We encountered basement at 462 mbsf (drilling depth [457.5 mbsf curated depth]) at Site 1203 (Hole 1203A) and cored a ~457-m volcanic section, with an average recovery of 56.5% (see "Operations"). The basement sequence consists of 18 basalt lava units (216 m of recovered core), 12 volcaniclastic interbeds with occasional biogenetic sediments, and 1 recrystallized silty chalk interbed that contains sand-sized vitric material (Figs. F5, F6, F7; Table T4). The general lithology of the basement units correlates well with logging and drilling rate data (Figs. F5, F6), and we note that the units exhibit a range of alteration intensity between slight and complete (see "Downhole Measurements" and "Alteration and Weathering").
Site 1203 is ~26 nmi south of Site 883 and 36 nmi southwest of Site 884 (see "Background and Scientific Objectives"), where basement rocks were recovered from Detroit Seamount during ODP Leg 145 (Rea, Basov, Janecek, Palmer-Julson, et al., 1993). Drilling at Site 883 (Hole 883E: penetration = 37.8 m; recovery = 63% and Hole 883F: penetration = 26.7 m; recovery = 41%) indicated that the uppermost carapace of the volcanic basement consists of altered plagioclase-phyric pillow basalt and massive basalt lavas. At Site 884 basalt lava flows from Hole 884E (penetration = 87 m; recovery = 66.5%) and 40 cm of altered aphyric pillow basalt from the last core of Hole 884B were recovered. The basalt units described from Hole 884E are aphyric to plagioclase-olivine-phyric with plagioclase phenocrysts up to 2.5 cm in size (Rea, Basov, Janecek, Palmer-Julson, et al., 1993). An 40Ar-39Ar age of 81 Ma for Site 884 basalt has been reported by Keller et al. (1995).
At Site 1203 the lava flow units (Table T4) are composed of aphyric to plagioclase-olivine-phyric basalt. Units 14 and 31 contain large plagioclase phenocrysts, similar to the basalt lava units reported from Site 884, but the size of the plagioclase crystals is generally smaller (e.g., Fig. F8). We define unit boundaries primarily on the basis of changes from lava flow to volcaniclastic deposit (e.g., the boundary between Units 8 and 9) (Fig. F9), major changes in lithology of the lavas (e.g., between Units 10 and 11), and the presence of weathered flow tops in the core (e.g., the boundary between Units 23 and 24) (see "Physical Volcanology and Igneous Petrology" in the "Explanatory Notes" chapter) The age of the basement and overlying sediment has been constrained by nannofossil assemblage identification (see "Biostratigraphy"). The age of the sediment immediately above basement is 50.6-51.5 Ma. The age of volcaniclastic Units 4, 7, and 12 has been determined to be 71.6-76 Ma, whereas that of Units 17 and 27 is 75-76 Ma, at least 5 m.y. younger than basement at Site 884 (Keller et al., 1995).
We briefly describe the major lithologies of the basement sequence (see Table T4). More detailed descriptions can be found in the visual core descriptions (see "Site 1203 Core Descriptions"). The major lithologic features of each unit are summarized in Figures F5 and F6 and Tables T4, T5, and T6.
The volcaniclastic units are composed of a range of sediment types, including primary tephra (pyroclastic and hyaloclastite) deposits and their resedimented derivatives, along with clastic and calcareous sediments. We identified the tephra deposits as basalt tuff, lapilli tuff, lapillistone, and breccia, whereas the clastic sediments are made up of vitric siltstones and sandstones that are often intercalated with nannofossil-bearing calcareous siltstones and mudstones. In total, we identified eight scoria fall and four hyaloclastite (Units 7c-7d, 17, and 25) tephra units along with six resedimented tuff and ten sediment units in the volcaniclastic sequences (Tables T4, T5). The volcaniclastic deposits and associated sediments are generally highly altered because their major component is basalt glass that has been converted to palagonite and/or replaced by other alteration products (green clay and/or zeolite). Pore spaces are typically filled with carbonate or zeolite. Note, however, that unaltered basalt glass is present in the tuffs of Unit 4.
The basalt tuff units are massive and laminated or thin- to medium-bedded tuffs with occasional trains of lapilli scoria. They generally exhibit good to moderate sorting, and the composition of these deposits is dominated by vitric basaltic fragments (Table T5) and minor amounts of plagioclase crystal fragments. The tuffs occasionally exhibit cross-bedding and are cemented by carbonate and zeolites.
The basalt tuffs are divided into three groups on the basis of their clast morphology, texture, and depositional structures: ash fall (Subunits 4c and 4f-4h), resedimented (Subunits 4b, 4d, 4i, 4k, and 4m), and hyaloclastite (Subunits 7c-7d and 31a and Units 17 and 25) tuffs (Table T5). The ash fall tuffs are characterized by massive to normal graded beds consisting almost entirely of highly vesicular basalt glass fragments that often exhibit fluidal morphologies (e.g., Figs. F10, F11). The resedimented tuffs contain a clast population identical to that of the ash fall deposits but feature structures indicative of deposition by traction and origin by resedimentation of primary fall deposits. The hyaloclastite tuffs are distinguished on the basis of their nonvesicular, splinterlike clast population (Table T5).
