In both younger and older pillow units, the alteration processes are mainly controlled by primary textural and structural features (i.e., the presence of impermeable glassy rims, the primary concentric structural zonation, and the vesicularity) and by the presence of radial, or irregularly distributed, fractures that crosscut the pillows from the outer glassy zone to the holocrystalline interior. Three main alteration environments with different degrees of alteration and secondary products have been distinguished: (1) alteration of glassy rims, (2) massive alteration along and around fractures (oxidation halos), and (3) pervasive alteration through diffusion.
The degree and the style of the glass alteration are strictly related to the presence of concentric and radial veins, which often form a complex network. The large set of veins crosscutting the glassy zone can be grouped in two main types on the basis of structural evidence: simple veins (Fig. 9A) and composite veins (Fig. 9B).
The simple veins (<0.1-0.5 mm) characterize the glassy rims of pillow basalts from all nine sites. They show a distinct symmetric layering, parallel to the original crack (Pl. 1; Fig. 3), characterized by sharp compositional and mineralogical variations (Fig. 9A). Where a complete layering is developed, three distinct layers are present, and they are characterized (respectively from the outer to the inner part) by brown undifferentiated palagonite, a dark brown intimate mixture of K-bearing clays and Fe-oxyhydroxides, and greenish brownish fibrous to granular cryptocrystalline Mg-rich smectites. In the adjoining selvages the alteration of the glass usually proceeds from (1) fibrous form, around cracks and veins, (2) to a mottled texture within the glass, and (3) to extensive areas of yellow-brown palagonite between the varioles in the variolitic zone.
The composite veins (2-5 mm) characterize samples from Sites 1206 and 1027 and developed as a consequence of repeated opening and crack filling along the median line of the primitive simple veins. The infilling secondary minerals are commonly symmetrically arranged, from the walls to the core of the vein, in the following sequence (Fig. 9B): (1) fibrous Mg-rich smectites, (2) fibrous radiating zeolites (mainly phillipsite), and (3) fibrous to blocky anhedral carbonates (calcite and aragonite).
The hyaloclastite breccias of Hole 1026B represent an exaggerated example of glass alteration; the glass shards smaller than 1 mm are generally completely altered to undifferentiated palagonite, whereas the largest clasts (1-6 mm) show a concentric zonation, and the extent of alteration decreases from rim toward the center (where sometimes unaltered isotropic glass is preserved; Fig. 9C). Clasts are cemented by a polymineralic matrix mainly consisting of the same mineral assemblages observed in the composite veins.
To define the major chemical variations during the alteration processes of glass crosscut by simple and composite veins, chemical analyses were performed by SEM-EDS. Chemical profiles indicate large variations between the four recognized zones (Fig. 10) in good agreement with textural and mineralogical observations (Fig. 9). The outer brown zone (Zone 4; Fig. 10), which represents the less evolved glass alteration, shows significant gains in K2O, SiO2, and Al2O3. In contrast Fe2O3t, MgO, CaO, and Na2O are strongly depleted (the last two elements remaining unchanged toward the central part of the vein). In the intermediate layer (Zone 3; Fig. 10), the strong enrichment of Fe2O3t and K2O is related to the formation of authigenic phases such as alkali-bearing clays (celadonite) and Fe-oxyhydroxides. The chemistry of the central part (Zones 2 and 1; Fig. 10) agrees well with the composition of the observed smectites, and no major variations between fibrous and cryptocrystalline granular part are evident. The chemical profiles across the altered glass around the composite veins and across the hyaloclastic fragments are generally comparable.
The composition of altered glass relative to the three main zones (from both vein types) normalized vs. the composition of the adjoining unaltered glass allows the evaluation of a general chemical balance for the fluid-glass interaction during the veining processes (Fig. 11). The major variations involve the increase of H2O and K2O, the strong loss of Na2O and a minor but significant depletion of CaO and MnOt. Total Fe2O3, MgO, NiO, and Cr2O3 show minor and comparable degrees of depletion, whereas SiO2 and Al2O3 remain almost unchanged.
The fracture systems that characterized almost all the analyzed pillow basalts represent the most important channel ways for the fluid circulation within the pillow units and consequently the preferential sites where fluid-rock interaction starts and develops into the adjoining rock. Fluid-rock interaction generates monomineralic or polymineralic coatings or fillings within fractures and variably sized oxidation halos in the adjoining selvages.
Several vein types and generations have been recognized throughout all Leg 168 pillow basalts (individual occurrences are recorded in the VEINLOG; see Appendix D of the Leg 168 Initial Reports; Davis, Fisher, Firth, et al., 1997).
We distinguish three main vein types on the basis of their geometric relationships: concentric veins, radial veins, and irregular veins. Most of the concentric and radial veins presumably represent filled shrinkage cracks that increase in frequency toward quenched surfaces. Irregular veins can be alternatively explained as filled shrinkage cracks variably distributed and oriented as a consequence of mechanical and lithological heterogeneities, or as secondary fractures developed in the "cold" rock during later spreading-related tectonism.
