freckled appearance to host rocks. This is the vicinity of the most prominent fault in the core.
The principal sulfide minerals are pyrite, pyrrhotite, chalcopyrite, and pentlandite, with at least two, and sometimes all four, of these minerals intergrown. These are typically globular to subrounded in shape or otherwise conform to the shapes of adjacent silicate minerals, and they are usually intergrown or associated with primary ilmenite and magnetite together with brown amphibole (Fig. F27D, F27E, F27F). All of these sulfide mineral phases in a sense have to be considered as something other than primary, because they crystallized at relatively low temperatures (~700şC) from what were originally immiscible sulfide droplets, containing Fe, Ni, and Cu, which segregated immiscibly from silicate melts at much higher temperature (e.g., Czamanske and Moore, 1977; Barton and Skinner, 1979). Secondary sulfides include rare bornite and chalcocite, found replacing primary chalcopyrite, cubic crystals, veins and vein networks of pyrite, and needles of pyrrhotite, all invariably associated with secondary clear or pale green amphibole and smectite.
The Leg 118 site report for Hole 735B (Robinson, Von Herzen, et al., 1989) notes the prominent occurrence of globular sulfides in the oxide gabbros, which are so abundant in cores from the topmost 500 m of the hole. Not only do the sulfides occur in the same rocks as the oxide minerals, but the two mineral groups are intimately associated texturally. That is, many very large globular sulfides, being aggregates of pyrite, pyrrhotite, and chalcopyrite (Natland et al., 1991), are embedded within large concentrates of ilmenite and magnetite. The two together often are present in shear zones or surround fragments of brecciated silicate minerals. Such rocks have their counterparts in many of the deformed oxide-rich zones cored between 504 and 1508 mbsf during Leg 176. Natland et al. (1991) speculated that the occurrence of oxides and abundant sulfides together was a consequence of extreme high-iron differentiation of abyssal tholeiitic basaltic magma, such that when oxide minerals began to precipitate, extensive sulfide segregation had to follow suit. Natland et al. (1991) also noted that chalcopyrite forms quite a high proportion of the sulfide aggregates and that glass margins to basalts dredged nearby have low concentrations of sulfur. They therefore inferred that basaltic melts in this region initially were not saturated in sulfide, and thus that residual liquids became enriched rather than depleted in Cu during magmatic differentiation, before sulfide saturation was achieved. Thus, when oxide minerals joined the liquidus, Cu-rich sulfides began to segregate in earnest. Alt and Anderson (1991), however, determined that primary sulfide mineral aggregates in olivine gabbros drilled during Leg 118 include pentlandite as well as chalcopyrite and the iron sulfides. This suggests that sulfide saturation actually occurred much earlier during magmatic differentiation than proposed by Natland et al. (1991).
The resolution of this conundrum is revealed in the thin sections of Leg 176. Ilmenite occurs in every rock, even in troctolites, where, although it is very rare, one or two grains a few microns in diameter are present in every thin section. There are two ways in which a fairly primitive olivine gabbro can reach the point of crystallizing ilmenite. An interstitial oxide mineral can crystallize from a small percentage of trapped intercumulus melt (Wager et al., 1960). If such a melt were basaltic in composition, then adjacent olivines, pyroxenes, and plagioclases might be zoned, and the modal proportion of those zoned minerals, plus the interstitial oxides, could be a measure of the percent of melt trapped in the rock. Usually, however, early formed silicate minerals are not zoned because slow cooling allows intracrystal homogenization by diffusion. In such cases, the amount of trapped melt is best estimated from the concentration of strongly excluded elements, such as P or U (Henderson, 1970). In portions of the core from Hole 735B, however, oxide minerals are concentrated in zones of deformation, suggesting that highly differentiated iron-rich melts flowed through channelized structures that developed in the rocks as a consequence of the deformation and compaction they experienced. This flow may have operated on the scale of individual grains, so that iron-rich melts were able to permeate the existing intergranular porosity structure of adjacent rocks, even olivine gabbros and troctolites, and there precipitate (or segregate) in very tiny proportions the same sorts of minerals that occur more massively in oxide gabbros. The observed oxide minerals and their associated globular sulfides thus might not represent trapped melt, but flowing melt.