We interpret the tuffs to be fall deposits formed by subaerial strombolian (explosive) eruptions, but deposited in water, because of the high vesicularity and fluidal shapes of the ash particles along with mature normal size grading and distinctive bedding. The resedimented tuffs represent primary tephra fall deposits that have been subjected to reworking and redeposition either during or shortly after the eruptions. The hydroclastite tuffs are typically associated with pillow lavas and are interpreted to have formed by quench fragmentation in association with subaqueous lava flow emplacement.
These deposits are brownish green to dark green, thinly to thickly bedded, clast-supported basalt lapillistones and lapilli tuffs composed of massive or normally size-graded beds. They are moderately sorted and made up of fine to medium lapilli scoria and occasionally contain clasts >25 mm in size. These deposits consist of highly vesicular (>70%) scoria (80 to >95 modal%), often featuring fluidal outlines (Figs. F12, F13). They also contain occasional cognate basalt lava lithics and blocks as well as armored lapilli, where a lava lithic is wrapped in vesicular glass (Fig. F12B). These units are cemented by zeolites (analcite and natrolite?) or carbonate. On the basis of the overall clast texture and depositional structures, along with the preponderance of delicate and highly vesicular scoria clasts, we identify these units as scoria fall deposits. The overall coarseness of the deposits suggests proximity to source because meter-thick lapilli scoria fall deposits are typically found within 5 km of their source vents (Cas and Wright, 1987; Thordarson and Self, 1993).
We recognize three breccias in the basement sequence. Two of these are Units 13 and 28, whereas the third forms the brecciated top of basalt Unit 21. All breccia deposits are massive, poorly sorted, and matrix supported, where the coarse fraction (20-125 mm) is composed of angular basalt lava lithics and blocks of basalt identical to those of the underlying lava unit (Fig. F14). In addition, a 20-cm interval at the top of Unit 28 consists of brownish white lapilli breccia, composed of 5- to 30-mm angular basalt and plagioclase-phyric dacite(?) lithic fragments that feature a felsic groundmass texture (i.e., feldspar and quartz after devitrified glass).
The matrix composition is variable. In Unit 13 and at the top of Unit 21, the matrix consists of 1- to 8-mm angular particles composed of sparsely vesicular, aphyric to highly plagioclase-phyric basalt glass and lava fragments with variolitic groundmass textures. The phenocrysts in the matrix fragments are identical to those found in the larger breccia clasts, hence indicating a monomict clast population. However, the color of the matrix fragments ranges from brownish gray to dark greenish gray to dark gray because of the differences in the degree and type of alteration that results from variations in the igneous groundmass textures of the clasts. These breccias are cemented by white carbonate (Fig. F14). In the lower part of Unit 28, the breccia contains small aphyric lava toes (4-7 cm thick) and pieces broken from larger basalt lava lobes resting in a very fine grained calcareous matrix that contains a well-preserved specimen of coralline red algae (Fig. F15). The lithology of the basalt toes and fragments is identical to that of the underlying lava. However, these lithic clasts show no indication of being produced by quenched fragmentation (i.e., jigsaw-fit texture or evidence for thermally induced disintegration of glassy lobe margins). Also, the calcareous matrix does not appear to have experienced any heating, as would be expected if it was invaded by hot lava.
We interpret Unit 13 and the brecciated interval on top of lava Unit 21 to be hyaloclastite breccia produced by quench fragmentation of the underlying pillow lavas. The presence of calcareous matrix in the basalt lithic breccia in Unit 28 indicates subaqueous deposition, although the exact mode of formation remains ambiguous. However, we regard a hyaloclastite origin for Unit 28 breccia to be unlikely because there is no direct evidence in the core for interaction between hot lava and water-logged sediment (e.g., Cas and Wright, 1987). Unlike the other breccia units, basalt + dacite(?), the origin of the dacite lithic breccia at the top of Unit 28 shows clear evidence of provenance from two lithologic sources, but its origin is unknown at this time.
These units are sequences of poorly to moderately sorted, laminated to thinly bedded vitric sediments that typically exhibit horizontal planar bedding with alternating beds of siltstone and sandstone. Most common colors are greenish and bluish grays and dark grays; coarser beds are typically lighter in color. Normally graded or cross-stratified beds and fine-scale laminations are common. Several intervals are variably disturbed by slumping and loading, as evidenced by an abundance of soft-sediment deformation structures (Table T5). These sediments are made up of variable proportions of highly to completely altered basaltic glass, siliciclastic fragments, and detrital carbonate. Clasts are subrounded to angular.
