IODP Proceedings    Volume contents     Search

doi:10.2204/iodp.proc.324.104.2010

Igneous petrology

Stratigraphic unit description and volcanology

In this site report the stratigraphic units are given Roman numbering (stratigraphic Units I–XVI), whereas the small-scale lithologic units identified during core description are given Arabic numbering (lithologic Units 1–82). We identified these lithologic units on the basis of sedimentological changes or, in the case of volcanologic units, criteria including the presence of chilled margins or contacts, identifiable flow tops, vesicle distribution, and the occurrence of intercalated volcaniclastic or sedimentary horizons. When considered together these criteria define the various sizes of individual volcanic inflation units. Accordingly, in the volcanic context, the term "stratigraphic unit" is used to combine smaller consecutive inflation units of similar character into lava packages or single eruptive sequences (see "Igneous petrology" in the "Methods" chapter).

Hole U1347A was drilled on the upper southeastern flank of Shatsky Rise on Tamu Massif (Fig. F1). The top of the drilled succession (Figs. F15, F16) commences with ~18 m of Lower Cretaceous sediment, including a chert-chalk series (Unit I), radiolarian-bearing glauconite-rich sandstones, and silicified limestones (Unit II) and a sequence of laminated volcaniclastic glauconite-rich sand- and siltstones (Unit III). Igneous basement was reached 157.6 mbsf in Section 324-U1347A-11R-1, after which drilling continued for 159.5 m into a volcanic succession dominated by five massive (~8–23 m thick) basalt flows and pillow basalt units. The massive flows appear in two eruptive sequences separated by a volcanically more complex ~75 m thick sequence of pillow basalt with individual inflation units ranging in size from 0.2 to 5.6 m. Based on the macroscopic core descriptions the volcanic units are thus divided into three groups (Fig. F15). In order of increasing depth, these are:

  1. Group 1 is the top basement sequence and consists of four massive lava flows (stratigraphic Units IV, V, VII, and IX), which together have a combined thickness of ~60 m.

  2. Group 2 comprises a ~75 m thick lava stack sequence of pillow basalt inflation units (Units X, XII, and XIV) interspersed with thicker, more massive basalt units (typically 1–2 and rarely >3 m).

  3. Group 3 is a succession of very thick (~23 m), massive lava flows (Units XV and XVI), located toward the base of the hole. The hole terminated ~2 m into the top of the second massive lava flow in this group; its true thickness is therefore unknown.

The core also revealed that the volcanic units are interbedded by four ~5 m thick, mostly undeformed sedimentary intervals (Units VI, VIII, XI, and XIII) (see "Sedimentology" for a more detailed description of these interbeds). A fifth interbed in Unit X was not recovered but has since been recognized in the downhole logging measurements as a ~5 m sedimentary interval of low resistivity and high natural gamma readings (see "Downhole Logging" for more details). The following is a description of the volcanic units and a preliminary assessment of the physical volcanology of the eruptive sequences drilled at Site U1347.

Group 1

First massive basalt flow (stratigraphic Unit IV, lithologic Unit 4)

Unit IV is a single 16.6 m thick (19.5 m in downhole logs) massive aphyric basalt flow (recovery = ~70%) with microcrystalline chill zones toward the upper and lower contacts, becoming coarser (fine-grained basaltic texture) toward the core. Thin section analyses of this coarser material reveal microphenocrysts of plagioclase intergrown with clinopyroxene and some traces of (pseudomorphed) olivine. Magnetic susceptibility readings are observed to increase concomitantly with grain size (Fig. F16), with the highest readings toward the lowermost region of Unit IV.

The lower chilled zone is somewhat thinner than the upper and is demarcated by a series of small pipe vesicles. Both the upper and lower contacts of the flow are well preserved, with a large proportion of fresh basaltic glass that surrounds and supports small blocks of basaltic rubble in what may be interpreted as an original brecciated flow top crust (interval 324-U1347A-11R-1, 21–28 cm). Vesicles are concentrated in both the upper chilled zone and uppermost part of the flow core but become sparse or absent in its middle and lower part. However, within the lower part of the flow core vertical arrangements, or zones, of small vesicles (i.e., pipe vesicle structures) are common and typically contain material of slightly darker color, subsequently identified as melt segregations. Geochemical analysis of these segregation materials reveals them to be of slightly different composition to that of the bulk flow (i.e., elevated K2O and Ni and lower MgO, V, and Cr) (Hartley and Thordarson, 2009), whereas diktytaxitic texture, indicative of elevated volatile content (Goff, 1996) is more common in the middle and lower core regions (see "Geochemistry").

Interestingly, preliminary data indicate a magnetic reversal may occur in the top half of the unit, with downhole logs revealing a decrease in resistivity at a similar interval (see "Paleomagnetism" and "Downhole Logging"). If confirmed, these data indicate that this massive flow unit actually represents the incomplete upper and lower portions of two separate inflation units.

Second massive basalt flow (stratigraphic Unit V, lithologic Unit 5)

This second massive flow is 12.5 m thick, has a very similar aphyric character to that of the overlying Unit IV, and also contains microphenocrysts of plagioclase intergrown with clinopyroxene together with traces of pseudomorphed olivine. Abundant fresh volcanic glass is preserved at the upper surface and reveals a pahoehoe-like flow-top structure (Figs. F17, F18). Fragments of radiolarian-rich sediment were also recovered within this surface and have apparently been baked by the lava flow; these are interpreted as part of a flow-top breccia. A good example is shown in Figure F19, where the contact between the lava and sediment clast is glassy and represents a narrow zone (~1 cm) of lava chilling and a profound baking of the mudstone clast. Moreover, a small but significant total natural gamma ray anomaly is apparent in the downhole logs; this most likely represents a thin sediment layer or lens of more than ~10 cm thick intervening between the two flows at this location (see also "Physical properties" and "Downhole Logging").