The concentric veins (<0.1-0.5 mm thick) are mainly concentrated in the glassy zone and subordinately in the outer hypocrystalline pillow zones, whereas radial and most of the irregular veins (<0.1-5 mm thick) crosscut the pillows from the outer glassy rims to the holocrystalline interior.
The mineralogical composition of the veins is extremely variable from site to site and in many cases strong variations occur even at the scale of the single sample. Nevertheless, two main vein types can be distinguished on the basis of the mineralogic assemblages of the filling products: veins characterized by oxidizing filling assemblages (V1) and veins characterized by reducing filling assemblages (V2).
V1 are common in pillow basalts from every site and are characterized by the following mineral assemblages, characteristic of the oxidizing style of alteration (as defined by Andrews, 1980): hematite, goethite, iddingsitic mixtures (Types 6-9; Table 2), celadonitic minerals (Types 3-5; Table 2), and Mg-rich saponites (Type 2; Table 2). These minerals are commonly arranged in submillimetric to millimetric symmetric layers, parallel to the original crack. In many cases, the transition between the different layers is graded and each layer progressively fades out toward the adjacent, suggesting dynamic growth with a progressive variation of the chemical conditions (Eh and pH) of the microenvironment. Although some exceptions have been recognized, the temporal depositional sequence can be summarized as follows: (1) Fe-oxides and Fe-oxyhydroxides and/or iddingsitic mixtures, (2) celadonitic minerals, and (3) Mg-rich saponites. The commonest variation from this general trend is the inversion between 1 and 2.
V2 have been recognized only in pillow basalts from Sites 1027, 1028, and 1032, and are characterized by "nonoxidative" mineral assemblages (Andrews, 1980) mainly represented by carbonates after Fe-rich saponite (Type 1;
Table 2). Sulfides (mainly pyrite) are irregularly distributed within this assemblage but mainly occur in the outer rims of the veins associated with saponite. In some cases, especially in veins from Sites 1027 and 1032, up to three generations of carbonates occur varying in texture and composition, from the rims to the median line of the vein. Columnar crystals, often deformed, commonly characterize the outer rims, after fibrous radiating sheaf, whereas the central part commonly displays a blocky arrangement of undeformed anhedral to subhedral crystals. Together with this textural variation, carbonates vary from almost pure aragonite (commonly intimately intergrown with saponite) to Mn-, Mg-, and Fe-bearing calcite (respectively, MnCO3 
15 mol%, MgCO3 9 mol%, and FeCO3 3 mol%) in the fibrous and columnar part of the vein. The blocky central part is often affected by recrystallization processes and is commonly characterized by almost pure calcium carbonates (mainly calcite). A significant example of the chemical variations in carbonates of the three generations is reported in
Figure 12 relative to Sample 168-1032A-12R-1, 9-16 cm (Piece 1).
In several cases from Sites 1027 and 1032 there is evidence that V2 mostly developed as a consequence of repeated opening and crack filling along the median line of the early celadonite + iddingsite ± Mg-rich saponite veins (V1). This is clearly shown by the presence of V1 millimetric fragments (mainly iddingsite and celadonite) arranged in regular trails running parallel to the vein walls along the central part of the V2 (i.e., within the blocky calcite).
Oxidation halos are present around V1 fractures and veins in pillow basalts from every site (with maximum abundance in samples from Sites 1026, 1027, 1029, and 1032) and are typically absent around V2 veins. These halos are a few millimeters to several centimeters thick and appear in hand specimens as dark borders on rock pieces that may be stained with orange or yellow. The percentage of the observed halos always decreases progressively from the outer quenched margin to the holocrystalline pillows interior. In every studied sample, oxidation halos always represent the zone of most intense alteration with respect to the nonhalo portions of the rock. Microscopic observations allow the recognition of three main types of alteration zones that, when present together, show the following spatial and temporal relationships (Fig. 13): reddish brown zone (RZ), greenish red to green zone (GZ), and greenish to light gray zone (TZ). The black halos found by other authors in pillow basalts from various oceanic settings (Böhlke et al., 1980, 1981, 1984; Alt and Honnorez, 1984; Alt et al., 1996; Buatier et al., 1989; Laverne et al., 1996) were not recognized in the studied samples.