The major lithologic features of each unit are summarized in Figures F5 and F6 and Table T4. Lavas in the upper part of the basement sequence (i.e., Cores 197-1203A-17R to 31R) generally show well-developed pillow structures with low vesicularity (typically <5%), whereas pahoehoe lava is more common in the lower part of the sequence (i.e., Cores 197-1203A-41R to 68R) and where the lavas exhibit a marked increase in vesicularity (15%-35%). Glassy lobe margins are present in all basalt units except Units 6 and 24. Unaltered glass is present in the chilled lobe margins (0.2-1.2 cm thick) throughout the sequence (e.g., Figs. F7, F11, F16, F17), whereas in the lower part of the sequence glassy lobe margins are bright green and highly to completely altered (Fig. F17B) (see "Alteration and Weathering"). They are identical to lobe margins that characterize pahoehoe and pillow lavas (Self et al., 1998).
Olivine-rich zones containing >10 modal% olivine are present in Units 11 and 16. We also note large plagioclase phenocrysts in Units 14 and 31 (Fig. F8). Phenocryst size is variable; plagioclase crystals and glomerocrysts in Units 14 and 31 reach 1.3 cm in length but are generally <6 mm in size elsewhere. On rare occasions the olivine phenocrysts have survived postemplacement alteration, but most commonly they are present as olivine pseudomorphs of carbonate, talc, serpentine, and/or Fe oxyhydroxide. Most plagioclase phenocrysts show evidence of incipient alteration to sericite, and interstitial groundmass glass (mesostasis) is partly to completely devitrified. Vesicles are variably filled with secondary minerals, of which calcite is the most common, but zeolite and secondary sulfide minerals are present in vesicles from the lower part of the basement section (see "Alteration and Weathering").
The volcanic succession at Site 1203 is composed of 18 lava units (representing at least 14 eruption units [i.e., lava flow fields]) and 14 volcaniclastic units. As described above, the latter lithologies consist of primary and resedimented basaltic tephra deposits, as well as clastic and calcareous sediments rich in volcanic components. Here we present more detailed descriptions of the internal architecture of the lava units at Site 1203 and the observed lithofacies associations.
The lava flow units at Site 1203 range from tholeiitic to alkalic basalt compositions (see "Geochemistry") and are 1.9 to ~63 m in (cored) thickness. Because of very good core recovery through the lava units (average = 69% and range = 40%-95% on the basis of curated thicknesses), we obtained comprehensive information about the internal architecture and the characteristic lobe structure of individual units. Key features of the lava units are listed in Table T6, and the main characteristics of each lava type are summarized below. On the basis of this information, we identified three lava flow types in the Site 1203 basement sequence. These lava flow types are pillow, compound pahoehoe, and simple pahoehoe lavas. A total of five lava units are categorized as "simple" lavas because in the core section they are typically composed of a single lobe (i.e., cooling unit). They range in thickness from ~2 to 10 m, comprise ~8% of the lava sequence, and have internal architecture and lobe margins consistent with that of inflated pahoehoe lobes in flood lava flow fields (Thordarson and Self, 1998). A total of 13 units are categorized as compound lavas (i.e., lava flow or an eruption unit consisting of multiple lobes) and are divided into pillow and pahoehoe lavas based on the characteristic internal architecture of their lobes (Table T6). However, Units 20 and 21 appear to be hybrid pillow-pahoehoe units. Together, the compound lavas make up ~91% of the lava sequence.
The pillow lava units are all tholeiites (see "Geochemistry") and are more abundant in the upper half of the Site 1203 succession (Figs. F5, F6; Table T6). They range in thickness from ~8 to 25 m and comprise ~24% of the lava flow units. The pillow lava lobes are bounded by smooth glassy lobe margins and range in thickness from 8 to 164 cm (average = 55 cm) (see Fig. F18). The lobe thickness distribution is skewed toward the smaller lobe sizes (see Fig. F19).
The distinguishing features of the pillow lobes are as follows. Thin (<10 mm), sparsely to nonvesicular glassy lobe margins are now generally altered to yellow-brown palagonite (Fig. F20), with occasional domains of sideromelane glass several millimeters across. The lava closest to the lobe margins is without exception defined by a variolitic groundmass texture and features a centimeter-thick microvesicular band with small (<3 mm wide and <10 mm long) tube- or drop-shaped vesicles. The lobe interiors are typically fine grained and nonvesicular and exhibit a distinct cubelike cooling joint pattern (Fig. F18).
The presence of a variolitic texture and cubelike jointing in the pillow lavas is indicative of high cooling rates and emplacement under water. The low vesicularity of the pillow lobes can be attributed to either suppressed degassing at great water depths (
1000 m), due to high external pressures, or to arresting of gas exsolution of partly degassed lava as it flows from land into water. We favor the latter interpretation because the lithofacies associations in the Site 1203 succession suggest a relatively shallow water depth (see below).
