IGNEOUS PETROLOGY

We recovered a 337.7-m-thick sequence of volcaniclastic rocks from Hole 1184A with an average recovery of 82.6%. We defined this sequence as Unit II and divided it into five subunits on the basis of changes in sedimentary structures, clast types, and characteristic grain size (see "Lithostratigraphy"). Because of the high proportion of volcanic glass shards, we assigned lithologic names for pyroclastic deposits as defined by Fisher and Schmincke (1984) (see "Igneous Petrology" in the "Explanatory Notes" chapter). Using this nomenclature, Unit II consists of a sequence of poorly sorted vitric and lithic tuffs, lapilli tuffs, and lapillistones.

Vitric and lithic clasts and armored and accretionary lapilli are present throughout Unit II, and coarse tachylite-rich layers form the upper and lower parts of Subunit IIC. The morphology and relative proportion of lithic to vitric clasts (ash to lapilli size) show systematic downhole variation (Fig. F20).

The volcaniclastic rocks consist of six principal components: ash- to lapilli-size lithic clasts and vitric shards, accretionary lapilli, armored lapilli, crystal fragments (plagioclase and clinopyroxene), and matrix and/or cement (Figs. F29, F30). Carbonized wood fragments are a minor component (Fig. F20). Thin section observations (summarized in Table T6) show that the matrix consists of fine-grained vitric and lithic ash, clay, and other alteration minerals cemented by zeolite or calcite. Titanomagnetite and, rarely, minor amounts of sulfide are also found within the matrix. The plagioclase and clinopyroxene crystal fragments are generally unaltered, but the titanomagnetite grains in the matrix show evidence of alteration to maghemite (Fig. F31). Some discrete plagioclase crystals exhibit irregular oscillatory zoning. Glass inclusions are found in the clinopyroxene and plagioclase crystals (Figs. F32, F33).

The lithic clasts in all five subunits of Unit II are composed of basalt, reworked volcaniclastic rock or diabase. They range in size from <1 to ~65 mm and are generally subangular to subround. The largest clasts are reworked volcaniclastic rocks, generally rip-up clasts of tuff and lapilli tuff (Fig. F18). Some rip-up clasts contain red, clay-coated accretionary and armored lapilli (Figs. F18, F34). Some moderately to highly vesicular lithic clasts contain smaller clasts of nonvesicular, aphyric basalt (Fig. F35).

The basalt clasts are predominantly nonvesicular to moderately vesicular and aphyric to sparsely plagioclase and/or clinopyroxene phyric (see Table T6). Scoriaceous basaltic clasts containing plagioclase and clinopyroxene are also present. Thin section examination demonstrates that some of the highly altered basaltic clasts have visible igneous textures, including intersertal, subtrachytic, and subophitic (Figs. F36, F37, F38). Several aphanitic basalt clasts contain clinopyroxene and plagioclase glomerocrysts. Rounded clasts of apparently more evolved basalt (Fig. F39) are distinguished by a greater abundance of plagioclase and titanomagnetite relative to clinopyroxene. In contrast to the aphanitic basalt, the diabase clasts contain unaltered plagioclase, clinopyroxene and titanomagnetite (Figs. F40, F41).

Armored and accretionary lapilli, though generally sparse, are often concentrated in layers 5 cm thick (e.g., Figs. F42, F43). Layers with concentrations of broken armored and accretionary lapilli grade into layers in which these lapilli are unbroken (Fig. F44). Where the armored and accretionary lapilli are broken, they are commonly found in a hydrodynamically stable, convex-upward orientation (Fig. F45), which suggests some reworking (see "Lithostratigraphy"). Both armored and accretionary lapilli have red (oxidized) margins in Subunits IIA and IIC (Fig. F15). Thin sections show that the armored lapilli have nuclei of vesicular basalt, aphanitic basalt, or altered glass (Figs. F46, F47).

The majority of vitric clasts are subangular to angular glass shards, <10 mm in size. The glass is generally blocky and nonvesicular, but minor amounts of sparsely to highly vesicular shards (some with scalloped edges or internal flow textures) are present (Figs. F48, F49, F50). In the upper 250 m of the cored section, glass shards are highly altered and commonly replaced by concentric layers of smectite, celadonite, and zeolite (see "Alteration," Fig. F49). Unaltered glass is generally present below Core 192-1184A-38R. Some glass shards contain small euhedral crystals of plagioclase and clinopyroxene (Fig. F51).

A highly altered, highly vesicular scoria clast type is also present and may represent devitrifed vesicular glass shards. The high degree of alteration gives these clasts the appearance of lithic fragments; as a result, the proportion of vitric to lithic clasts may be underestimated.

The lapillistones forming the upper and lower parts of Subunit IIC have a more compacted texture than the rest of Unit II (Fig. F19). In thin section, the lapillistones contain large amounts of tachylite (glass made nearly opaque by the presence of Fe-Ti oxide microlites) (Fig. F52). Plagioclase laths within the tachylite define a subtrachytic-to-trachytic texture around flattened or elongated lithic clasts (Figs. F53, F54).

We selected six samples of tuff for whole-rock analysis by inductively coupled plasma-atomic emission spectrometry (ICP-AES). The samples fall into two compositional groups on the basis of position in the cored section and the presence or absence of unaltered glass. Unaltered glass is rare above Core 192-1184A-39R but is present from Section 39R-1 to the bottom of the hole. The three samples from above Core 192-1184A-39R (192-1184A-18R-6, 43-45 cm; 31R-1, 140-142 cm; 36R-7, 74-76 cm) have higher total alkali contents than do the other three samples (Fig. F55). The apparent alkalic character of the upper three samples is probably a result of alteration rather than an original magmatic signature, because their incompatible element abundances are low (Table T7) compared to abundances in alkalic basalts on Malaita and Santa Isabel (Tejada et al., 1996). All six samples have relatively high weight loss on ignition (5.8-7.4 wt%) (Table T7).

