Geochemical techniques reliant upon bulk sample analysis are generally not thought suitable for tephrochronological purposes, albeit that they may satisfy petrological requirements to characterize geochemical provinces. Grain discrete methods are required in order that detrital contamination or the variable influence of (micro)phenocryst phases can be excluded. EPMA is the most widely applied of the grain discrete techniques and is adapted to analyze the principal glassy phases of tephra in order to determine the most subtle geochemical differences both between the products of individual volcanic centers and of discrete eruptions within those centers.
The geochemical study of tephra by EPMA has afforded considerable advances in tephrochronological studies and remains the principal means of characterizing major element tephrogeochemistry of glass shards. The method is, however, not without its problems; the most significant of these for tephrochronology is the deleterious effect of the probe's electron beam in inducing alkali metal mobility within the matrix of the glass shards. This effect has been outlined and detailed by many authors both in natural and synthetic glasses (Keller, 1981; Nielsen and Sigurdsson, 1981; Hunt and Hill, 1993) and in albite (Autefage and Couderc, 1980). The most significant effect of this mobility, if inappropriate analytical conditions are utilized, is the apparent loss of sodium from the sample. In terms of the final geochemical quantitation, the tephra will appear to have low Na2O concentrations and slightly elevated SiO2 and Al2O3 concentrations. As the analytical conditions that impose these effects are commonly employed in standard petrological analysis, it is unsurprising that there are numerous examples of erroneous analyses in the literature. These have been explored by Hunt and Hill (1994, 1996) and Hunt et al. (1998). Given that the severity of the potential impact of poor analysis includes incorrect attribution of tephras to volcanic regions (Hunt and Hill, 2001) and potential confusion of key tephra isochrons, great care is required in adopting appropriate analytical conditions.
Geochemical analyses of tephras in this paper were performed by wavelength dispersive spectrometry (WDS) on a Cambridge Instruments Microscan V, under the following conditions: accelerating voltage = 20 kV, beam current = 15 nA (measured by Faraday Cup), peak count per element = 10 s, and a defocused (5-10 µm) beam. Na2O and SiO2 concentrations were determined first to minimize the effects described above. The knock-on effects of Na2O mobility on remaining elements, later in each analysis, were minimized by blanking the electron beam during spectrometer repositioning and reducing the electron beam exposure time to 50 s. Both oxides and simple silicates were used as standards. Corrections were made for counter dead time, atomic number (Z) effects, fluorescence (F), and absorption (A) using a ZAF procedure described by Sweatman and Long (1969). Andradite garnet and Lipari obsidian (Hunt and Hill, 1996) were employed as secondary standards to assess for stability of beam and reliability of glass data. These analyses are available on request. WDS data are found to be of much higher precision than those obtained through energy dispersive spectrometry (EDS) as demonstrated, inter alia, by Hunt and Hill (1996). As a consequence of this rigorous approach, we are confident that the analyses (Tables AT1, AT2) reflect the best possible data that can be obtained from tephras in the Japanese marine record, although we recognize that the data set could have been further enhanced by determination of volatiles (fluorine and chlorine) and minor elements (e.g., phosphorus).