In this section, we describe the primary textures and mineralogy of igneous basement units. Units 1-5 and 18-19 belong to a felsic series, and Units 6-17 belong to a trachybasaltic series (Fig. F44) at Site 1139 (see "Magmatic Characteristics").

Unit 1 is composed of four variably altered and brecciated volcanic subunits and an interbed of bioclastic sandstone (Subunit 1B). We observe that the volcanic subunits are generally composed of felsic lavas and volcaniclastic rocks (Subunits 1A and 1C-1E). Flow banding is common in Subunits 1A and 1C, and highly vesicular material (pumice) is present in Subunit 1C. Quartz phenocrysts are present in the rhyolite of Subunit 1A, but quartz is only a minor groundmass phase in Subunits 1C-1E, where the major phenocryst phase is sanidine. Subunit 1C exhibits a striking perlitic texture (Fig. F14).

Units 2 and 4 are vesicular rhyolites that may represent parts of a welded pyroclastic flow. They are macroscopically similar and, although highly altered, the groundmass exhibits a flow texture (Fig. F45A). They contain relatively unaltered sanidine (8%-15%) and minor (1%) quartz phenocrysts (Fig. F45B). Unit 3 is a crystal vitric tuff-breccia consisting of a highly altered, glassy, perlitic groundmass that contains sanidine and quartz phenocrysts (Fig. F46). In addition, lithic clasts are present and occasionally the sanidines are brecciated or shattered.

Unit 5 is a trachyte containing phenocrysts of sanidine and completely altered mafic phenocrysts (probably clinopyroxene), which have been replaced by siderite (Figs. F47). The moderately to highly altered feldspathic groundmass exhibits a trachytic texture and contains minor quartz that is possibly of secondary origin.

Units 6-17 vary from trachybasalt to basaltic trachyandesite and are all moderately to highly altered. The top of Unit 7 is a generally clast-supported sediment, rich in sanidine crystals (Fig. F48A) and with very little matrix (now altered to opaque clay and hematite) (Fig. F48B). The sanidine crystals in this sediment and their absence as liquidus phases in the trachybasalt and basaltic trachyandesite lavas indicate derivation of the sediment from a separate, more evolved eruptive source.

Each mafic unit displays trachytic intergranular texture and is either aphyric or sparsely plagioclase-phyric (Figs. F49, F50A). The groundmass is composed of plagioclase (55% ± 10%), clinopyroxene (35% ± 10%), titanomagnetite (5% ± 3%), and mesostasis. Generally, the clinopyroxene and glass are moderately to completely altered, the plagioclase is moderately altered (to clay and replaced by carbonate), and the titanomagnetite is fresh to moderately altered (to maghemite and hematite) (Fig. F51). Fresh titanomagnetites exhibit rare maghemite exsolution. These lavas contain slightly more titanomagnetite, as much as 10 vol%, than lavas from other Leg 183 sites. A sulfide phase (pyrite? or pentlandite?) in Unit 17 may be primary because it is found as inclusions in unaltered primary minerals; the small size of this sulfide phase precludes positive identification.

Several petrographic features are significant. Firstly, every trachybasalt and basaltic trachyandesite contains large, euhedral sanidine crystals surrounded by a sieve-textured reaction rim (Fig. F50A). Rare examples have sieve-textured interiors. Some sanidine crystals exhibit highly resorbed outlines (Fig. F50B). In contrast, the presence of euhedral plagioclase microphenocrysts (i.e., Figs. F49, F50A) indicates that plagioclase is the probable liquidus phase. We conclude that the sanidine in these lavas is xenocrystic.

The second feature is the absence of albite twinning in many plagioclase phenocrysts. Only Carslbad twinning (Fig. F52A), which can cause plagioclase to be mistaken for alkali feldspar, is evident. It appears that strong compositional zoning in many of these laths (Fig. F52B) has inhibited the formation of albite twins, allowing only those of the Carlsbad variety to develop. Similar features were observed in some of the plagioclase glomerocrysts from other Leg 183 sites, especially at Site 1137, in basement Unit 10 (see "Igneous Petrology and Geochemistry" in the "Site 1137" chapter).