Beneath the uppermost ~1 m thick chilled zone, the top ~3 m of the second massive flow is highly vesicular (~15%–20%). However, because recovery in Unit V is poor, details of the internal structure of this massive basalt flow cannot be recorded. Again, on the basis of downhole logging data, this unit may be divided into three zones, with a less fractured and more massive middle zone yielding higher resistivity readings apparently bracketed by regions of lower resistivity. The lower contact of the second massive flow is defined by a sand–siltstone succession (Unit VI; Fig. F15) that appears as a ~4.5 m thick intercalation in the downhole logging data (of which only ~50 cm was recovered in the core).

Third massive basalt flow (stratigraphic Unit VII, lithologic Units 7–9)

The third massive flow of Group 1 in the downhole logging measurements is apparent as a ~8 m thick interval that follows beneath the sedimentary interbed of stratigraphic Unit VI (see "Sedimentology"). This flow contains <3% plagioclase microphenocrysts, only a trace of clinopyroxene, and no olivine, and thus it is the most aphyric of all basalts cored at Site U1347. Recovery is relatively good (>60%) with the topmost ~3 m being highly vesicular (20%–30%), passing down into ~2–3 m of nonvesicular basalt in the flow core. However, unlike the overlying unit, the lower part of this flow is absent.

Geochemical analyses reveal that the lower (nonvesicular) part of this unit is characterized by high TiO2 and Ni abundances compared to the top part and the overlying massive lava flows of Units VI and IV (see "Geochemistry" for more details). Such a difference could be interpreted as a compositional break reflecting different magma sources. If correct, this flow "unit" may actually represent juxtaposed portions of two separate flows (Fig. F15), which in downhole logging measurements may be evident from a pronounced decrease in resistivity at or near the level of their contact (see "Downhole Logging"). In addition, the lower flow is also characterized by high magnetic susceptibility, which diminishes significantly toward its base (Fig. F16). Unit VII is underlain by a well-bedded 4–5 m thick radiolarian-bearing clay-sandstone (Unit VIII; Fig. F15). This is of finer grain size than sediment of Unit VI and contains abundant signs of bioturbation (see "Sedimentology").

Fourth massive basalt flow (stratigraphic Unit IX, lithologic Unit 11)

The fourth and lowermost of the massive basalt units of Group 1 follows beneath the sedimentary intercalation of stratigraphic Unit VIII; it is ~12 m thick, moderately vesicular (especially in its upper part), and plagioclase-pyroxene phyric with phenocryst abundances of as high as 20% (Fig. F16). These petrographic characteristics distinguish it from the overlying three flows. Despite moderate recovery (<50%), many volcanic features are observed, including the baking of sediment at the upper contact, a well-defined chilled upper zone, a sparsely vesicular core, and a chilled lower margin at its base. This fourth massive flow is obviously much less vesicular than the flows above, but its core contains well-developed vertical vesicle pipes and associated small segregation zones containing darker melt material.

Group 2

Heterogeneous stack of pillow lavas and minor flow units (stratigraphic Units X, XII, and XIV)

The heterogeneous lava stack comprising Units X, XII, and XIV is 75 m thick and consists primarily of pillow lava inflation units with glassy margins and massive aphanitic to plagioclase-phyric interiors (Figs. F20, F21). Inflation units within the pillow lava–dominated sections have glassy margins 1–3 cm thick; these consist of plagioclase and pyroxene in sideromelane and display a transition to spherulites at the edge of the margin (Fig. F21). Unaltered glass is present in all of the chilled pillow unit margins throughout the pillow lava succession. Irregular tube- and drop-shaped vesicles (or amygdules) are present along the inside of some pillow margins. Most of the vesicles in the pillow lava succession are filled with secondary minerals. The interiors of pillows are relatively free of macroscopic features but rarely preserve megavesicles, cooling joints, or sparse spherical vesicles (<1 mm in diameter). The groundmass of the pillow lavas varies from cryptocrystalline on the margin, to very fine grained in the interiors with sparse tabular and glomerocrystic plagioclase phenocrysts.

Since pillow lava inflation units are typically subspherical or oblate, it is important to note that the curated thickness may, in some instances, only represent marginal sections through these pillow units and thus represent a significant underestimation of the individual pillow thicknesses; however, we deduce that these types of inflation units generally have individual thicknesses ranging from ~0.2 to 1.2 m (Fig. F22). From the recorded curated size distribution, ~1 m appears to be the modal maximum thickness for individual pillow inflation units in this stack. Examination of individual unit thicknesses within the stack as a whole indicates a broader size distribution including three larger (~3–6 m) massive units, nine or ten medium-sized inflation units (~1–2 m), and the intervening successions consisting of numerous small pillow inflation units (~0.2–0.8 m). The pillow units are intercalated with sediments ranging from sandy siltstone to volcaniclastic limestone, in some cases bedded sediment layers are as thick as ~5 m (e.g., Units XI and XIII), and in other instances they occur as small interpillow sediment pockets (e.g., intervals 324-U1347A-19R-3, 57–78 cm, and 22R-5, 19–30 cm).

For ease of description, Group 2 has been arbitrarily divided into upper, middle, and lower sections on the basis of the intervening sedimentary intercalations (Units XI and XIII).

Upper pillow lava stack (stratigraphic Unit X, lithologic Units 12–42)

This upper lava stack follows beneath the fourth massive flow of Group 1, lacks any (cored) intervening sedimentary bed, and consists predominantly of small pillow inflation units less than ~1 m thick. In total, >20 small pillow inflation units can be identified ranging in recovered thickness from 0.15 to 1.35 m, yielding a total recovered thickness of ~40 m. Vesicularity within these pillows is variable, with only rare development of well-defined concentric vesicle zones, and chilled margins often lack visible radial cracking and the associated vesicle concentrations. Petrographically, these pillow lavas are composed of massive aphyric, sparsely phyric, and plagioclase phyric basalt, with plagioclase phenocryst abundances of <1%, 3%–5%, and 10%, respectively. Plagioclase "phenocrysts" often occur in the form of anhedral plagioclase laths arranged into glomerocrysts of ~2–3 mm in size. Groundmass textures vary from cryptocrystalline through microcrystalline to very fine grained (and rarely to medium grained) groundmass sizes from chilled margins into flow interiors.