The compound pahoehoe lava flow units except for Unit 19 are all alkali basalt (see "Geochemistry") and are confined to the lower half of the Site 1203 basement sequence (Figs. F5, F6; Table T6). They vary in thickness from ~8 to 64 m and account for 50% of the total thickness of lava in the succession. The thickness of lobes in the compound pahoehoe ranges from 9 to 252 cm (average = 56 cm), and the lobe thickness distribution is strongly skewed toward the smaller lobe sizes (Fig. F19). Individual lobes are defined by smooth glassy lobe margins (Figs. F16, F17).
The distinguishing features of the compound pahoehoe lobes are as follows. The glass at lobe margins is typically highly vesicular and completely altered to yellow to dark brown palagonite, clay, and/or zeolite minerals. The groundmass of the lava in the vicinity of the lobe margins is typically aphanitic (hypohyaline to hypocystalline) with an intersertal texture, whereas the lobe interiors are fine grained and characterized by an intergranular texture. Most commonly, the lobes have highly vesicular upper and lower crusts and sparsely to nonvesicular lobe interiors, although some of the smaller lobes are moderately to highly vesicular throughout (Fig. F21). They often feature small pipe vesicles at the lobe base and small vesicle cylinders along with horizontal vesicle sheets and irregular pockets of vesicular segregated material extending up through the massive lobe interior.
The internal architecture and vesicularity of the compound pahoehoe lavas at Site 1203 are very similar to those found in modern pahoehoe flow fields in Hawaii and elsewhere (e.g., Wilmoth and Walker, 1993; Mattox et al., 1993; Hon et al., 1994), and, therefore, we interpret these to be lavas emplaced in an subaerial environment.
The hybrid pillow-pahoehoe lava units are tholeiitic in composition (see "Geochemistry") and are present in the Hole 1203A core where the transition from subaerial pahoehoe- to subaqueous pillow lava-dominated succession occurs (Fig. F5; Table T4). The volcanic architecture of these hybrid lavas is transitional between that of the pillow and compound pahoehoe and indicates emplacement in relatively shallow water. Their characteristic features are as follows.
They consist of multiple decimeter- to meter-thick pillowlike lobes (Figs. F16, F22), with typical lobe thicknesses between 20 and 200 cm (average = 57 cm). As in other compound lavas at Site 1203, the lobe thickness distribution is skewed toward smaller lobe sizes (Fig. F19). The lobes are defined by sparsely vesicular glassy (normally altered to palagonite) lobe margins (Fig. F16). As in the pillow lobes described above, the lava closest to the lobe margins is characterized by variolitic groundmass texture, containing a centimeter-thick microvesicular band and small pipe vesicles. In the lobe interior, the lava is fine grained and often features a cubelike joint pattern indicative of water-enhanced cooling (Fig. F22). The lobes of these hybrid lavas are typically finely vesicular, and vesicle abundance is highly variable (range = <5%-30%). However, the vesiculation pattern is distinctive and different from that found in pillow and pahoehoe lavas. The lava itself is usually sparsely to nonvesicular (<5 modal%), but it contains domains (i.e., vesicle cylinders, horizontal vesicle sheets, and irregular pockets) filled with highly vesicular (up to 40 modal%) segregated material (Fig. F22). This vesiculation pattern implies that the water pressure at the site of emplacement was high enough to suppress primary (first) vesiculation during initial stages of lobe emplacement but low enough to allow secondary vesiculation (i.e., "second boiling") during melt segregation. Typically, such melt segregation in basalt lavas is driven by volatile enrichment (mainly H2O; increase by a factor of 1.5-2) in residual melts after ~35%-45% crystallization (Goff, 1996; Self et al., 1998; Thordarson and Self, 1998). If the volatile content of the lava at the point of deposition is known, this vesiculation pattern can be used to constrain the water depth at the time of emplacement.
A total of five units are identified as simple pahoehoe lava. They are all tholeiitic basalt (see "Geochemistry") and are more abundant in the upper half of the Site 1203 succession (Figs. F5, F6; Table T6). They range in thickness from ~2 to 10 m and account for 8% of the total thickness of lava flows in the succession.
We were able to accurately document the volcanic architecture of Units 5, 11, and 16 because of very good recovery (Table T6). The upper and lower lobe boundaries of these units were recovered and are in all cases smooth glassy pahoehoe surfaces. Units 11 (~4.6 m thick) and 16 (~10.4 m thick) feature a distinct threefold division into vesicular upper crust, massive lobe interior, and vesicular lower crust (Fig. F23), which is the diagnostic structure of inflated pahoehoe sheet lobes (e.g., Thordarson and Self, 1998). In both lobes, the upper and lower crusts are characterized by relatively high vesicularity (10-30 modal%), except for the top 50 cm of the upper crust, where the lava is finely vesicular (diameter < 1 mm) and characterized by unusually lower vesicularity (
8 modal%). The pahoehoe sheet lobe of Unit 5 is ~2 m thick and sparsely vesicular (5 modal%) throughout.