Many groundmass plagioclase laths also lack albite twinning and exhibit only Carlsbad twins, but unlike the plagioclase phenocrysts, no compositional zonation was observed. These groundmass feldspars exhibit undulatory extinction, and the Carlsbad twin plane is more diffuse than in the phenocrysts and in primary sanidines of the felsic rocks of Units 1-5 and 18-19 (Fig. F53). For example, in Figure F52A a groundmass feldspar exhibiting a "diffuse" Carlsbad twin (as well as undulatory extinction—see the inset box in Fig. F52A) is contrasted with that of the saniline phenocryst in this photomicrograph. We conclude that the groundmass plagioclase has been partially replaced by alkali-feldspar (plagioclase is still seen in the groundmass) and that the anomalous twinning and undulatory extinction result from this alteration. Similar features have been described elsewhere; for example, the altered basalts from the Ninetyeast Ridge (Frey et al., 1991). The replacement of plagioclase by potassic feldspar could indicate that the groundmass plagioclase was relatively Na-rich prior to alkali exchange with percolating K-rich fluids. Relatively sodic groundmass plagioclase is consistent with these lavas being more evolved than basalt.

The lowermost two units at Site 1139 (Units 18 and 19) are moderately sanidine ± clinopyroxene-phyric trachyandesite and trachyte, respectively. These units are completely altered, except for the top of Unit 18 and the bottom of Section 183-1139A-73R-3 in Unit 19, which are only moderately altered. In Unit 18, the sanidine phenocrysts are larger (8 mm) than in Unit 19 (3 mm) (compare Fig. F54A and F54B) Rare titanomagnetite phenocrysts, now altered to maghemite and hematite, are also present. Mafic phenocrysts are completely altered to clay and/or replaced by siderite (Fig. F55; see "Alteration and Weathering"), as in Unit 5 (see Fig. F47). The sanidine phenocrysts and vesicles can be traced through zones of high (pink brown) to complete (white) alteration. In these areas of Unit 19, veins of carbonate and/or quartz pervade the rock, to the point of giving the rock a brecciated appearance (Fig. F56). Nevertheless, the groundmass in the completely altered areas preserves an original trachytic fabric (Fig. F57). In the least-altered samples, the groundmass is composed of alkali feldspar (perhaps formed by the secondary replacement of plagioclase), quartz (perhaps primary although quartz-bearing veins are present), titanomagnetite (partially replaced by hematite), and a mesostasis composed of an intricate network of microcrystalline opaques set in a felsic glass (Fig. F58). All mafic minerals have been completely altered to clay or replaced by siderite and quartz. The microcrystalline clays suggest moderate alteration temperatures (see "Alteration and Weathering"). At the base of the borehole, Unit 19 exhibits a recrystallized groundmass containing abundant small, oriented laths of feldspar poikilitically enclosed in larger grains of a second feldspar and quartz (Fig. F59).

We report XRF data for 23 samples from Site 1139 (Table T11). Interpretation of these geochemical data is complicated by the moderate to high degree of alteration that has affected all Site 1139 basement samples. In the SiO₂ vs. total alkalis (Na₂O + K₂O) diagram (Fig. F60), the samples form an alkalic suite. Two samples (183-1139A-65R-3, 90-93 cm, Unit 10, and 66R7, 8-12 cm, Unit 13) that plot within the tephrite/basanite field are highly altered (>7% loss on ignition [LOI]) (Table T11) and probably have lost SiO₂ and/or gained alkalis during alteration (Fig. F60A). Yet, these samples are petrographically similar to the other trachybasalts. In the following discussion, these samples are grouped with the trachybasalts. The petrographically highly altered crystal-vitric tuff (Sample 183-1139A-56R-3, 92-95 cm, Unit 3) lies in the dacite field (Fig. F60A) but also has high LOI (~5.7%). The low-alkali, high-LOI, and high-MgO contents suggest that the tuff gained MgO during alteration, particularly since 4.2 wt% MgO is unusual in a rock with abundant quartz phenocrysts and high normative quartz (36.7%) (Table T11).

A change in the type of alteration is reflected in a downhole increase in CO₂ (Fig. F61). The CO₂ data, combined with identification of siderite in XRD analyses and petrographic observations of carbonate in the groundmass of many samples, suggest that fluxing of CO₂-rich fluids played an important part in the alteration process of Units 7-19 (see "Alteration and Weathering").

