A total of 58 samples of unconsolidated ash were collected for this study (Table T1; Fig. F2): 3 from Hole 1207A, 18 from Hole 1208A, 5 from Hole 1209A, 8 from Hole 1209B, 7 from Hole 1210A, 1 from Hole 1210B, 1 from Hole 1211A, 2 from 1211B, 5 from 1212A, 6 from 1212B, and 2 from Hole 1213A. The samples were acquired during the leg and on a subsequent trip to the Gulf Coast Core Repository at Texas A&M University in College Station, Texas. Each sample was first air-dried and then dry sieved to separate the sand-sized fraction (0.063–2 mm) from finer (<0.063 mm) material. Several of the collected samples were very fine and contained no sand-sized fraction (Table T1). Sand fractions for 41 of the samples were then epoxied to glass slides and ground to a thickness of 30 µm. Thin sections were etched with hydrofluoric acid and stained using the procedure outlined by Marsaglia and Tazaki (1992). The staining served to differentiate Ca from K feldspar, but other siliceous components also stain. Thus, this technique provides information on the Ca and K content of vitric components as well.
Point-count categories included mineral grains (crystals) and glass fragments, subdivided by texture/morphology, degree of vesicularity, and composition (color/stain) (summarized in Table T2). Much of the description and interpretation of shard morphology has been based on observations of ash using a scanning electron microscope (e.g., Heiken and Wohletz, 1985). In thin section, the 30 µm thickness provides some idea of the three-dimensionality of the textures we describe but, in essence, we are dealing with two-dimensional slices or views. Therefore, we devised a classification scheme (Fig. F3) that makes use of our two-dimensional slices of three-dimensional fragments and allows us to expand on the cuspate, platy, and pumice shard textures outlined by Fisher and Schmincke (1984).
An automated-stage point-counting system attached to a petrographic microscope was used to determine compositional modes from the thin sections. A total of 400 points were counted on each section using the Gazzi-Dickinson method (Dickinson, 1970; Ingersoll et al., 1984); bioclasts were noted but were not included in the 400-point total. Raw point count data are presented in Table T1 and were recalculated in an Excel spreadsheet program using the parameters listed in Table T2. Recalculated parameters are listed in Table T3 and plotted in Figures F4, F5, F6, and F7. In cases where the volcanic material was of insufficient quality or quantity to count 400 grains, the volcanic components in the samples were qualitatively described (Table T1). Age is an important parameter in ash bed correlation. To facilitate ash bed correlation, the ages of Leg 198 ash beds were estimated using two different methods (Table T1). The first method extrapolated ages from age-depth plots created by shipboard scientists using biostratigraphic (nannofossil and foraminiferal) datums (Bralower, Premoli Silva, Malone, et al., 2002). The second method extrapolated ages from shipboard paleomagnetic data as interpreted by Bralower, Premoli Silva, Malone, et al. (2002) and postcruise data (H. Evans, pers. comm., 2005) according to the geomagnetic polarity timescale of Cande and Kent (1995). As Natland (1993) points out, linear interpolations of age using such methods require a constant rate of sedimentation and lack of vertical core distortion. We used these data to target likely correlative pairs of ash beds from different holes or sites. Then, for each ternary plot, the minimum percent difference (any of the three parameters) between the two ash samples was entered into a correlation table (Fig. F8, Table T4). An average minimum percent difference was calculated based on the number of ternary plots used in each case (number changes because some ash beds do not contain all glass stain types). This number ("Total percentage" in Table T4) provides a means to evaluate the likelihood of correlation. The correlations were quantified as being excellent (0%–3%), good (3%–5%), moderate (5%–8%), or poor (>8%). We also used this method to test the similarity of ash beds that were not time-correlative.
Additional data were collected from marine tephras recovered from the Izu-Bonin intraoceanic arc during ODP Leg 126. The samples were previously described in Marsaglia (1992). For this study, selected samples from Leg 126 were recounted using the same ash classification scheme used for the Leg 198 samples (Fig. F3). In the Leg 126 samples, a total of 300 points (vitric components) were counted instead of the 400-point total counted in the Leg 198 samples owing to lower vitric to crystal ratios in the Leg 126 samples (Table T1). The data were recalculated using the parameters in Table T2.