Despite reasonable core recovery (49% and 68%) (see Table T6), only parts of the sheet lobes that make up Units 6 and 24 (i.e., vesicular basal crust, the massive lobe interior including horizontal vesicle sheets, and small portions of the vesicular upper crusts) were recovered. Partial preservation is the most likely explanation of incomplete recovery, suggesting that these lavas were subjected to significant erosion prior to burial by later formations.
The sheet lobes of Units 5, 11, and 16 all have conformable contacts with water-lain volcaniclastic sediments, which implies subaqueous flow emplacement. The unusually low vesicularity along with very small vesicle size in the Unit 5 sheet lobe as well as in the top 0.5 m of the vesicular upper crust of the lobes of Units 11 and 16 may indicate suppression of volatile exsolution resulting from water-enhanced cooling of the lava. In the case of the two latter units, vesiculation appears to have progressed in a normal fashion when the upper crust had thickened enough to seal the molten lava interior from its surroundings, whereas the Unit 5 sheet lobe never achieved such insulation because of its small thickness.
The compound pahoehoe and the pahoehoe sheet lobes at Site 1203 commonly exhibit a threefold division into vesicular upper crust, massive lobe interior, and vesicular lower crust and contain segregation structures such as pipe vesicles, vesicle cylinders, and horizontal vesicular sheets (Table T6), which are the characteristic architecture of subaerial pahoehoe flows formed by endogenous emplacement (i.e., transport of lava under an insulating stationary crust and growth by lava inflation).
The similarity between compound pahoehoe and pillow lavas in terms of lobe sizes and overall morphology, along with the presence of segregation structures in the pillowlike lobes of Units 20 and 21, is consistent with the notion that the emplacement mechanism of these two lava types is essentially the same. They differ only in the environment of emplacement and cooling rates.
The most striking lithofacies change at Site 1203 is the shift from the dominance of compound pahoehoe lava flows in the lower part of the succession to pillow lava flows in the upper part. Intuitively, this implies a change from a subaerial to a submarine setting for the Site 1203 environment, which most likely was caused by gradual subsidence on a regional scale. The presence of hybrid pillow-pahoehoe lavas along with a more frequent occurrence of hyaloclastite tuffs and breccias in the middle part of the succession as well as thicker volcaniclastic sequences in the upper part are compatible with this inference (Table T4; see also Fig. F56 in the "Leg 197 Summary" chapter). However, the presence of calcareous interbeds in the lower part of the succession as well as the presence of thick primary tephra fall deposits in the upper part shows that land and sea were never far apart.
In thin section we characterize the volcaniclastic lithologies by the abundance of vitric particles (fresh and altered sideromelane) along with subordinate amounts of lithic fragments (both juvenile and cognate) and occasional crystal fragments. Vitric particles typically comprise up to 99% of the clast population.
In the primary basalt tephra fall deposits (e.g., Units 4f-4h, 9, and 22b), highly vesicular scoria with fluidal shapes are by far the most abundant clast type, exhibiting delicate morphologies, highly cuspate clasts, and containing occasional plagioclase laths (Fig. F13). These units contain tephra clasts (Fig. F10) that have an elongate and twisted outline and numerous elliptical vesicles. The presence of these characteristics is consistent with our interpretation that these deposits were produced by fire-fountaining or explosive activity in subaerial Hawaiian or Strombolian eruptions (Fisher and Schmincke, 1984).
The vesicles in tephra clasts are commonly filled with radiating zeolite minerals (natrolite or mesolite?) or analcite. Hairline cooling fractures often penetrate the scoria clasts and are formed during quenching of the scoria clasts during transport through the air or upon contact with water when still hot. The preservation of such delicate structures demonstrates that these deposits have not been reworked after transport and deposition and adds support to our interpretation that these deposits are primary tephra fall deposits. The lithic population is composed mainly of cognate crystalline basaltic fragments.
The hyaloclastite tuffs (Table T4) are characterized by abundant nonvesicular platy or splinterlike clasts. The lack of vesicles and the elongate shardlike morphologies are consistent with our interpretation that these tuffs were produced by nonexplosive fragmentation when the underlying lava flowed into water (Table T5).
Inspection of the fine-grained matrix in the basalt-lithic breccia that comprises the lower part of Unit 28 shows that it consists of partly recrystallized calcareous mudstone containing dispersed basalt glass and plagioclase fragments and one specimen of fossilized coralline red algae (Fig. F15). This observation is consistent with our interpretation of subaqueous deposition for this breccia unit. Furthermore, the composition of the volcaniclastic sandstone in Unit 27 and the breccia interval at the top of Unit 28 appears to differ from the other clastic sedimentary units. It contains a significant amount of dacite(?) lava lithic fragments (mixed with basalt fragments) in which feldspar phenocrysts are imbedded in felsic groundmass. The presence of dacite lithic fragments suggests influx from an external sediment source.
