The point-count data presented in Table T1 indicate that the ash recovered from Shatsky Rise during Leg 198 consists of predominantly colorless glass exhibiting a range from blocky to bubble-wall to pumice textures (e.g., Fig. F3) and lesser amounts of crystal phases. Plagioclase is the dominant crystal component in these ashes, ranging as high as 12% of the total composition but averaging <3%. Traces (<2%) of quartz, pyroxene, amphibole, biotite, and K-feldspar were also noted. The dominant components are colorless vitric pyroclasts, but as much as 5% of the vitric components exhibit tan, brown, or black groundmass. In turn, these darker fragments are generally nonvesicular and blocky (Lvv) (Fig. F3), which, coupled with their color, suggests that they are mafic to intermediate in composition. The colorless glassy fragments were classified according to their stain and texture/morphology including degree of vesicularity (Fig. F3). They were mostly either K stained or exhibited both Ca and K stain, but in a few samples the glass is predominantly Ca stained. The relative proportions of Ca- and K-stained samples are shown in the ternary plot in Figure F4. The few unstained colorless fragments noted in Table T1 may have been completely embedded in protective epoxy that prohibited their staining.
The vitric clast classification scheme (Fig. F3) is based on variations in shard shape and degree of vesicularity of glass, ranging from nonvesicular to highly vesiculated pumice. We observed a wide assortment of vitric grain assemblages within the samples, so several schemes of recalculated parameters (Tables T2, T3) were devised to illustrate the textural variability of the ash beds. From these, ternary plots showing relative proportions of ash texture were constructed (e.g., vesiculated vs. shards vs. blocky pyroclasts; slightly/moderate vesiculated vs. highly vesiculated/pumice vs. shards; and U-shaped bubble wall vs. Y-shaped bubble wall vs. platy shards) (Figs. F5, F6, F7). Plots for the subsets of shard populations according to stain (K, Ca, and Ca + K) provide less information and are presented elsewhere (Gadley, 2005). In the first and second plots (Figs. F5, F6), the data spreads are consistently large with a similar data distribution for all stain combinations and no distinct populations. Both of these plots illustrate the tendency for the samples from Site 1208 to be more vesiculated. The third plot (Fig. F7) shows the overall dominance of platy over Y- or U-shaped shards.
The variability among the textural and compositional proportions of the ash beds shown in Figures F5, F6, and F7 and Table T3 suggests that these attributes could be used to "fingerprint" ash for detailed bed-by-bed correlation. First approximations of correlativity are provided by the age estimates (biostratigraphic and magnetostratigraphic) in Table T1. Initially the ash layers were correlated based on age and then quantified based on textural composition. Correlations between the ash layers of similar age within holes at a site as well as from site to site were evaluated (Table T4; Fig. F9). The shallowest three ash layers found in Holes 1212A and 1212B were the first to be evaluated using this technique because this series of ash beds appeared to be very similar in both holes based on lithology and magnetic susceptibility (as observed by K.M. Marsaglia and other shipboard scientists). Correlation tests for the three ash layers within Holes 1212A (Samples 198-1212A-2H-6, 77 cm; 3H-1, 92 cm; and 3H-2, 95 cm) and 1212B (Samples 198-1212B-2H-7, 65 cm; 3H-1, 144 cm; and 3H-3, 12 cm) are good to excellent (Table T4; Fig. F9). Furthermore, the correlation test for a pair of slightly older ash beds in these two holes (Samples 198-1212A-4H-4, 37 cm, and 198-1212B-4H-4, 58 cm) also rates as excellent. At Site 1209 ash layers similar in age to the two sets discussed above do not correlate as well compositionally, but these exhibit a poor compositional correlation between holes (1209A and 1209B). Sample 198-1207A-2H-4, 59 cm, shows a good correlation to Sample 198-1208A-5H-2, 1 cm. Samples 198-1209A-2H-6, 72 cm, and 2H-6, 85 cm, provide good correlations to Samples 198-1209B-3H-1, 129 cm, and 3H-2, 11 cm, and Sample 198-1209A-3H-1, 30 cm, is an excellent correlation to Sample 198-1209B-3H-3, 88 cm. Sample 198-1209A-3H-1, 30 cm, is also good correlation to its depositional counterparts, Samples 198-1209B-3H-3, 88 cm, 3H-4, 48 cm, and 3H-134, 4 cm. Samples 198-1209A-4H-3, 24 cm, and 4H-5, 55 cm, are good and excellent correlations to Samples 198-1209B-4H-6, 104 cm, and 5H-1, 110 cm, respectively. Although very similar in both age and depth, Samples 198-1210A-11H-1, 142 cm, and 198-1210B-10H-6, 42 cm, show only a moderate correlation.
The correlations in Figure F9 are not unique; to show this we also tested samples of different ages. A total of 113 tests of compositional/textural similarities were evaluated (Table T4). Of these, 19% showed excellent compositional similarity, 28% showed good compositional similarity, 27% showed moderate compositional similarity, and 26% showed poor compositional similarity. These percentages are similar to those generated for the age equivalence/correlations in Figure F9. Results for various sample pairs with excellent to good compositional/textural similarities are shown in Figure F10.