In rocks with dispersed rather than massive primary oxide minerals, the oxides and sulfide occur consistently in a single association within which brown amphibole is also a prominent phase. The several minerals, in this association, are always present as intergranular phases (Fig. F28A). In some, troctolites, olivine gabbros, gabbros, and gabbronorites, brown amphibole is quite abundant, forming up to 2% of measured modes, and in such rocks it is more abundant than the oxide minerals and sulfides it contains. Because in many rocks there are both brown and green amphibole, with the latter clearly being secondary in origin, there is some uncertainty as to whether brown amphibole should also be considered a primary mineral, especially where it is intergrown with the green amphibole (Hébert and Constantin, 1991). This issue cannot be resolved using thin sections only, nor is it likely that all brown amphibole should be considered primary (Mével, 1987). However, combined electron-probe and ion-probe analysis can be used to establish the magmatic origin of certain brown amphiboles in abyssal gabbros (Gillis, 1996). In Leg 176 gabbros, however, the brown amphiboles are intimately intergrown with both magmatic oxide minerals and globular sulfides, which are segregations of immiscible sulfide melt. In some cases, ilmenite is intergrown symplectically with olivine (Fig. F28B), orthopyroxene, or amphibole. In the absence of green amphibole in many of these rocks, symplectites of ilmenite and brown amphiboles are likely to be magmatic in origin.
In many samples, the minerals of this particular intergranular association extensively penetrated the fabrics of their host rocks. Commonly, elongate ilmenite, blebs of sulfide, and patches of brown amphibole are within individual coarse grains of clinopyroxene, especially along cleavage planes (Fig. F28C). Whereas the amphibole by itself might be construed as a replacement of the pyroxene, most of these inclusions are bounded by faceted pyroxene crystal surfaces, and all of them are present together in the same mineral grains. They have not eaten away at the pyroxene; they have simply filled in spaces within it that were originally the consequences of crystal growth.
Ilmenite is even found along partings and fractures of olivine crystals. In the rare rocks with Cr-spinel, one can find the spinel partly jacketed by ilmenite (Fig. F28D) or ilmenite penetrating into fractures within the spinel (Fig. F28E). All of this is evidence for the ability of a fluid from which oxide minerals, sulfide globules, and brown amphibole were simultaneously forming, to penetrate the very finest-scale interstices of the fabric and mineral structure of primitive rocks. The mineral association suggests a highly evolved penetrating hydrous melt that was simultaneously rich in Ti, Fe, and sulfide.
Pentlandite, the apparently anomalous mineral in this association, can be seen as fine flaring or flamelike structures, somewhat more reflective than enclosing pyrrhotite, in many of the intergranular sulfides, especially in the olivine gabbros. Rarely, the outlines of incipient cubic crystals are present. The pentlandite flares often emanate into the pyrrhotite from the surfaces of intergrown yellowish chalcopyrite. Black-and-white photomicrographs fail to do justice to these features.
These pentlandite-bearing sulfide globules are clearly part of this characteristic intergranular mineral association that also includes ilmenite, intergrown magnetite, and brown amphibole. In rare examples, the pentlandite occurs within intergrowths of iron and copper sulfides and ilmenite, all surrounded by brown amphibole. The nickeliferous sulfide thus appears to be part of a late-crystallizing intergranular mineral assemblage in these rocks. It does not represent any early-segregating sulfide assemblage and is, therefore, not evidence for early saturation of sulfide during crystallization differentiation of parental basaltic magmas.
In part, this interpretation depends on negative evidence, namely that globular sulfides, recrystallized to mineral intergrowths, occur neither as inclusions in the olivines of troctolites and olivine gabbros nor in hypidiomorphic intergrowths of plagioclase with olivine. This is where the descriptor "none" becomes important even though it is difficult to establish. Sulfides at grain boundaries do not count; only those truly enclosed within early-crystallizing minerals provide proof of early sulfide saturation. None was found. The sulfides thus segregated after these minerals crystallized, and the melt was not at sulfide saturation while they did. One can find sulfides in cracks or at grain boundaries, and these are commonly associated especially with ilmenite but sometimes even amphibole, within these grains. There are also globular sulfides in some pyroxenes and plagioclases. However, the crucial question is the stage of differentiation at which sulfide segregated. This was not early during the magmatic evolution of these gabbros.