We find no downhole change in the types of vitric and lithic clasts that are present throughout Unit II, except that tachylite and red, oxidized lithic clasts are much more abundant in Subunit IIC. Differences in the relative abundance of lithic to vitric clasts correlate broadly with the subunit boundaries (Fig. F20), and there is a peak in the proportion of lithic clasts within Subunit IIC. Armored and accretionary lapilli are also present throughout Unit II but are generally least abundant in Subunit IID. This subunit comprises vitric lithic tuffs with inclined layers, suggesting current reworking, perhaps above storm-wave base (see "Lithostratigraphy"). Reworking of these tuffs could have destroyed accretionary lapilli that may originally have been present. The upper 20 m of Subunit IIA may have been extensively reworked (see "Biostratigraphy"). However, Subunit IIA contains the highest abundance of accretionary lapilli and the highest proportion of angular vitric clasts, indicating that little reworking occurred during deposition of the lower part of this subunit.

Subunit IIC is defined by the presence of compacted tachylite-rich layers, a higher ratio of lithic to vitric clasts, abundant red (oxidized) clasts, and high magnetic susceptibility and paleomagnetic intensity (see "Paleomagnetism" and "Physical Properties").

The volcaniclastic sequence recovered from Hole 1184A represents several broadly similar eruptive events. Significantly, several features of the volcaniclastic rocks are consistent with eruption in a shallow-water to emergent setting.

The abundance of blocky glass shards suggests that the volcaniclastic sequence formed through the explosive interaction of basaltic magma with water in hydroclastic eruptions. The accretionary and armored lapilli indicate the presence of steam-rich subaerial eruption columns. Such eruption columns are typical of hydroclastic volcanic activity and generally originate from vents at water depths of <300 m (e.g., Fisher and Schmincke, 1984). Alternatively, the eruption could have occurred on land, with ground or surface water providing the moisture necessary for formation of accretionary lapilli.

The presence of tachylite clasts and red (oxidized) lithic clasts in the volcaniclastic succession strongly suggests a subaerial setting for at least some of the eruptions. Tachylite clasts are most abundant in Subunit IIC, but vesicular tachylite clasts are sparse throughout the succession. Tachylite is generally thought to form by rapid cooling of basaltic magma droplets in air or in littoral environments where subaerial lava flows quench and fragment as they enter the sea (e.g., Fisher and Schmincke, 1984). The local existence of land is supported by the presence of carbonized wood fragments at several intervals that coincide with subunit boundaries (see "Lithostratigraphy"), suggesting that the wood fragments accumulated at the sites of deposition between major depositional and/or eruptive events.

The presence of rip-up clasts and bands of convex-upward accretionary lapilli fragments (e.g., Fig. F45) suggests some reworking of the volcaniclastic sediments, which is consistent with the sedimentological observations for Unit II (see "Lithostratigraphy"). The absence of blocks and bombs suggests that the sequence did not originate close to the eruptive center, but the fact that accretionary lapilli are abundant suggests relatively short transport distances during reworking. Estimating with any certainty the distance to the eruptive vent(s) is impossible, but the small positive free-air gravity anomalies and bathymetric highs around Site 1184 suggest three possible sources (see Fig. F3; also see "Background and Objectives").

One of the objectives for Site 1184 was to determine the nature of the basement on the eastern salient of the Ontong Java Plateau (see "Background and Objectives"). The recovery of a sequence of volcaniclastic sediments containing a high proportion of vitric clasts and accretionary and armored lapilli indicates that the eastern salient was the site of explosive hydroclastic eruptions in a shallow marine to emergent environment. The presence of volcaniclastic rocks with intercalated wood fragments at Site 1184 provides the first clear evidence of emergent, explosive volcanism on the plateau. The inferred middle Eocene age for the volcaniclastic sequence (see "Biostratigraphy") is nearly the same as that of the 44-Ma alkalic Maramasike Formation on the island of Malaita (Tejada et al., 1996) and broadly comparable to that of the most recent eruptive event producing tholeiitic basalts on the island of San Cristobal (Makira) at 36 ± 3 Ma (Birkhold-VanDyke et al., 1996).

Immobile-element abundances in the six samples analyzed by ICP-AES are similar to those of basalt from Site 1185 (lower group), Site 1186, and Units C-G at Site 807. One sample from Hole 1184A plots on the border of the Singgalo Formation field, defined by basalt flows from Malaita (Fig. F56), and is similar to some 90-Ma Site 803 basalts. This pattern is less clear on a plot of TiO₂ vs. Mg# (Fig. F57), but the MgO contents may have been affected by alteration. The three samples from below Core 192-1184A-39R have similar TiO₂ contents to those of the Kwaimbaita Formation basalts, Units C-G from Site 807, and the lavas from Sites 1183, 1185 (lower group), and 1186, but Cr (and Ni) abundances are higher (Fig. F58).

The inferred middle Eocene age of eruption for volcaniclastic rocks at Site 1184, coupled with magmatism of a broadly similar age on Malaita and San Cristobal, suggests that magmatic activity was widespread on the Ontong Java Plateau during the Paleogene. The fact that the Site 1184 Unit II bulk samples are tholeiitic and have incompatible element abundances similar to Ontong Java Plateau basalts erupted at ~122 Ma implies that similar mantle sources, melting to similarly high degrees (Neal et al., 1997), were available as long as 80 m.y. after the initial plateau magmatism.