Thicker inflation units of 1–2 m occur toward the lower part. Much larger units of 3–5 m thickness (lithologic Units 37 and 42) form the base of this section but are themselves separated by four or five successions of larger pillow units. The thickest of these larger units (lithologic Unit 42) lies directly upon a thin sedimentary intercalation of silty claystones and a coarse-grained volcaniclastic limestone (stratigraphic Unit XI; Fig. F15), which defines the boundary between the upper and middle lava stack sections.

Middle pillow lava stack (stratigraphic Unit XII, lithologic Units 46–58)

This middle lava stack consists of ~10 small pillow inflation units, ranging in thickness from ~0.2 to 1.5 m, and is intercalated with a single thick flow unit of ~5.6 m, giving it a total recovered thickness of ~20 m. Vesicularity in the pillow units is variable, with only rare development of well-defined concentric vesicle zones, and chilled margins often lack visible radial cracking and associated vesicle concentrations. Petrographically, these pillow lavas are composed of massive aphyric, sparsely phyric, and plagioclase phyric basalt. The core of the larger unit contains pipe vesicles and segregation features. Downhole logging indicates low resistivity throughout much of this section, probably due to the stacking pattern of the small pillow units; by contrast, the 5.6 m thick flow in the middle of Unit XII is characterized by a consistently higher resistivity. The middle lava stack is underlain by a thin ~15 cm (~50 cm by downhole logging) sedimentary intercalation of highly altered volcaniclastic sandstone (stratigraphic Unit XIII; Fig. F15), which defines the boundary between the middle and lower lava stack sections.

Lower pillow lava stack (stratigraphic Unit XIV, lithologic Units 60–80)

This lower section consists of ~10 large inflation units typically ranging in recovered thickness from ~0.5 to <2.0 m, and which are intercalated with a few much smaller pillow units of ~0.1–0.2 m, giving a total recovered thickness of ~14 m. This is close to the true thickness since recovery throughout this lava stack approached 100%. Vesicularity in the larger inflation units typically occurs as weakly concentrated accumulations of vesicles within the upper chill zone but increases in density (~1%–2%) toward the top glassy contact. Pipe vesicles and magmatic segregation features occur within the core of the units. The lowermost units of this stack are a succession of particularly small pillows (~0.1–0.2 m thick) containing thin intercalations of highly baked sedimentary material. These small pillows lie directly upon the top of the massive flow units of Group 3.

Group 3

Fifth and seventh massive basalt flows (stratigraphic Units XV and XVI, lithologic Units 81 and 82)

The largest of the volcanic units recovered in Hole U1347A occurs immediately below the lowermost pillow basalts of Group 2. The uppermost lava flow (Unit XV) is ~23.1 m thick and is itself immediately underlain by Unit XVI, which is similar in character but only cored to a depth of ~2 m before drilling was terminated at the site. Based upon observable internal characteristics in the recovered core this is likely to have been of a similar "massive" dimension. Both flows have well-developed chilled zones at the top (glassy to microcrystalline in the topmost ~1–2 m), with a somewhat thinner one (<0.5 m) at the base of the upper flow. By comparison with the overlying lava stack, relatively little glass formed at these flow contacts, but intense sediment baking effects were observed to have occurred where a thin sandstone or mudstone had intervened between the two flows (interval 324-U1347A-29R-4, 70–82 cm). Vesiculation is confined to the upper 2–3 m, below which the flows become very homogeneous and largely nonvesicular. Moreover, pipe vesicles and associated magmatic segregation features are less well developed compared to the four thick units (IV, V, VII, and IX) that occur in Group 1 at the top of this volcanic succession.

Texturally, these units consist of thick, homogeneous, and very fine grained flow cores, progressively passing into microcrystalline and ultimately cryptocrystalline patches through the chilled zones and at the chilled margins. Within the massive flow of Unit XV, several 30–70 cm thick zones of marginally coarser texture may be identified; these may reflect cooling (and crystallization) heterogeneities or multiple injections during inflation (see "Petrography and igneous petrology" for more details).

Preliminary assessment

It is known that the stacking and horizontal arrangements of lava successions are complex, though such studies are largely restricted to continental examples (e.g., Jerram and Widdowson, 2005). From these studies, the hierarchical organization of lava units may be described in terms of lava flow-fields, lava flows, and lava lobes (Self et al., 1997). The term "flow-field" designates the entire product of an eruption event and may consist of one or multiple lava flows. The term "flow" refers to a lava body formed by the solidification of a single outpouring of lava and represents a single episode of magma effusion during the eruption event. However, lava flows rarely form a laterally continuous stretch of lava; instead they consist of successions of lobate segments or inflation units, which can occur on several scales. Continual inflation of such lobes (or coalescence of several lobes) produces a broad lava body with a very flat upper surface that could be identified as a "sheet" lobe (Hon et al., 1994). Meter- or kilometer-scale sheet lobes can themselves be found to be composed of centimeter- to meter-scale lava lobes in the field. The term "toe" is given to small, centimeter- to decimeter-sized budding lava lobes. Importantly, in marine eruptions, these toes typically form glassy rinds, inflate, and often separate from the main lobes to produce pillow lava units. All these different scales of inflation units are analogous to those of the different unit types recovered in Hole U1347A.