The lava flow units are aphyric to olivine- and/or plagioclase-phyric. They consist of variable amounts of glomerocrystic plagioclase and/or olivine in a groundmass of plagioclase, clinopyroxene, titanomagnetite, with varying abundance of mesostasis consisting of glass and postemplacement alteration products. Plagioclase is the most abundant phenocryst phase, usually present as glomerocrysts forming aggregates up to 10 mm in size; individual phenocrysts reach up to 1.3 cm in size in Units 14 and 31. Individual crystals within the glomerocrysts are typically oscillatory zoned (Fig. F24), fractured, and, occasionally, glass inclusions are present (Fig. F25). Some phenocrysts have rounded or embayed margins (Fig. F26), indicating disequilibrium. Olivine, where present, is also often found as glomerocrysts (Fig. F27). It is typically pseudomorphed by a combination of carbonates (calcite and, rarely, magnesite and siderite), iddingsite, and green/brown clay or Fe oxyhydroxide (see "Alteration and Weathering"); it is recognized by its equant form and preservation of the characteristic fracture patterns. Occasionally, it is skeletal in form (8:1 aspect ratio) and up to 3 mm in length (Fig. F28). Unaltered olivine is present in several flows, where it comprises up to 35 modal% of the rock in certain thin sections (e.g., Fig. F29) and commonly contains rare inclusions of glass and Cr spinel (Fig. F30).
Toward lobe margins the lavas show evidence of quenching where groundmass plagioclase laths are skeletal and are surrounded by acicular parallel arrays of clinopyroxene (usually in optical continuity) in a subvariolitic texture, along with dendritic titanomagnetite (Fig. F31). Immediately adjacent to lobe margins a well-developed variolitic texture is present, with spherical structures formed from plumose clinopyroxene radiating around minute plagioclase microlites (Fig. F20). This is indicative of high cooling rates (quenching) of the lava. Unaltered glass is often present in the areas between varioles. In the larger lobe interiors and massive basalt units, intergranular to subophitic textures predominate (e.g., Fig. F32).
Vesicle-rich segregations are common in the interiors of some lava flows (e.g., Units 20, 21, 23, and 26). They consist of usually vertically oriented concentrations of pipe vesicles and vesicle cylinders partially to wholly filled with vesicular segregated material. These are generally finer grained than the surrounding groundmass and are composed of comb-textured acicular clinopyroxene and abundant acicular and dendritic titanomagnetite, along with interstitial devitrified glass (Fig. F33). Plagioclase is typically sparse in these regions. Titanomagnetite can locally comprise up to 20% of the segregated material in, for example, Units 23, 26, 29, and 30 (e.g., Figs. F33, F34).
The opaque mineralogy of the Site 1203 basalt is dominated by oxide minerals, primarily titanomagnetite, with minor primary pentlandite. Secondary pyrite is present in some sections, but this is restricted to mesostasis, glass-rich portions of the basalt groundmass, which are most susceptible to alteration and replacement, or veins and vesicles. The sulfides are generally extremely small (<0.01 mm in size) blebs included in primary phases, except in the olivine-rich zones of Units 11 and 16, where pentlandite comprises up to 1% of the modal mineralogy. The presence of pentlandite indicates that subsolidus reequilibration has occurred (Augustithis, 1979).
The magnetic characteristics of basalt are recorded in the micron-sized oxide phases, predominantly magnetite and titanomagnetite. Titanomagnetite is a solid solution between ulvöspinel (Fe2TiO4) and magnetite (Fe3O4), and it can be oxidized by two mechanisms (Haggerty, 1991): (1) at low pressure and <600°C to produce cation-deficient titanomagnetite or metastable titanomaghemite that can, on occasion, be subsequently converted to members of the hematite-ilmenite series and (2) at low to moderate pressures and >600°C with the direct formation of minerals from the hematite-ilmenite solid-solution series. In the Site 1203 basalt units, we see trellis-type ilmenite "oxidation exsolution" (cf. Buddington and Lindsley, 1964) and maghemite/titanomaghemite lamellae in the titanomagnetites (Figs. F35, F36). Note that in this and other site chapters, we do not make the distinction between maghemite and titanomaghemite, as they are indistinguishable using reflected-light microscopy.
The secondary mineralogy demonstrates that the basement sequence at Site 1203 has experienced oxidative alteration (see "Alteration and Weathering"); the textures of the primary titanomagnetite support this and indicate that oxidation has occurred both above and below 600°C. For example, in several basalt lava units micrometer-sized and larger titanomagnetite grains visible under reflected-light microscopy have been completely replaced by maghemite (e.g., Units 1, 3, and 29) (Fig. F37), and in others, ilmenite oxidation exsolution lamellae are present (Fig. F35). We note partial alteration to maghemite in Units 29 and 30 (Fig. F38). Titanomagnetite is, for the most part, unaltered in Units 6, 11, 14, 16, 18-21, 23, and 24 (e.g., Fig. F39). The condition of titanomagnetite in the Site 1203 basalt units is summarized in Table T7.