Highly evolved, very iron-rich basaltic magmas contain almost no nickel. They do not produce nickel-rich immiscible-sulfides. In the Leg 176 samples, however, the Ni evidently came from the olivine, next to which the intergranular amphibole-oxide-sulfide assemblages commonly occur. Apparently, reaction of the olivines with the through-going fluids resulted in dissolution of the Ni, and its later but nearby separation from those fluids in the form of immiscible sulfide droplets. The original interpenetrating melts therefore need not have had much nickel.
One final note concerning overall trends in the core. In the deepest rocks recovered from Hole 735B during Leg 176, between ~1400 and 1500 mbsf, there is very little ilmenite. Intergranular amphibole is also scarce. Sulfides, of the usual multiphase variety, are much more abundant, especially in proportion to the very small quantities of brown amphibole and ilmenite in the same rocks. In the classical cumulate sense, these rocks could be described as almost ideal adcumulates; that is to say, they are rocks that retain only the tiniest fraction of trapped intercumulus melts (Wager et al., 1960), as evidenced by the extremely low proportion of ilmenite and associated magnetite they contain. In the alternative context of the potential for these mineral phases to represent some sort of migrating, iron-rich, hydrous melt phase, it is clear that if such a melt phase was present it was largely expelled from these rocks before much crystallization of amphibole-ilmenite-sulfide intergrowths occurred. Nevertheless, the sulfides are fairly abundant, suggesting that sulfide segregation occurred before crystallization of ilmenite and amphibole from the throughgoing hydrous melts and that the sulfide globules, once formed, were retained in their host rocks while surrounding fluid was expelled. This presumably had to do with the much higher density and viscosity of immiscible sulfide than of hydrous silicate melt.
Details of this process need to be worked out with more care using data from electron and ion microprobe analyses. The general problem of these intergranular mineral assemblages is likely to be tied very closely to the formation and migration through the rocks of iron-rich melts in general and of the formation of massive oxide-sulfide concentrates in other parts of the core.
Felsic veins were recovered throughout Hole 735B from 500 to 1500 mbsf during Leg 176. These veins vary in size, geometry, and detailed mineralogical and chemical composition. All such veins, taken together, make up ~0.54% of the total volume of the core. Despite the small volume, their occurrence as an integral part of the gabbroic sequence has important implications for the physical and chemical processes taking place in the context of magma evolution and crustal accretion. Figure F29 shows that despite the scatter, the overall vein abundance recovered during Leg 176 clearly decreases with increasing depth downhole except for the last occurrence at 1430 mbsf.
The felsic veins vary in geometry from regular planar dikes/dikelets to irregularly shaped pockets, and in size/thickness from a few millimeters to several centimeters. Many of the veins are apparently of igneous origin with characteristic primary igneous textures and sharp intrusive contacts (Fig. F30). Some preserve igneous textures, but have clearly undergone subsequent high- and low-temperature alteration and have developed diffusive or reactive boundaries with the host gabbros. Still other veins, particularly some of the thin veinlets/pockets, have uncertain origins. Those associated with late-stage fractures with reaction halos may be of hydrothermal origin, but others could be either igneous or metamorphic.
Most of the veins are leucodiorite dominated by plagioclase plus small amounts of green amphibole. Other common lithologies include diorite, trondhjemite, and tonalite with variable amounts of dark-green amphibole, quartz, and biotite. Granitic veins were also recovered (e.g., an X-ray fluorescence chemical analysis from Sample 176-735B-99R-4, 106-108 cm, has ~28% normative quartz and 23.5% normative orthoclase). They often are irregularly shaped and coarse grained; have abundant K-feldspar, plagioclase, quartz, and biotite; and have a well-developed micrographic texture. Less abundant primary minerals associated with these veins include pyroxene, zircon, apatite, and oxides. Superimposed on the primary phases in many of the veins are high-temperature and low-temperature metamorphic mineral assemblages including actinolite, secondary plagioclase, epidote, titanite, chlorite, quartz, clays, sulfides, and oxides. Some of the felsic veins with these assemblages may, in fact, be of hydrothermal origin.