Pillow lava stacks

Pillow lava units of Group 2 are similar in dimension to those cored in Hole U1346A, but they differ in key respects. For instance, the pillows cored in Hole U1347A are much less vesicular, with only rare development of concentric vesicle zones or banding, and they often lack visible radial cracking around their margins. In addition, they are highly remarkable for their well-developed chilled margins and dark glassy rinds (~1–3 cm thickness), in which fresh basaltic glass (showing vitreous, conchoidal fractures) is commonly preserved (Figs. F20, F21). Structures within the glassy chilled margins also reveal details of deformation that resemble those of typical pahoehoe-like crusts, such as cross-sections through wrinkled and folded lava crusts. In other cases these chilled margins provide evidence for the coalescence of adjacent pillow units, occurring between lobes at the centimeter to meter scale (e.g., Sections 324-U1347A-19R-2 and 20R-2). Most basalts in Group 2 appear petrographically similar. However, inspection of the stacking pattern, together with differences in the sizes of the inflation units and the positioning of intercalated sedimentary horizons (Fig. F15), reveals a repetitive pattern. For example, the pillow basalt sequences all appear to begin with larger massive units, pass upward into predominantly medium-sized inflation units, and then finally pass into an overlying succession of small, closely packed pillow lava units before the cycle is repeated. This stacking pattern may indicate several eruptive pulses, each producing separate flow-fields, during which the rate of lava effusion diminishes over time. In such a scenario, the thicker, larger units (>3 m) would be produced at the onset, followed by the production of progressively thinner units (1–2 m), and concluding with a more prolonged but reduced-magmatic-flux eruptive phase producing the smaller pillow units. Hiatuses between the eruption of these putative flow-fields are indicated by the occurrence of sedimentary intercalations.

Flow-field organization of the massive sheet flows

The thicker massive flows of Groups 1 and 3 have well-developed upper chill zones and thinner ones developed at their lower contacts. They also have thick homogeneous cores with the degree of crystallinity increasing subtly away from their margins. These petrographic changes define a broadly tripartite internal structure that divides each flow into a vesicular upper crust, a massive core, and a sparsely vesicular basal zone. The upper crust has glassy folds and concentrations of vesicles (and amygdules) in the top 1–3 m (10–30 vol%). The massive core of each unit is typically homogeneous, sparsely vesicular with rare megavesicles, and often incorporates well-developed pipe vesicle zones containing dark melt segregation material. Vesicle density in these unit interiors is usually negligible (<2%), and these vesicles are randomly distributed, rounded, subrounded, or sometimes coalesced. A large proportion of the vesicles are filled with secondary minerals (calcite and clay minerals); however, the diktytaxitic vesicles remain open. Contacts between the large units typically reveal a glassy margin, sometimes with brecciated basalt clasts. The upper crust exhibits folds and stretched vesicles typical of inflated pahoehoe sheet flows (e.g., Hon et al., 1994) with glassy margins containing spherulites (Figs. F17, F18). These crusts display several alternating horizontal layers of larger and smaller vesicle horizons and rare large (>5 mm) vertically elongate, irregularly shaped vesicles.

Petrographically, the massive interiors of the larger flows are composed of a microcrystalline to very fine grained, typically aphyric to sparsely plagioclase phyric basalt, with groundmasses consisting of plagioclase and clinopyroxene microlites and altered glass (see "Petrography and igneous petrology"). Phenocrysts occur either as 2–3 mm clots and/or intergrowths of two distinct generations of plagioclase. These plagioclases are either characterized by cores of individual tabular plagioclase crystals, of up to 2 mm, surrounded by later acicular microlites, or else they appear as smaller 0.3–0.6 mm glomerocrysts composed entirely of plagioclase microphenocrysts. Only occasionally does the tabular plagioclase occur independently as separate crystals.

Ubiquitous development of segregation features associated with pipe vesicles in the four upper massive flows of Group 1, and in the larger inflation units of the underlying Group 2 succession, indicates emplacement and subsequent inflation through continued high-rate magma supply, followed by stagnation, cooling, degassing, and solidification. These thick units are comparable in both dimension and internal structure with those commonly observed in continental flood basalt provinces (Thordarson and Self, 1998); accordingly, it may be deduced from their thickness dimension that they are also likely to be laterally extensive sheet flows.

Volcano-sedimentary relationships

The sediments lying above, and intercalated within, the volcanic succession recovered from Hole U1347A are indicative of the availability of basalt-derived material both during and after the eruptive episodes. Sedimentary structures, extensive bioturbation, and macro- and microfossil content indicate that moderate-depth to shallow-marine depositional conditions were established during noneruptive periods (see "Paleontology" and "Sedimentology"). To better understand how the sedimentary record may have responded to lava eruption, it is useful to briefly outline the dynamics of lava flow emplacement and the evolution of a lava field or pile. Lava fields evolve continually over periods of years to decades. During their development, different areas will become active due to changes in the supply of magma from the vent sources to the propagating tips of individual lava flows that constitute the evolving lava field. Accordingly, areas of an active lava field become periodically abandoned and elsewhere rejuvenated over the period of active volcanism. These changes can result in a highly dynamic physical environment across the surface of a single lava field, which will have profound effects upon the distribution of coeval sedimentation. For instance, until the whole of the lava field becomes inactive, the development of sedimentary deposits will be localized and patchy. Once eruption does cease altogether, the whole of the volcanically generated surface becomes a potential site for alteration and/or deposition. Development of sedimentary deposits is far more likely in a submarine environment, compared with subaerial flows. The intra- and supra-volcanic deposits recovered in Hole U1347A can thus be respectively interpreted as patchy, localized, volcanically coeval deposits, developing into a more continuous post-eruptive sedimentary succession in a shallow- to offshore-marine environment.

Petrography and igneous petrology

Variably porphyritic plagioclase-clinopyroxene basalt pillows and larger inflation units were recovered in Hole U1347A. Polished thin sections of rims and interiors of 73 of these flows were prepared, including all instances where inductively coupled plasma–atomic emission spectroscopy (ICP-AES) analyses were made. The chemical analyses reveal that a variety of distinctive and fairly differentiated rocks were obtained (TiO2 > 1.5%; Mg# < 60), all of incompatible element depleted tholeiitic character (see "Geochemistry"). Fresh glass was recovered at many locations, readily enabling identification of the microphenocrysts and phenocrysts preserved within it. From these exterior portions of flows and pillows, general crystallization histories could then be traced into their interiors.