In the olivine-rich sections of Units 11 and 16, Cr spinel is present as inclusions in olivine (Fig. F40). Where the Cr spinel has been in contact with the magma, it has either reacted completely or partially to titanomagnetite (e.g., Fig. F41) but remains unaffected where armored by olivine. The presence of Cr spinel in this basalt is significant, as the Cr content of such spinels can be used to estimate melt compositions as well as fractional crystallization processes (cf. Allan et al., 1988, 1989; Sack and Ghiorso, 1991).
We conducted major and trace element analyses by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) (see "Physical Volcanology and Igneous Petrology" in the "Explanatory Notes" chapter) on 26 samples from the basalt units, 1 basaltic clast each from volcaniclastic Units 13 and 22, and 1 whole-rock volcaniclastite (Table T8). In an alkalis (Na2O + K2O) vs. SiO2 classification plot, samples from the lava flow units range from tholeiitic to alkalic basalt (Fig. F42). The volcaniclastite sample has a very high Na2O + K2O (6 wt%) that we interpret to be a secondary feature reflecting the altered nature of the basalt glass shards, ash, and lapilli particles. Hence, for the basalt samples we evaluate the compositional effects of postmagmatic alteration processes. First, loss on ignition (LOI) in Site 1203 basalt generally exceeds 1 wt% and ranges from 0.86 to 6.7 wt%, except for the volcaniclastite with 11.7 wt% (Table T8), whereas recent Hawaiian lavas have LOI < 1 wt% (e.g., Rhodes, 1996). All of the Site 1203 basalt samples that extend well into the alkalic field have LOI > 2 wt% (Fig. F42). Some but not all of these alkalic basalt samples have a relatively high K2O content (Table T8); the sum of Na2O + K2O in these basalt samples overlaps with the range in basalt from Site 883 (Fig. F42). The mobility of K2O is clearly shown in Figure F43, where abundance of the incompatible elements Na, Ti, P, and Zr are highly correlated but K abundances are scattered, especially to high values. The Site 1203 samples with the highest K2O contents (1.3-1.8 wt%) (Table T8) are from Unit 20, with high LOI values (5.2 wt%), and the three samples from Units 29 and 30 (LOI = 4.0-6.7 wt%) that contain a high proportion of clay-filled vesicles.
A similar result is seen for data from Site 883 Detroit Seamount basalt samples (Fig. F43). There is also evidence for CaO loss during alteration, most conspicuously in the volcaniclastite sample (2 wt%) (Table T8). Despite the evidence for K2O mobility during alteration, we conclude that the Site 1203 lava compositions range from alkalic to tholeiitic. In particular, samples from flow Units 23, 26, 29, and 30, which have variable K2O because of alteration, are all characterized by relatively high abundances of Na, Zr, Ti, and P (Fig. F43) and they are truly alkalic basalt. The alkalic basalt units range in thickness from 8 to 63 m. These units are from the lowermost 160 m of the ~450 m of basement penetration. Units 23 (63 m) and 26 (41 m) are the thickest flow units recovered (Table T4); both units are dominantly aphyric, highly vesicular basalt. In detail, the Ti vs. Zr plot shows different trends for basalt from Units 29 and 30 compared to that defined by basalt from Units 23 and 26. Note that Units 29 and 30 are similar; the boundary between them was defined by the presence of a 127-cm section of breccia in Section 197-1203-66R-1, a change from aphyric (Unit 29) to sparsely plagioclase-phyric (Unit 30) and a slight decrease in TiO2 and Zr content.
Another important feature of basalt from Units 23, 26, 29, and 30 is that it has relatively constant Sr and Ba contents despite a Zr content that ranges from 200 to 315 ppm (Fig. F43). Unlike tholeiitic basalt samples from Detroit Seamount, Sr and Ba were not highly incompatible elements during the petrogenesis of these alkalic lavas. Four samples from Unit 23 were analyzed (Table T4). Relative to the other three samples, Sample 197-1203A-52R-6, 12-14 cm, is enriched in the incompatible elements Ti, Zr, P, Y, and Ba by ~20%, whereas Sr contents are similar. These enrichments can be explained by the abundance of highly evolved segregated melt in this sample (Fig. F44). The absence of Sr enrichment is consistent with this evolved melt being in equilibrium with plagioclase.