Oxide-mineral concentrations and felsic veins are commonly associated. Oxides occur either as euhedral to subhedral grains or aggregates within veins, or in anhedral aggregates along the margins of the veins and extending into the host lithologies. This association has two possible explanations: (1) the felsic veins may, in fact, have intruded pre-existing oxide-rich horizons or veins. (2) The oxides may have been precipitated from a highly evolved, oxide-saturated, magma. This latter process will inevitably result in a more silica-rich residual melt-the vein material.
The origin of the igneous veins is not certain. Most of them likely result from extreme degrees of differentiation. For example, oxide crystallization at late stages of basaltic melt evolution will result in more silica-rich melts. This could also explain the oxide-felsic vein association on local scales (see above). The igneous felsic veins may serve as "conduits" for subsequent fluid migration, which explains why many of these veins are hydrothermally altered. Alternatively, some of the alteration may have resulted from fluid exsolution during the final stages of felsic magma crystallization. Further detailed petrographic, mineralogic, and geochemical studies are required to better understand the genesis of the felsic veins.
Layering as used here refers to planar magmatic features that crosscut the core at angles of 0°-40ş. The lateral extent of these features is not known and cannot be determined on the basis of a 5- to 6-cm-diameter core. These features are inhomogeneities defined by variations in grain size, modal proportions, and textural appearance. They do not include the fine-grained microgabbros that cut the core with irregular, often sinuous, sutured and intrusive contacts. The types of layering observed are (1) grain-size layering characterized by either sharp breaks in grain size or gradational variations in grain size (graded), (2) modal layering marked by distinct changes in the abundance of plagioclase, olivine, clinopyroxene, and Fe-Ti oxide, (3) magmatic foliation (igneous lamination) defined by the preferred orientation of plagioclase and in some cases olivine and clinopyroxene (see "Structural Geology"), and (4) layering defined by textural variants such as crescumulate texture.
Grain-size layering is by far the most common type of layering observed and is present in well-developed rhythmic intervals in several sections of the core (Fig. F2 , 827 to 914 mbsf and 1138 to 1220 mbsf). Figure F31 shows rhythmic cycles in grain size between medium and coarse-grained olivine gabbro from Section 176-735B-171R-4 and an isolated coarse-grained layer in a medium-grained olivine gabbro from Section 176-735B-186R-4. In Sections 176-735B-134R-2 to 176-735B-135R-2 (837 to 845 mbsf), grain-size variations are normally graded, coarsening downward, with plagioclase increasing in size from 2-7 mm to 10-20 mm and clinopyroxene increasing in size from 3-5 mm to 10-15 mm within each layer. Figure F32 shows a histogram of layer thickness for 34 layers identified from this part of the core. Layer thickness, which refers to the distance from the coarse base of one layer to the coarse base of the next layer, ranges from 6 to 22 cm with a dominant thickness of 11 cm.
Modal layering often accompanies the grain-size layering. Generally, the coarser parts of a layer are more mafic than the finer parts. In the graded grain-size layers mentioned above (Sections 176-735B-134R-2 to 135R-2), the mode typically varies from 60% plagioclase, 30% clinopyroxene, and 10% olivine in the medium-grained top of a layer to 55% plagioclase, 40% clinopyroxene, and 5% olivine in the coarse-grained base of a layer. A similar variation was observed in layers between 0 and 18 cm in Section 176-735B-138R-7 and was verified by point counting. In this case, the finer grained fraction has a mode of 58% plagioclase, 37% clinopyroxene, 4.5% olivine, 0.2% opaque minerals, and 0.3% amphibole, whereas the coarser grained fraction has a mode of 51.7% plagioclase, 36.8% clinopyroxene, 10.4% olivine, 0.4% opaque minerals, and 0.7% amphibole.
