『Abstract
　Exhumation of the Himalayan-Tibetan orogen is implicated in the marked rise in seawater ⁸⁷Sr/⁸⁶S ratios since 40 Ma. However both silicate and carbonate rocks in the Himalaya have elevated ⁸⁷Sr/⁸⁶S ratios and there is disagreement as to how much of the ⁸⁷Sr flux is derived from silicate weathering. Most previous studies have used element ratios from bedrock to constrain the proportions of silicate- and carbonate-derived Sr in river waters. Here we use arrays of water compositions sampled from the head waters of the Ganges in the Indian and Nepalese Himalaya to constrain the end-member element ratios. The compositions of tributaries draining catchments restricted to a limited range of geological units can be described by two-component mixing of silicate and carbonate-derived components and lie on a plane in multicomponent composition space. Key elemental ratios of the carbonate and silicate components are determined by the intersection of the tributary mixing plane with the planes Na = 0 for carbonate and constant Ca/Na for silicate. The fractions of Sr derived from silicate and carbonate sources are then calculated by mass-balance in Sr-Ca-Mg-Na composition space. Comparison of end-member compositions with bedrock implies that secondary calcite deposition may be important in some catchments and that dissolution of low-Mg trace calcite in silicate rocks may explain discrepancies in Sr-Ca-Na-Mg covariation. Alternatively, composition-dependent precipitation or incongruent dissolution reactions may rotate mixing trends on cation-ratio diagrams. However the calculations are not sensitive to transformations of the compositions by incongruent dissolution or precipitation processes provided that the transformed silicate and carbonate component vectors are constrained. Silicates are calculated to provide 〜50％ of the dissolved Sr flux from the head waters of the Ganges assuming that discrepancies between Ca-Mg-Na covariation and the silicate rock compositions arise from addition of trace calcite. If the Ca-Mg-Na mixing plane is rotated by composition-dependent secondary calcite deposition, this estimate would be increased. Moreover, when ⁸⁷Sr/⁸⁶S ratios of the Sr inputs are considered, silicate Sr is responsible for 70±16％ (1σ) of the ⁸⁷Sr flux forcing changes in seawater Sr-isotopic composition. Since earlier studies predict that silicate weathering generates as little as 20％ of the total Sr flux in Himalayan river systems, this study demonstrates that the significance of silicate weathering can be greatly underestimated if the processes that decouple the water cation ratios from those of the source rocks are not properly evaluated.』

1. Introduction
2. Study area
3. Sampling and analytical methods
4. Calculation of silicate- and carbonate-derived Sr fractions
　4.1. Correction for rainfall and hot-spring inputs
　4.2. Modeling two-component mixing
　　4.2.1. Tributary chemistry: Lesser Himalayan Deopryag catchment
　　4.2.2. Processes that may modify water chemistry
　　4.2.3. Calculation of Sr inputs to the Deopryag catchment
　　4.2.4. Influence of secondary calcite and trace calcite on Sr partition calculations
　　4.2.5. Estimates of uncertainties
　　4.2.6. Sr inputs in the Kanwana catchment
　　4.2.7. Sr inputs in the Deoban catchment
　　4.2.8. Calculation of Sr inputs in the High Himalayan Crystalline catchment
　　4.2.9. Calculation of Sr inputs in the Tibetan Sedimentary Series catchment
　4.3. Constrains on ⁸⁷Sr/⁸⁶Sr ratios of carbonate and silicate end-members
5. Calculation of Sr and ⁸⁷Sr/⁸⁶Sr fluxes
6. Conclusions
Acknowledgments
References