Six analyses of trachyandesite and trachyte from Units 18 and 19 demonstrate some of the characteristics of element mobility associated with alteration (e.g., in Fig. F62, the Unit 19 analyses are arranged in order of increasing alteration, as inferred from petrographic study and CO₂ and H₂O contents). There is a progressive decrease in SiO₂, Na₂O, and K₂O, and an increase in total Fe₂O₃* (* indicates the total iron as Fe₂O₃) and MgO, as the degree of alteration increases. In contrast, TiO₂ and Al₂O₃ have relatively constant concentrations and appear to be unaffected by the alteration. CaO and P₂O₅ have concentrations below detection limits in the two moderately altered samples and higher concentrations in the others. The alteration behavior of trace elements in Unit 19 is illustrated by Zr, Nb, Rb, and Ba. The first three show a decline with increasing alteration, whereas Ba varies more erratically. In contrast, the most altered sample from Unit 18 has a higher abundance of SiO₂, but a lower abundance of MgO, Al₂O₃, and Na₂O, and highly incompatible trace elements (Fig. F60A; Table T11). From these results it is clear that the chemical effects of alteration are complicated and that both the major and trace element data should be interpreted with caution.

Although the sequence has suffered relatively intense alteration, it is possible to extract some information about the magmatic characteristics of Site 1139 samples. The abundance of sanidine phenocrysts is consistent with the alkalic compositions of these rocks. When the petrographically more altered samples (Units 3, 10, 13, and 18) are excluded, compositions range from trachybasalt to rhyolite and plot entirely in the alkalic field (Fig. F60A). In general, these lavas are more alkalic than lavas from other Leg 183 sites (Fig. F60B). MgO and Ni contents are low (<4.3% and <13 ppm, respectively), and total alkalis are always >4.5% (Table T11). Concentrations of incompatible trace elements are also relatively high compared to the other Leg 183 sites (e.g., Zr contents are 292-420 ppm in the trachybasalts and basaltic trachyandesites) (Fig. F63).

Compositions change downhole from rhyolite and trachyte in Units 1-5, to trachybasalt (and one basaltic trachyandesite) in Units 6-13, to basaltic trachyandesite in Units 14-17, to trachyandesite in Unit 18 and trachyte in Unit 19 (Figs. F60A, F64). On the basis of TiO₂ and SiO₂ (Fig. F64), we separate the Site 1139 rocks into two groups that will be referred to subsequently, for convenience, as the mafic group (trachybasalts and basaltic trachyandesites) and the felsic group (trachyandesite, trachytes, and rhyolites). Although the Unit 3 crystal-vitric tuff has MgO and Al₂O₃ contents similar to those shown by Units 6-17, this may be a result of alteration, and we include this sample in the felsic group, especially since perlitic textures are identified in thin section.

In primitive mantle-normalized trace element diagrams (Fig. F65), samples from the mafic group have generally subparallel patterns that are surprisingly consistent considering the degree of alteration experienced by these units. In the felsic group, patterns are more variable as a result, at least in part, of the fractionation of feldspar and incompatible element-rich accessory phases (e.g., low Ba, Sr, and Ti relative to adjacent element concentrations). The crystal-vitric tuff sample (Unit 3) is again distinctive, having the lowest Zr and highest Nb compared to the other felsic units, but this may be a result of the high degree of alteration experienced by this sample.

As noted above, volcanic rocks from Site 1139 are distinguished from all other samples drilled or dredged from the Kerguelen Plateau basement by their relatively evolved and alkalic character (Fig. F60B) and high concentrations of incompatible trace elements (Fig. F63). They more closely resemble the alkalic volcanic series from the Kerguelen Archipelago. The chemical trends shown by the Kerguelen Archipelago evolved suites have been interpreted to result from fractionation of feldspar, Fe-Ti oxides and accessory phases, such as apatite (Weis et al., 1993, 1998).

