Most studies of the grain size of tephra fall deposits in marine sediments have relied on a complete analysis of all components (Kennett, 1981). However, the extensive alteration of many of the Leg 165 tephra fall layers makes this approach impossible. In spite of the degradation of glass shards in the layers as they are converted to smectite, the majority of phenocrysts in the layers appear to remain relatively unaltered. Their size is thus a remnant proxy of the original size distribution of the entire layer. This, however, is based on the following assumptions: (1) the fragmented source mixture from an explosive eruption contains a subpopulation of crystals of sufficient size range to supply crystals that are aerodynamically equivalent to the largest juvenile (glassy) particles at distances >100 km downwind; and (2) the largest crystals are likely to be found at the base of a tephra fall layer. The first assumption is supported by many studies of terrestrial and marine tephra fall deposits in which phenocrysts are a ubiquitous component of distal deposits and are of a size that indicates aerodynamic equivalency in comparison to the less dense but larger juvenile components of the deposits (e.g., Carey and Sigurdsson, 1982; Cornell et al., 1983; Ninkovich et al., 1978). The second assumption is supported by observations of crystal-rich bases for many marine tephra fall deposits (Kennett, 1981) and the larger contrast in terminal settling velocity between glass shards and crystals when they are settling in water as compared to air (Cashman and Fiske, 1991). This difference in settling velocity would lead to an enhanced separation of the components during settling through the water column to the seafloor. Crystals would settle out ahead of aerodynamically equivalent glass shards and be concentrated at the base of layers.
An advantage to using crystal sizes as a proxy of grain size, and thus an index of dispersal, is that this component has a relatively constant density (~2.7 g/cm3), and crystal shapes are more uniform (aspect ratio of ~0.6). Comparison of whole sample grain-size distributions can be problematic because the morphologies and compositions of the dominant glass shards can be quite variable and influence the settling velocity in a complex way (e.g., Wilson and Huang, 1979).
Maximum feldspar sizes were measured in 26 tephra fall layers of early Miocene to Pleistocene age from Site 998 (Table 1). The long and intermediate axes were measured for each crystal. A typical set of measurements from a single sample is shown in Figure 2. The distributions are generally symmetrical, and from these the median and standard deviation of the measurements were calculated for each sample (Table 1). Median sizes of the longest crystal axis range from 50 to 300 µm, with the coarsest layers found in the lower to middle Miocene sites (Fig. 3). The majority of layers fall in a relatively narrow range, with median crystal sizes between 150 and 220 µm. There does not appear to be any correlation between the thickness of the layers and the maximum crystal sizes (Fig. 4). This result is somewhat surprising because it might be expected that thicker layers would correspond to larger eruptions and that these would be able to disperse coarser particles over a greater area.
For the Miocene as a whole, the complete data set of crystal measurements were combined to calculate the average dispersal properties of the tephra fall layers (Fig. 5). The median of the maximum feldspar sizes for the Site 998 Miocene tephra layers is 190 µm.
Site 999 contains the highest accumulation rate of volcanic ash and the most number of tephra layers in the western Caribbean sites. Feldspar size measurements were carried out on 36 tephra fall layers of Miocene to Pleistocene age (Table 2). The median size of the 20 largest crystals varies from ~100 to 350 µm. There is also a general correlation between the median size of the crystals and the ash accumulation rate at this site (Fig. 6). The coarsest layers are in the middle to upper lower Miocene, similar to the size variation observed at Site 998 (Fig. 3). This correlation may be an indicator of (1) more energetic eruptions during this period, (2) increased wind strength, or (3) a higher probability of sampling coarser layers within sequences with a higher number of tephra fall layers.
As with Site 998, there does not seem to be any correlation between tephra layer thickness and the size of the largest feldspar crystals at Site 999 (Fig. 7). The median size of the Miocene fall layers as a whole is virtually identical to that of Site 998, with a value of 190 µm (Fig. 8).
Site 1000 did not penetrate beyond the Miocene sediments ending in sediment with an age of ~20 Ma. Feldspar measurements were carried out on 19 tephra fall layers (Table 3). This site exhibits the largest variation in median size of the 20 largest crystals, ranging from ~150 to ~600 µm. The median size is well correlated to the ash accumulation rate at this site, with the coarsest layers found to be in the upper lower Miocene (Fig. 9).
In contrast to Sites 998 and 999, there does appear to be some correlation between the median size of the largest feldspar crystals and the thickness of the tephra fall layers (Fig. 10). This site does, however, have a greater range in tephra fall thicknesses than Sites 998 and 999, with one layer up to 55 cm thick. The median size of the Miocene fall layers as a whole is slightly coarser than those at Sites 998 and 999, with a value of 260 µm (Fig. 11).