Based on lithology, units of the basalt succession were divided into three groups (Fig. F15). The distribution of these groups and lithologic units, together with a summary of the relative abundances of phenocrysts and the presence of glass, is shown in Table T6, and more precise visual estimates based on thin section descriptions are plotted in Fig. F23. The occurrence of fresh glass has been logged in 324GLASS.XLS in LOGS in "Supplementary material." The following gives a summary of the petrographic description of the major lithologic groups using thin sections. A discussion follows of the crystallization histories based on rock textures and mineralogy as illustrated by photomicrographs.

Group 1 upper massive basalt flows (stratigraphic Units IV, V, VII, and IX)

The petrography of the upper massive basalt lava flow units (four massive flow units: IV, V, VII, and IX) is based on 28 thin sections from all identified flow units. These are of low to moderate vesicularity (1%–15%), are aphyric to sparsely plagioclase-phyric (1%–5%), and commonly contain glomerocrystic plagioclase and/or clinopyroxene. Plagioclase is the most abundant phenocryst phase, usually present as glomerocrysts forming aggregates as large as 5 mm. Relatively high abundances (5%–7%) of plagioclase phenocrysts occur in lower Unit IX (Figs. F16, F23). Olivine and clinopyroxene are observed as microphenocrysts. Although olivine is typically pseudomorphed by brown clay (see "Alteration and metamorphic petrology"), a few unaltered olivine microphenocrysts are included in the more glassy parts of Units IV and V. Some are skeletal in form and as large as 0.7 mm. Olivine microphenocrysts commonly contain rare inclusions of altered glass and (only) one olivine crystal contains Cr spinel. Clinopyroxene microphenocrysts usually subophitically enclose acicular microcrysts of plagioclase. In groundmasses of the larger massive flow interiors, intersertal to subophitic textures predominate. Unaltered glass is present in the marginal areas of massive flow units. The unaltered glass in the rims merges within centimeters to zones of dark brown coalesced spherulites (e.g., Fig. F24).

Group 2 pillow lava stack (stratigraphic Units X, XII, and XIV)

The petrography of pillow lavas (Units X, XII, and XIV) is based on 36 thin sections from 19 of the 57 inflation units identified (see 324UNIT.XLS in LOGS in "Supplementary material"). These have 0%–3% vesicles that range in size from 0.1 to 1 mm in the glassy rims but are larger in pillow interiors, except the lowermost massive flow of Unit X (Fig. F15), which has a higher vesicularity (7%–15%). The basalts are aphyric, sparsely plagioclase-phyric (0%–5%), and sparsely plagioclase-clinopyroxene phyric (5%–20%). Plagioclase phenocrysts are commonly glomerocrysts forming aggregates as large as 3 mm. On the other hand, pyroxene and olivine occur only as microphenocrysts (<1 mm diameter) throughout the pillow lava stack. Most pyroxene microphenocrysts are intergrown with acicular or tabular plagioclase. Olivine microphenocrysts are rare in most thin sections (several grains per thin section) and are almost completely replaced by secondary minerals. With the exception of the upper pillow lava succession of Unit XII, the relatively thick inflation units (>1.5 m) within the pillow lava successions have no olivine, whereas many of the thinner inflation units (<1.5 m) contain olivine (Table T6). Many glassy margins of pillow tops, sides, and bottoms are preserved as fresh glass, most usually mantling basaltic rock with spherulitic textures (see 324GLASS.XLS in LOGS in "Supplementary material"). Groundmass minerals of flow interiors are microcrystalline to very fine grained and are composed of acicular plagioclase, dendritic clinopyroxene, and tiny skeletal titanomagnetite. Textures around pipe vesicles or gas blisters, as well as around magmatic segregations, are aphyric and cryptocrystalline, darker in color, and contain more titanomagnetite than surrounding rock.

Group 3 lower massive basalt flows (stratigraphic Units XV and XVI)

The petrography of the lowermost massive basalt flows (Units XV and XVI) is based on nine thin sections. These are aphyric to sparsely plagioclase-phyric (<3%) and commonly contain both plagioclase pheno- and microcrysts and olivine microphenocrysts. Clinopyroxene phenocrysts are not present. Plagioclase is the most abundant phenocryst phase, usually present as glomerocrysts forming aggregates as large as 3 mm. Individual tabular phenocrysts of plagioclase reach as large as 2 mm. Olivine occurs as microphenocrysts and is typically pseudomorphed by brown clay (see "Alteration and metamorphic petrology"). Some of it is skeletal in form, and as large as 0.5 mm in diameter, but it also forms glomerocrystic aggregates with plagioclase phenocrysts and microphenocrysts. Groundmasses in the massive flow interiors have coarsely spherulitic to intersertal and subophitic textures (e.g., Fig. F25). The grain size of intergrown groundmass plagioclase and clinopyroxene in Unit XV (0.2–0.4 mm) is greater than Unit XVI basalt below it (0.1–0.2 mm).

Crystallization history of basalts

Figures F24, F25, F26, F27, F28, F29, F30, F31, and F32 depict the crystallization histories and many other attributes of the basaltic rocks in Units IV–XVI. Captions in all cases identify the curated depths and the stratigraphic unit (I–XIV) of the thin sections, including their main characteristics (flow unit, pillow succession, etc.) and positions (rim, top, upper, middle, etc.) within the sampled stratigraphic unit or interval.

In hand specimen, many rocks were deemed aphyric, whereas in thin sections even the glassiest samples were observed to contain tabular/elongated to partly skeletal crystals of plagioclase, with variable proportions of these plagioclases intergrown with anhedral pale brown clinopyroxene. These crystals could be termed "microphenocrysts" because they are tiny (<1 mm). The two-phase intergrowths are either isolated pairings of the minerals or crowded multicrystal clumps and clusters and are of different grain sizes both within and between samples. Many thin sections contain (a few) much larger phenocrysts, megacrysts, and glomerocrysts of the plagioclase/clinopyroxene aggregates, in some cases including small amounts of olivine. Except at extreme glassy margins, the crystals are usually mantled with dark brown spherulitic overgrowths that consist chiefly of needles of plagioclase and intergrown fibers of dendritic clinopyroxene. The crystallization histories of the rocks, therefore, can be considered in two stages: (1) a preeruptive interval crystallizing larger crystals (and aggregates) and (2) a syn- and posteruptive interval with smaller crystals growing in the melt at the time of eruption. Synchronicity between plagioclase and clinopyroxene crystallization in the preeruptive stage is indicated by their combined nucleation on the planar surfaces of faceted plagioclase, whereas spherulitic overgrowths imply a sudden and abrupt increase in cooling rate at the point of eruption.