The mafic rocks from Site 1139 and the Kerguelen Archipelago generally overlap in K₂O, SiO₂, and Al₂O₃, but at a given MgO content, Site 1139 lavas trend to higher contents of TiO₂ (Fig. F66), Fe₂O₃* (not shown), and Y (Fig. F67) than their counterparts from the Kerguelen Archipelago. In the plots of SiO₂ and MgO vs. TiO₂ (Fig. F66), it is unclear whether these differences reflect real differences in magma compositions, caused perhaps by differences in crystallization histories, or result from alteration because we demonstrate that both silica and magnesium are affected by alteration. However, the high Y concentrations at Site 1139 likely indicate either a different parental magma composition or a different evolution trend than those for the Kerguelen Archipelago series. Primitive mantle-normalized trace element patterns of the mafic group (Fig. F65A) are strongly fractionated as indicated by Nb/Zr and Zr/Y ratios, but the Kerguelen Archipelago series extends to even higher ratios (Fig. F67). Even so, Nb/Zr ratios in the Site 1139 mafic group are significantly higher than those of other Kerguelen Plateau basement lavas. Therefore, there is no indication of the crustal component that was inferred in basalts from Sites 1137 and 738 on the basis of their low Nb/Zr ratios (Fig. F63).

The felsic group from Site 1139 (Units 1-5, 18, and 19) not only shows major element abundances offset from the Kerguelen Archipelago values, but also shows markedly different trends in TiO₂, Al₂O₃/TiO₂, and P₂O₅ vs. SiO₂ caused by the strong silica enrichment (Fig F66). The felsic rocks are strongly enriched in Nb, Zr, and Y, but show pronounced relative depletion of Ba, Sr, Ti, and P (Figs. F65, F66B, F67), which is the signature of felsic magmas that have evolved through the fractionation of feldspar, Fe oxides, and apatite (Weis et al., 1993). Because fractionation of these phases has also been invoked for the Kerguelen Archipelago series, the major element trends suggest that this fractionation continued to increase silica in the Site 1139 felsic magmas, whereas other elements had already been depleted by the removal of the fractionating phases (~0 wt% MgO, CaO, and P₂O₅) (Fig. F65; Table T11).

The Site 1139 basement sequence ranges from trachybasalt to trachyte and rhyolite and is the most alkalic and evolved suite of igneous rocks encountered during Leg 183. On the basis of macroscopic features, we conclude that this sequence erupted subaerially. In spite of the intense alteration that subsequently affected these rocks, petrographic and geochemical analyses reveal information regarding the igneous evolution of these lavas. The mafic lavas are generally aphyric and contain rare plagioclase phenocrysts and sanidine xenocrysts. The felsic lavas also contain sanidine phenocrysts, and the rhyolitic samples have quartz phenocrysts and microphenocrysts. The presence of the sanidine xenocrysts and disequilibrium textures within the mafic lavas and the occurrence of felsic lavas both above and below the mafic units suggest that magma mixing or wall-rock assimilation may have affected the trachybasalts and basaltic trachyandesites. However, geochemical evidence of magma mixing is not obvious from the current data set and will be the topic of further investigation.

Major and trace element data provide a compelling picture of feldspar, Fe-Ti oxide, and apatite fractionation in the petrogenesis of the Site 1139 lavas. In spite of the overprint of alteration, including evidence of silicification in Unit 18, the behavior of trace elements such as Ba, Sr, Zr, and Ti suggest that high degrees of fractionation are required to form the felsic lavas and volcaniclastic rocks. Shore-based isotopic data and mineral chemistry data will reveal whether the lavas were generated from similar parental melts or whether they originated from unrelated magmatic systems. We characterize postemplacement alteration as including both silica and rubidium loss and magnesium gain, in addition to mobilization of calcium, phosphorus, and highly incompatible trace elements (Fig. F62). The fluids responsible for the alteration seem to have had varying amounts of H₂O and CO₂.

Site 1139 volcanic and volcaniclastic units cannot have evolved from tholeiitic melts that are the dominant lava type at other Kerguelen Plateau sites. Instead, the Site 1139 suite appears to have a magmatic history similar to the evolved lava suites exposed on the Kerguelen Archipelago, where the evolved lavas have been interpreted as products of feldspar, Fe-Ti oxide, and apatite fractionation from alkali basalt and basanite parental magmas. Primitive mantle-normalized ratios of Nb/Zr relative to Zr/Y provide, however, a preliminary indication that some aspects of the magmatic evolution of the Site 1139 sequence are different from the Kerguelen Archipelago series (Fig. F67). We expect that isotope analyses will further constrain similarities and differences between the formation of Site 1139 and Kerguelen Archipelago lavas.