Two of the analyzed samples are basalt clasts in volcaniclastic Units 13 and 22. The compositions of these clasts are not identical to the underlying basalt lava units (Table T8), but they are transitional in composition like other Site 1203 basalt units with <2 wt% LOI (Fig. F42). Two other samples are from olivine-rich portions of Units 11 and 16. They are MgO rich and SiO2 poor relative to other Detroit Seamount basalt samples (Table T8; Fig. F42A), a result of olivine addition. The olivine addition trends are especially clear in plots of Na2O vs. Zr and MgO vs. Ni, Sc, CaO, and Sr (Figs. F43, F45).
Although the high Ni content of the two picritic Detroit Seamount lavas is consistent with olivine accumulation, the Ni abundances of the remaining basalt (50-280 ppm) do not correlate with MgO content (Fig. F45). Except for Site 884 lavas, Sc abundance from basalt recovered from Detroit Seamount tends to increase with decreasing MgO content, thereby showing that basalt from Sites 883 and 1203 did not experience major fractionation of clinopyroxene. The CaO-MgO panel shows a distinctive field for the alkalic basalt samples at Site 1203; they have relatively low CaO and range to higher MgO contents than the transitional and tholeiitic basalt samples. The Sr-MgO plot shows that at a given MgO content, basalt from each drill site has a different range of Sr content, with the lowest abundances in the Site 884 tholeiitic basalt samples and the highest abundances in the Site 1203 alkalic basalt. At each drill site, Sr and MgO abundances are inversely correlated, suggesting that despite the presence of plagioclase-phyric lavas, plagioclase fractionation was not a major process controlling the lava compositions.
An objective of Leg 197 is to compare the compositions of basalt lavas from the Emperor Seamounts and the Hawaiian Islands. We use Mauna Kea Volcano, Hawaii, for this comparison because the compositions of the dominantly alkalic postshield stage and dominantly tholeiitic shield-stage lavas are well documented (Frey et al., 1990, 1991; Rhodes, 1996). In the classification plot (Fig. F42), Detroit Seamount lavas from Site 884 are within the tholeiitic shield field. Site 883 lavas are in the alkalic field, but Keller et al. (1995) concluded that these may be transitional lavas that plot in the alkalic field because of K2O addition during alteration. Site 1203 lavas range from tholeiitic to alkalic basalt. Although some Detroit Seamount lavas appear similar in major element composition to Hawaiian lavas, those from Sites 883 and 884 differ significantly in abundances of incompatible elements (Keller et al., 2000; M. Regelous et al., unpubl. data). For example, in plots of MgO vs. TiO2, P2O5, Zr, Ba, and Sr at a given MgO content, Detroit Seamount lavas from Site 883, and especially Site 884, are offset to lower incompatible element abundances than the Hawaiian trend defined by lavas from the Mauna Kea and Mauna Loa Volcanoes on the island of Hawaii (Fig. F46). In general, the Site 1203 transitional to tholeiitic lavas fill the gap between lavas from Sites 883 and 884; the Site 1203 alkalic basalt lavas (Units 23, 26, 29, and 30), however, overlap with those of the postshield stage from Mauna Kea Volcano in Ti, Zr, and P, but they have lower Sr and Ba contents (Fig. F46).
An important characteristic of Hawaiian volcanoes is their temporal evolution from preshield-, to shield-, to postshield-, to rejuvenated-stage lavas. Lava compositions and isotopic ratios change systematically with time (e.g., Chen and Frey, 1985; Clague and Dalrymple, 1987; West et al., 1987). Therefore, it is of interest to compare the temporal geochemical variations in extensive volcanic sections through the Emperor Seamounts with sections from Hawaiian volcanoes. Suiko Seamount, an Emperor Seamount drilled during Deep Sea Drilling Project Leg 55, shows a classic Hawaiian transition from older tholeiitic lavas to younger alkalic lavas (Jackson et al., 1980). In contrast, at Detroit Seamount the alkalic lavas, distinguished by a relatively low Ti/Zr value, occur deep in the Site 1203 section (Fig. F47). The Site 1203 alkalic basalt lavas that erupted as compound pahoehoe lavas in subaerial to shallow-marine environments may be analogous to the dominantly alkalic postshield-stage lavas that erupt as Hawaiian volcanoes migrate away from the hotspot. If the overlying tholeiitic to transitional pillow lavas at Site 1203 erupted from the same volcanic center as the alkalic lavas, the only analogous alkalic-to-tholeiitic transition in the Hawaiian Islands is the preshield stage represented by Loihi Seamount (e.g., Clague and Dalrymple, 1987). A scenario that could explain the alkalic-to-tholeiitic transition at Site 1203 is that the drilling penetrated lavas erupted from two distinct volcanic centers that were in different stages of growth. For example, such a lava sequence would be encountered in Hawaii where tholeiitic shield-stage lavas from Mauna Loa Volcano overlie alkalic postshield-stage lavas erupted from Hualalai Volcano. This possibility receives support from the observation that Unit 24, erupted within the alkalic sequence (Units 23, 26, 29, and 30), plots in the tholeiitic field and has a Ti/Zr value similar to the younger tholeiitic lavas.