The presence of up to several percent of tabular plagioclase crystals (e.g., within and adjacent to glassy material) intergrown with well-formed anhedral clinopyroxene, the rarity of olivine, and the almost complete absence of Cr spinel distinguishes these rocks from their less differentiated counterparts at Sites 1213 and U1346 elsewhere on Shatsky Rise. Evidently, eruption did not change the identity of the crystallizing minerals because during every stage of crystallization the basalts appear to be saturated in both clinopyroxene and plagioclase. The rocks were, therefore, in a condition of low-pressure plagioclase-clinopyroxene cotectic crystallization during their entire cooling and differentiation history. Olivine was a minor (trace to 1%) coprecipitating mineral during only the earliest stages. It was later supplanted by titanomagnetite, which only is visible between crystals of the silicate minerals in the rapidly cooled spherulitic portions of pillows and flows and within intersertal patches in their interiors.

Crystallization history of pillow and flow units following eruption

Pillow basalts in Hole U1347A are restricted to a central stack of ~75 m in the middle of the basalt sequence (Group 2) and occur in packages sandwiched between thicker flow units of up to ~23 m thick. In general, the pillows recorded the same pattern of crystallization, from rims to interiors, as those of pillows in Hole U1346A, with the exception that olivine is rarely present and Cr spinel is absent (except for a single tiny crystal in one thin section). The rocks become increasingly crystalline away from glassy rims, progressing from the outer zones dominated by spherulites of plagioclase toward the interior of pillow inflation units, where branching acicular plagioclases are intergrown with smaller grains of dendritic to subhedral clinopyroxene and skeletal titanomagnetite. The rocks in Hole U1347A, however, are significantly less altered than those in Hole U1346A, and, therefore, the textures and morphologies of all the minerals can be easily observed with the microscope. The modal proportion of fresh rock as seen by crystallinity in reflected light (percentage of polished mineral surfaces) is never <50% and, in some examples, is as high as 95%.

More information about crystallization can be determined from the thicker flows. Figures F24 and F25 depict the crystallization of two of the five thickest cooling units in Hole U1347A. These, respectively, are the third massive basalt flow (Unit V; Flow 3; ~8 m thick based on logging) near the top of the basalts in Group 1 and the fifth massive basalt flow (Unit XV; Flow 5; ~23 m thick based on logging and curated coring depths) in Group 3 near the base of the cores (see Fig. F16 for locations).

In both of these massive flows, general increases in grain size (e.g., plagioclase observed in photomicrographs) and crystallinity are apparent through the flow interiors. Grain size and crystallinity increase from both tops and bottoms but not in the same way. For instance, the degree of crystallinity, and the average grain size of Flow 3 at 4 m from its base, is far less than that observed in Flow 5 at only 1.5 m from its base. This is probably because Flow 3 is much thinner and therefore cooled more efficiently from its top than from its insulated base (Kirkpatrick and Hodges, 1978). The stubbiness (width to length ratio) of plagioclase crystals in Flow 3 is also greater throughout. Along with a lower degree of crystallinity, this may be a consequence of a more differentiated composition, or else the melt of Flow 3 was simply more viscous on eruption and thus less prone to form elongate and acicular plagioclases (Kirkpatrick, 1975).

Flows 3 and 5 also can be distinguished in another way. Many subaerial basalt flows tend to crystallize in frameworks of minerals (Jerram et al., 2003). Isolated crystals and clumps at their margins continue to grow in both length and width in interiors until they touch and form interlocking aggregates and networks. At this point, because the delicate networks are rarely broken, almost no motion of interstitial liquid is present and the networks will continue to crystallize in situ. In submarine tholeiitic pillow lavas, two types of networks can be distinguished, even in the centers of small pillows: (1) those consisting mainly of interlocking acicular plagioclase (e.g., Kirkpatrick, 1979; Natland, 1979) and (2) those consisting of aggregates or networks of intergrown plagioclase and clinopyroxene, forming a so-called "sieve texture" around much finer grained dark intersertal patches.

In these terms, Flow 3 is more clearly allied with the first type, with plagioclase crystals first touching and then forming a weakly linked network in its interior (Fig. F24D). Flow 5 exhibits clumps or clusters of intergrown plagioclase and clinopyroxene that touch and form sieve textures in its interior (Fig. F25B–F25F). Similar sieve-textured plagioclase-clinopyroxene networks separating intersertal patches are also well developed in Flows 1 and 4. Some portions of the interior of Flow 5, however, are not nearly so coarsely crystalline as most other flows in Hole U1347A, nor do they have well-developed sieve texture. This is partly evident in Figure F25D, whereas Figure F25E clearly shows two zones of different crystallinity, crystal size, and network characteristics on the left and right. In these zones, the finer grained material is evident as darker splotches and streaks consisting of dark glass, finely crystalline dendritic clinopyroxene, and elongate skeletal titanomagnetite, suggesting they likely are more differentiated than the bulk of the rock. The patches and streaks of finer grained and darker material within Flow 5 do not appear to have separated from the rest of the rock and created an internally hybrid lava flow.

