Major elements taken from du Bray et al. (2005). http://pubs.usgs.gov/ds/2006/155/ GEOREF# 2006-021428 The authors report: Several different schemes are possible for reporting iron contents. In addition, reported abundances of ferrous versus ferric iron in these rocks are unlikely to represent magmatic values, because of oxidation during devitrification, zeolite facies metamorphism, and (or) postmagmatic hydrothermal alteration. Consequently, total iron abundances were recalcu-lated as ferrous iron oxide and denoted as FeO*. Interaction with postmagmatic fluids caused compositions of many West-ern Cascades volcanic rocks to change in other ways as well. In particular, many of these rocks were hydrated (as indicated by secondary clay minerals, sericite, and (or) chlorite), and others were affected by fluids that precipitated calcite. Both processes caused volatile contents of the affected samples to increase, and correspondingly caused relative abundances of all other constituents to decrease. Therefore, to facilitate meaningful oxide abundance comparisons among samples, all analyses were normalized to 100 percent on a volatile-free basis. The resulting data are reported in columns identified by SiO2n, TiO2n, Al2O3n, FeO*n, MnOn, MgOn, CaOn, Na2On, K2On, and P2O5n. All data are reported as weight percent. Total = One measure of major oxide analytical accuracy is how nearly the sum of the determined constituents approaches 100 percent. Consequently, the database includes a column that reports initial analytical totals as reported by the source. Some sources do not include totals; totals for these samples were computed and added to the database. Initial analytical totals reported in the sources were spot checked for accuracy; discrepancies were noted and corrected in a number of cases. Many sources present abundances for the oxides listed above but include no abundance data for volatile constituents. Initial analytical totals for these samples tend to be several to 5 or 6 percent less than 100 percent. Unfortunately, it?s impos-sible to determine whether these low initial totals result from inaccurate analyses and (or) unreported volatile constituent abundances.