The most striking example of modal layering in the Hole 735B core is the occurrence of Fe-Ti oxide-rich layers (see "Opaque Minerals"). These layers generally contain 2%-20% Fe-Ti oxide together with 30%-50% plagioclase, 0%-5% olivine, 0%-5% orthopyroxene, and 30%-50% clinopyroxene and in many cases appear to be localized along shear planes. Melanocratic layers rich in clinopyroxene are rare. They occur at 47 to 54 cm and 110 to 118 cm in Section 176-735B-135R-1 (843 mbsf), 131 to 149 cm in Section 176-735B-144R-1 (914 mbsf), and 5 to 12 cm in Section 176-735B-203R-1 (1431 mbsf). Fe-Ti oxide is abundant (3%-10%) in the lower three of these layers.
Pegmatitic layers with an average grain size greater than 30 mm are intermittent throughout the core (Fig. F2). These layers range in thickness from 5 cm to 50 cm and consist mostly of plagioclase and clinopyroxene with lesser, variable amounts of olivine or orthopyroxene, Fe-Ti oxides, and sulfides. In some instances, the large crystals are oriented in a subvertical direction. In Section 176-735B-163R-3 from 11 to 18 cm, for example, single olivine crystals extend vertically for 5 cm or more. These crystals, together with plagioclase crystals, produce a crescumulate texture.
Two major paradigms form the backdrop for studies of the lower oceanic crust. First is the concept of layered basic intrusions formed by the cooling and crystallization of large batches of basaltic magma over tens of thousands of years. Layering in such intrusions is derived from a number of processes, including mechanical sorting, variations in cooling and crystal growth rates, crystal-liquid fractionation, and injection of multiple batches of melt (Wager and Brown, 1968; Cawthorn, 1996). A second paradigm that has shaped our ideas about oceanic crustal structure is the ophiolite model, characterized by a stratigraphic sequence of pillow lavas, sheeted dikes, isotropic gabbro, layered gabbro, and deformed ultramafic rocks. Recent conceptions of crustal generation at ocean ridges incorporate aspects of both layered intrusions and the ophiolite model. Sinton and Detrick (1992), for example, proposed that eruptions at fast-spreading ocean ridges are fed by a magmatic system consisting of a large crystal mush zone that is overlain by a narrow melt lens. During rifting episodes, melt is transported upward from the lens through dikes and onto the seafloor to form lavas. With time, the solidified mush (gabbro), dikes, and lavas are carried away from the ridge axis by seafloor spreading.
The importance of the present study is that Hole 735B represents the only deep hole (1.5 km) through an in situ section of the lower oceanic crust, providing a unique opportunity to investigate the composition and structure of this crust and to compare these observations with comparable ones on ophiolites and layered intrusions. The lithologies recovered during Legs 118 and 176 represent a generalized evolutionary sequence ranging from primitive troctolitic gabbros to highly evolved oxide gabbronorites. These lithologies, however, do not change systematically through the section but, rather, occur in separated enclaves that in places interpenetrate one another. Although the details of this interpenetration cannot be worked out in a single drill core, it is clear that large batches of magma evolving in situ, such as might be expected by analogy with layered intrusions, were not involved. The main phase of crystallization produced troctolites, olivine gabbros, and gabbronorites. Low abundances of incompatible elements suggest that interstitial melt is largely absent from these rocks.
We subdivided the igneous rocks recovered during Leg 176 from 504 to 1508 mbsf into 458 intervals, grouping them into seven units. The contacts between intervals were chosen to differentiate rock types based on grain size or mode. In some cases, these contacts appear to represent intrusive boundaries; in most cases, however, they separate local variations within a single intrusion. Likewise, the unit boundaries are chosen on the basis of changing proportions of rock types in the section but do not necessarily represent separate magmatic events. The sequence recovered during Leg 176 consists mainly of olivine gabbro (70%), gabbro (15%), and troctolitic gabbro (2%), with similar major and trace element compositions. The section also contains orthopyroxene-bearing gabbro and gabbronorite (6%), oxide gabbro (5%), and orthopyroxene-bearing oxide gabbro and oxide gabbronorite (3%). The bulk composition of the section, calculated from weighted averages of the rock types, is similar to MORB.