Finally, further support for this cooling history for these thick massive lava flows comes from variations in magnetic susceptibility measured within Flow 5 (see "Physical properties"). This is controlled by the quantity of titanomagnetite in the rock, which, in submarine basalts, depends on crystallinity. Those rocks in which all glass has been crystallized into minerals (including all vestiges of intersertal glass) have the most titanomagnetite (corresponding to the coarsest crystalline portions of the core). Accordingly, the glass in quenched pillow rims should have the least titanomagnetite and thus have a lower magnetic susceptibility. On this basis, the nearly continuously recovered Flow 5 records the least magnetic susceptibility at its "quenched" top, fluctuations thereafter, and a general increase toward its base, below which the susceptibility again decreases toward the contact (Fig. F33). These data support petrographic inferences that the flow cooled asymmetrically, is coarsest grained and most crystalline toward its base, has chemical variability, and thus has intrinsically different proportions of titanomagnetite within it. A similar increase in magnetic susceptibility with depth occurs in Flow 1 (~19 m thick) but without the fluctuations (Fig. F33).

Phenocrysts, megacrysts, and glomerocrysts

Large crystals in these basalts are tabular, equant, or irregularly shaped and are significantly larger than the smaller tabular plagioclases and intergrown clinopyroxenes that are plentiful in the glass (Fig. F26). These larger crystals were all formed prior to eruption. Some have skeletal or sievelike interiors or occur as broken zoned fragments, or occur within multicrystalline aggregates and glomerocrysts. They are variously present in glass, in spherulitic portions of pillows and flows, and in their coarse-grained interiors, and their proportions can vary from one thin section to the next, between hand specimens, and within individual cooling units. In many cases, grain size is sufficiently small that the phenocrysts and microphenocrysts were more readily discerned in thin sections than during visual description of the core. However, special aspects of their crystal morphologies (in particular zoning, skeletal morphologies, or broken outlines) permit them to be distinguished from posteruptive crystals of similar size.

Table T6 shows the estimated presence and relative abundance of phenocrysts as estimated using thin sections. Most notable are the abundance of plagioclase phenocrysts in Flow 4 and the rarity or absence of clinopyroxene phenocrysts in the lower pillow stack (stratigraphic Units IX and XIV, respectively), as well as their absence in Flows 5 and 6 (stratigraphic Units XV and XVI). Olivine is persistent in trace abundances throughout the core.

Olivine: This rare mineral is almost always replaced by pale brown clay minerals (Fig. F27A–F27C), calcite, or both (Fig. F27D). Despite alteration, its crystal outlines are usually maintained even if the crystals have delicate dendritic extensions (Fig. F27A, F27B). Some olivine is intergrown with tabular plagioclase (Fig. F27C), and some grains have broken edges (Fig. F27D). Small dark brown circular areas within the crystals (Fig. F27B) are probably altered glass inclusions. Very rarely, some relict fresh olivine remains (Fig. F27E), with even rarer inclusion of Cr spinel (the only grain of this mineral seen). Overall, the few olivine phenocrysts appear to have crystallized mainly before extrusion but continued to crystallize with dendritic extensions afterward.

Plagioclase: Plagioclase displays the most varied crystal morphologies. Here we consider only the larger crystals and glomerocrysts that formed prior to eruption. Some crystals are tabular but have a skeletal interior, often with a nonskeletal mantle or normally zoned exterior (Fig. F28A). Others are elongate synneusis aggregates of several crystals, each having a central twin plane (Fig. F28B), or else they appear as moderately complex multicrystalline aggregates made up of several crystals bounding each other along portions with skeletal interiors (Fig. F28C). All of the above crystals appear to be a single array of exterior oscillatory and normal zones. Finally, some glomerocrysts are complexly multicrystalline, with melt inclusions within and between crystals and grains exhibiting complex sequences of skeletal growth and oscillatory zoning (Fig. F28E, F28F). Oscillatory zoning observed in a single crystal (Fig. F28F) reveals the different rates of growth along different crystal facets when it still was isolated in the melt. The crystal then became attached to the glomerocryst (Fig. F28E), after which the oscillatory zoning continued to develop in the aggregated crystal. Many of the observed glomerocrysts exhibit similar complex patterns of growth. The path from simple faceted and tabular morphologies to a complex aggregate of crystals can be readily observed, even in megacrysts approaching 0.5 cm (Fig. F29), and some glomerocrysts preserve unaltered glass inclusions both within and between crystals (Fig. F30). These inclusions may record the degassing histories of these rocks that, on the basis of their vesicularity and characteristics of interbedded sediments, appear to have erupted in a shallow-water environment.

Clinopyroxene: Figure F31 shows typical attributes of the larger clinopyroxene population. The most usual occurrence in the glassy portions of the flow is as small crystals intergrown with tabular or at least faceted plagioclase. Some of the pyroxenes have a somewhat lumpy but mostly faceted outline (Fig. F31A), whereas others are intergrown with skeletal plagioclase (Fig. F31B). Toward flow interiors, in the more finely spherulitic portions of the rocks, the intergrowths shown in Figure F31A combine into loosely strung aggregates (Fig. F31C), which seem to be precursors to the network sieve structure described earlier. In coarser grained rocks, the clinopyroxene grains also are larger, and these aggregates become fully interlocking (Fig. F31D). Examples of truly coarse grained aggregates can be found with clinopyroxene that subophitically enclose angular plagioclase fragments (Fig. F31E). These examples suggest a substantial period of cotectic crystallization of clinopyroxene and plagioclase in a slowly cooled magma prior to eruption. Euhedral clinopyroxene phenocrysts occur only in a few glassy samples. Examples from flow tops are shown in Figure F31F and F31G, where clinopyroxene appears as isolated small crystals.

Titanomagnetite

Titanomagnetite has importance in determining certain rock magnetic properties such as magnetic susceptibility. Many of the rocks are fresh enough to preserve detailed crystal morphologies of this mineral (Fig. F32). Glass rims from pillow units typically have no titanomagnetite, which is confirmed after a close inspection of the glassy chilled margins, which reveal swirling spherulitic patterns of fibrous plagioclase and brighter dendritic clinopyroxene containing only a very few titanomagnetite grains (Fig. F32A, F32B).

