A total of 144 samples was collected during Leg 149 for use in this study. Sample intervals for Sites 897, 898, 899, 900, and 901 are shown in Figure 3. Unconsolidated samples were air-dried and then hand-sieved for the sand fraction; for each very fine to fine sand sample the sand concentrate was sprinkled onto a glass slide spread with epoxy; the epoxy was cured, and then the sand/epoxy mixture was ground to a thickness of 30 microns. For coarser samples, for which grain orientation might be a problem using this method, a second stage was added: the sand/epoxy mixture was ground flat, and then another slide was attached with epoxy to the smoothed surface and the original slide was ground away. Lithified samples were vacuum-impregnated with blue-dyed epoxy under high pressure prior to thin-section preparation. Thin sections were stained for potassium and calcium feldspar recognition using the method outlined in Marsaglia and Tazaki (1992).
Approximately two-thirds of the shipboard samples collected for this investigation proved unsuitable for sand provenance studies in that they contained an insufficient sand fraction. Many of the unsuitable samples are coarse silts/siltstones and bioclastic sediments/rocks. Only 50 samples contained sufficient siliciclastic sand for point-counting purposes (Appendix). For these samples, 250 to 300 points were counted per thin section using the Gazzi-Dickinson method of point counting (Dickinson, 1970; Ingersoll et al., 1984). The grid spacings used were larger than the maximum grain size, and thus fewer points were counted for some coarser samples and for samples with only a minor terrigenous component. Uniformity among operators was maximized by recounting the thin sections by multiple operators and by frequent cross-verification of grain types throughout the counting process.
Counted grains were placed into 22 monomineralic and polymineralic compositional categories (Table 1). Because the sand-sized bioclastic component in some samples was so great, bioclastic debris was not tallied in six samples ("nc" in the Bio category in the Appendix). Point-count categories and recalculated parameters (e.g., QFL, QmKP, LmLvLs) are defined in Table 1, raw point-count data are presented in the Appendix, and recalculated parameters are listed in Table 2. Recalculated parameters, means, and standard deviations were calculated using a Quattro Pro spreadsheet program. For each sample, mean grain (sand) size was petrographically estimated; these data are listed in Table 2.