The reddish brown zone (1-5 mm thick) always occurs immediately around fractures or veins of Type 1 (V1) and is characterized by a massive precipitation of secondary minerals, mainly represented by goethite, hematite, undifferentiated Fe-oxyhydroxides, iddingsitic mixtures, and subordinately by celadonite and celadonite-bearing mixtures. Partial or complete oxidation of titanomagnetite to titanomaghemite is also common, especially in samples from Site 1029. RZ are always more extensively altered than the other adjacent zones and the total amount of alteration, determined by point counting the secondary minerals in thin section, varies from 30% to 50% with maximum values variable between 50% and 60% in some samples from Site 1027. Secondary phases, often showing complex temporal and spatial relationships, occur as vesicles or miarolitic void fillings, pseudomorphic replacement of mafic minerals (mainly olivine), and irregular patches or disseminated aggregates within mesostasis. A statistical study of the mineralogical zonation of the vesicle fillings (Fig. 13A) shows that, despite some exceptions recognized in every site, the most common sequence of mineral formation in RZ is the following: (1) goethite or undifferentiated Fe-oxyhydroxides cryptocrystalline aggregates, (2) hematite lamellae, (3) iddingsite, (4) iddingsite-bearing mixtures, and (5) celadonite or celadonite-bearing mixtures. This temporal and spatial sequence agrees well with that observed within the adjacent veins (Fig. 13B).
The greenish red to green zone (1-2 mm thick) is located immediately adjacent to RZ and occurs systematically only in samples from Sites 1026, 1027, 1029, and 1032. The total amount of alteration varies from 30% to 40%, and secondary phases occur as in the reddish brown zone previously described. The most common minerals are celadonite and celadonite-bearing mixtures (both with iddingsite and Mg-rich saponite) associated with an extremely variable amount of iddingsite, Fe-oxyhydroxides, and subordinate amounts of Mg-rich saponite (Type 2; Table 2). The sequence of the alteration inferred by the most common mineralogical zonation of the vesicle fillings (Fig. 13A) suggests that celadonite and celadonite-bearing mixtures occur, in most of the cases, after Fe-oxyhydroxides and iddingsite. When present, the last phase to form is the Mg-rich saponite.
The greenish to light gray zone (1-3 mm thick) represents the transition between the oxidation halos and the nonhalo portion of the rock. It represents the zone with the most complicated spatial and temporal relationships between the different secondary phases, which are represented by celadonite, celadonite-bearing mixtures, Mg-rich saponite, Fe-rich saponite, and minor iddingsite. The alteration intensity is extremely heterogeneous, even at the scale of the single sample, and varies from <10% to about 30%. A temporal and spatial sequence of secondary mineral formation representative of all studied samples is very difficult to identify. Nevertheless the following switched sequences are two of the most common that were observed
(Fig. 13A): (A) iddingsite
celadonite Mg-rich saponite Fe-rich saponite; and (B) Fe-rich saponite Mg-rich saponite celadonite iddingsite. Sequences (A) and (B) can be repeated more than one time even in the same vesicle. The complex relationships between secondary phases of this zone can be explained as the result of local equilibrium variations of the microenvironment that presumably occurred during the transition from the oxidizing alteration, in a water-dominated system, to the reducing alteration in a rock-dominated system.
Because oxidation halos flank only V1 veins and fractures and are typically absent around later V2 carbonate-bearing veins (Fig. 13B), it seems that they have developed early and always before the reducing alteration marked by the pervasive precipitations of Fe-rich saponites, carbonates, and sulfides in the gray interior (Fig. 13). The concentric configuration and the sharp boundaries between the different zones within the halos suggest that they developed under the control of diffusion from the solutions that occupied the fractures, which penetrated into the rock roughly perpendicular to the original fractures. Because the alteration products are mainly represented by Fe-oxyhydroxides and K-bearing clays, it is likely that the alteration phase related to the formation of the halos is triggered by normal or slightly evolved seawater.
The gray rocks that are unaffected by oxidation halos show a degree of alteration and a secondary mineralogy extremely variable from site to site. Generally their alteration intensity increases systematically with the distance from the ridge axis (i.e., from younger to older pillow basalts). The only exception is represented by the pillow basalts from Site 1028, which in some cases show a degree of alteration comparable to the oldest site drilled during Leg 168 (i.e., Site 1027).
In this portion of the pillow interior, the dominant mechanism that triggers the alteration processes is pervasive fluid seepage occurring primarily along the grain boundaries of the igneous minerals and primary voids. Alteration is mainly concentrated around phenocrysts, within gas and segregation vesicles, and miarolitic voids.
Fe-rich saponites and carbonates are the main secondary phases characterizing this stage of the alteration, but carbonates are completely absent in pillow basalts from Sites 1023 to 1025, 1029, and 1031. They both occur as pseudomorphic replacement of olivine (Pl. 1; Fig. 4) and sometimes of clinopyroxene, as dense patches within mesostasis, and as vesicle and fracture fillings and coatings in all the studied samples. Talc is a subordinate secondary product and has been recognized in Sites 1026 to 1028 and 1032, whereas sulfides are always a minor constituent. As evidenced by the spatial variations in vesicle and vein fillings (Fig. 13) and by the temporal relationships of the crosscutting veins, the carbonates seem to be always the last secondary minerals to form. The formation of talc suggests that a significant increase of Si activity occurred at Sites 1026 to 1028 and 1032 in the latest stages of alteration.