However, within ~2–3 cm of these glassy outer rims, the spherulitic swirls of plagioclase and dendritic clinopyroxene become populated with hundreds of tiny titanomagnetite grains (Fig. F32C). In high-magnification reflected light, these titanomagnetite grains reveal incipient skeletal morphologies, whereas coarser titanomagnetite with better developed skeletal morphology can be seen in the menisci of segregation vesicles despite the high degree of alteration (Fig. F32D). Intersertal patches in the coarser grained portions of the flows and pillows also tend to be altered, but not so extensively that the alteration destroys the titanomagnetite crystals. In Figure F32E, an elongate skeletal titanomagnetite grain spans a large portion of basaltic rock (more reflective, white), with interlocking clinopyroxene-plagioclase (light gray) and two altered (nonreflective, dark gray) intersertal patches into which the crystal grew lengthwise. In Figure F32F, many of the titanomagnetite grains (small, less reflective grains) are completely enclosed by intergrown plagioclase (dark gray) and clinopyroxene (light gray), yet none appears in the altered intersertal portions of the rock (nonreflective). In Figure F32G, an intergrowth of titanomagnetite and plagioclase is shown next to an altered space with clay minerals. Typically, titanomagnetite has alteration pits and cracks, but on the whole its morphology is well preserved.

Finally, Figure F32H shows an unusual cross section through a partially crystalline glass inclusion in a plagioclase crystal. The dark crystals are a trellis of dendritic titanomagnetite, with the larger crystals and their pointed tops representing a larger "trunk" and smaller orthogonal crystals being "branches." The mineral is set in glass that, in this thinned out view, is only pale brown. The tiny clear crystals are dendritic clinopyroxene. There is no acicular plagioclase evident.

Preliminary assessment

From their compositions, Shatsky Rise basalts in Hole U1347A are more differentiated than basalts in either Hole 1213B or Hole U1346A (see "Geochemistry"). In petrographic terms, this is evident from the predominance of plagioclase-clinopyroxene intergrowths at all stages of crystallization, the scarcity of olivine phenocrysts, and the almost complete absence of Cr spinel. These phenocryst assemblages and the character of intergrowths compare well to those of rather evolved, low-temperature gabbros that formed in "shallow" crustal magma chambers (such as those occurring beneath superfast spreading ridges). The fastest spreading ridge currently is the East Pacific Rise, yet basalts there are almost always aphyric, whereas basalts of ridges spreading at slow and intermediate rates, such as the Mid-Atlantic Ridge and Costa Rica Rift, are often porphyritic (e.g., Hekinian and Morel, 1980; Bryan, 1983; Natland et al., 1983). Natland and Dick (2009) attribute this aphyric nature to a filtration effect occurring in the nearly consolidated gabbros beneath the East Pacific Rise and the necessity to drive only eruptive liquids through this and into a steady-state "melt lens" at the top of these gabbros; importantly, this lens is not seen on slowly spreading ridges (Sinton and Detrick, 1992). This melt lens collects only evolved residual liquids that happen to work their way buoyantly through a dense filtration network or series of flow channels in the gabbros. Whereas melt leaks into the lens, the phenocrysts are instead retained as textural relics in the gabbros. Eruption then taps the aphyric magma in the melt lens. In the case of Shatsky Rise basalts, the high abundances of cotectic plagioclase and clinopyroxene phenocrysts present makes it probable that there was no melt lens present beneath Site U1347, or beneath Sites 1213 and U1346, despite the postulated high spreading rates for the spreading surrounding Shatsky Rise.

Experimental petrology on mid-ocean-ridge basalt (MORB) (e.g., Stolper 1980; Fujii and Bougault, 1983; Kinzler and Grove, 1992), and on rocks from the Ontong Java Plateau (OJP) large igneous province (LIP) (Sano and Yamashita, 2004), indicates that precursor magmas only have olivine on the liquidus at low pressure. This is then followed by plagioclase and finally clinopyroxene at lower pressures and a temperature interval of only ~50°C. The rocks in Hole U1347A are multiply saturated in these minerals, with plagioclase and clinopyroxene predominating, suggesting that these minerals have crystallized at similar temperatures, but at very low pressures upon eruption in a shallow submarine environment. The basalts in Holes 1213B and U1346A were only slightly different in their crystallization sequence, as they had both olivine and plagioclase phenocrysts when they erupted, soon followed by clinopyroxene. The supposition that primitive MORB and LIP liquids always have olivine alone on the liquidus, and thus must always be picritic, has been questioned previously (e.g., Kushiro and Thompson, 1972; Natland et al., 1983; Fisk, 1984). Shatsky Rise now appears to be an additional location where this conventional wisdom may be questioned.

Primitive basaltic liquids that are multiply saturated in several silicate phases raise two other petrogenetic problems. First, experimental petrology (Fisk, 1984) suggests that they originated by partial melting at low pressure, at depths perhaps only 10–20 km in the mantle. Alternatively, computer simulations suggest that low to moderate H2O content in the magma may have caused phase boundaries to converge at low pressure (Almeev et al., 2008), especially in enriched basalts that have higher concentrations of K2O and related incompatible elements. This raises the question: did Shatsky Rise basalts result from partial melting of either "dry" or fairly "wet" mantle at shallow depths? Second, along spreading ridges, multiply saturated and fairly primitive rocks exist along portions of the Central Indian Ridge, where gabbroic and ultramafic rocks have been dredged from many fracture zones (e.g., Engel and Fisher, 1975). They also occur in Deep Sea Drilling Project Hole 504B on the Costa Rica Rift (Natland et al., 1983), where the seismically measured thickness of ocean crust is only ~4 km. In a similar fashion to the rocks in Hole 504B, the rocks of Shatsky Rise are depleted and have low concentrations of elements such as Zr, Y, and TiO2 (see "Geochemistry"). Indeed, the concentrations of these elements are low even within the MORB spectrum, and this may also mean that the rocks are dry. However, if this interpretation is correct, why, then, should rocks that are depleted, and which cotectically crystallized plagioclase and clinopyroxene, occur at Shatsky Rise where the crust is relatively thick?