Tephra geochemistry has been established for selected Pliocene-Pleistocene tephras from both Leg 186 Sites (1150 and 1151). These have been compared with published data for named/previously identified tephras both on land and in the marine record from the Japan region (Cadet and Fujioka, 1980; Fujioka et al., 1980; Furuta and Arai, 1980a, 1980b; Machida, 1999; Machida et al., 1984, 1985, 1987; Pouclet et al., 1986). As the majority of previously characterized tephras are principally found in the Quaternary record (see Machida, 1999), we have confined our discussion of potential correlations to the Pleistocene succession.
The data considered in this study comprise multiple electron probe analyses of tephric glass shards. Although the Atlas of Japanese Tephrochronology (Machida and Arai, 1992) presents multiple analyses from discrete tephras, difficulties arise in comparing tephrogeochemistry from our study with many published data sets, principally due to a widespread tendency to publish single mean data points and standard deviations rather than representative individual analyses (e.g., Machida, 1999). This difficulty is compounded further by the publication of normalized data without an indication of assumed volatile (including water) content—thereby losing an indication of (1) the potential effects of alkali metal mobility (loss) and (2) the effective enhancement of the higher-abundance element oxides (e.g., silica and alumina). An additional source of difficulty arises in drawing comparisons with Japanese tephrochronological data that are based on tephra characterization by refractive index (RI) of glass shards. Where only RI data are available, comparison between many onland Japanese tephra is often problematic, as ODP/DSDP tephra studies are frequently based on EPMA data alone. As RI is based on the density of the glass, it is possible that any single range of RI values could arise through two or probably many more discrete geochemical compositions; Furuta et al. (1986) suggest that the transition metal (and particularly ferric and ferrous) oxides control RI. Additionally, the effects of variable shard hydration on RI values must be considered. These factors appear, initially, to suggest that EPMA data for shard geochemistry may have greater utility in distinguishing and correlating tephras than RI methods. Therefore, if the RI approach is to gain international acceptance, a widely disseminated rigorous comparative exercise is required between RI and EPMA data sets. The work of Furuta and Arai (1980a, 1980b) and Furuta et al. (1986) begins to address this issue, but their results (see Furuta et al., 1986, p. 140, fig. 6), from a limited number of tephras, can be interpreted as indicating greater problems with RI as a tool in tephra correlation, particularly compared with EPMA.
In different volcanic regions, particular element pairs determined by EPMA prove useful in distinguishing between tephras produced from the same or different eruptive centers. In Iceland, for example, FeO/CaO ratios and FeO/TiO2 ratios are frequently the most useful indices for identifying particular tephras. In Japan one of the most useful oxide pairs in this approach is TiO2/K2O (see Machida et al., 1987), although additional examination, in particular of MgO, FeO, and CaO, offers further assistance. In attempting to establish tephra correlations, all major elements have been examined from readily available published data and our own extensive analysis of Leg 186 tephras. For each of the 59 tephras whose geochemical compositions are reported herein (see Tables AT1, AT2), up to 30 individual shards have been analyzed. A minimum number of 15 analyses was sought and where this was not achieved it reflects the number of shards exposed on the probe slide surface and available to the electron beam. Data that clearly reflect sample damage during analysis, with subsequent low totals, have been removed. This problem is particularly acute in highly vesicular (pumiceous) thin-walled shards.