Oxide gabbros and orthopyroxene-bearing oxide gabbros and gabbronorites are major lithologies throughout almost 1000 m of Hole 735B and are present in minor proportions throughout. The formation of oxide gabbros clearly follows the main phase of crystallization, in that the oxide gabbros cut previously crystallized rocks. The strong correlation between deformation and oxide-rich gabbros in the Hole 735B section argues that magmatic differentiation was not driven by the filter pressing of residual melts out of cumulates during deformation, with those expelled melts accumulating in undeformed rocks. In such a model, oxide-rich gabbros should be most abundant in rocks with the least evidence for deformation and should be rare in highly deformed samples. The relationships in the Hole 735B core instead require that residual melts are depleted in rocks that show little evidence of deformation or compaction and are concentrated in rocks that are the most highly deformed. At least three models are supported by the data: (1) lenses of late-stage magma may have acted as zones of weakness along which deformation was concentrated; (2) the late-stage magmas from which the oxides crystallized may have been concentrated in zones that had been previously sheared, because these zones have greater high-temperature permeability; or (3) active shear zones acted as conduits for melt transport through the section. One or more of these processes must have occurred in the Hole 735B section.
The high concentrations of oxides present in some samples require that large volumes of melt were transported through these rocks. A ferrobasaltic liquid migrating through a system of fractures and shear zones would be unlikely to precipitate more than 5% oxides. Some samples contain in excess of 15% oxides, and the shear zones within these samples contain even higher oxide abundances. A ratio of transported melt to fractured rock in excess of 5 to 1 would be necessary to account for these samples. Previous reports of oxide-rich gabbros localized along shear zones in ophiolites are limited to the single instance of the Lizard complex (Hopkinson and Roberts, 1995), and oxide gabbros in layered intrusions are generally present as undeformed, conformable layers or segregations formed during the final stages of crystallization as a result of progressive magma enrichment. The oxide gabbros recovered during Legs 118 and 176 require a differentiation mechanism unlike those previously reported in ophiolites and layered intrusions.
The microgabbros from Hole 735B record relatively late intrusive events. Contact relationships indicate that they were intruded into solid or near-solid-state, coarse-grained, gabbroic host rocks, and as a result, we interpret the microgabbros as a later, separate magmatic cycle. The microgabbros, however, cover a wide range of compositions and rock types and, therefore, they cannot be explained by a single late parental magma. As a result, the microgabbros appear to represent a late episode of intrusion covering approximately the same magma compositions as were responsible for the earlier coarse-grained gabbros. The microgabbros may represent channels through which melt was transported through the earlier coarse-grained, gabbro section. Additional evidence for melt transport through the section is present in the form of monomineralic plagioclase channels that appear to replace the gabbro host rock but preserve only high-temperature igneous minerals, and have fresh igneous textures in thin section. The relative timing of the microgabbros, the plagioclase-rich channels, and the Fe-Ti oxide gabbros is not clear, but all three appear to postdate the main phase of crystallization of the coarse-grained gabbroic rocks.
Felsic to dioritic dikelets and veins with clear crosscutting relationships represent a final intrusive event in the sequence. These commonly intrude fractures or shear planes within the earlier gabbroic sequence. In some cases, the felsic veins are bordered by oxide-rich gabbros, the oxides apparently derived from the late-stage melt. Many of the felsic veins are overprinted by later deuteric or hydrothermal alteration.
Igneous layering is an important feature of both ophiolites and layered basic intrusions where parallel layers can be traced for some distance in weathered outcrops. No well-documented layering, however, has ever been reported in gabbros sampled from the ocean floor except for those recovered during Legs 118 and 176 from Hole 735B. Many styles of layering are present in the Leg 176 sequence, including rhythmic grain-size, modal, and textural layers. Some sections of the core are rhythmically layered, with grain-size or modal layering repeating at regular intervals of 10 to 20 layers through a 1- to 2-m section. Thick, intensely layered sequences, such as have been reported from layered intrusions and in the lower parts of some ophiolites, are not present, however, in the Leg 118 and Leg 176 cores. Likewise, the systematic cryptic variations in bulk rock chemistry common in some layered intrusions and ophiolite sequences are not found in the Hole 735B section. In addition, the abundance of oxide gabbros localized along shear planes in the Hole 735B sequence, clearly set it apart from gabbroic sequences in typical ophiolites or layered basic intrusions.