The rates and stoichiometry of hornblende dissolution were studied in batch reactors in 0.01 M HOAc-LiOAc buffers at pH 3.6-4.0 and 25°C for three consecutive weathering cycles of 36, 45, and 34 days (total 115 days). The dissolution rates were obtained for two particle size fractions (0.11-0.25 and 0.25-0.50 mm in diameter) with initial surface areas of 0.098 and 0.086 m2 g-1, respectively. Results showed that dissolution rates were nonlinear and nonstoichiometric in the early stage of reaction. The rates were greatest in the first cycle, decreased with time, and became more linear and more stoichiometric as the experiment proceeded into the second and third cycles. A very rapid release of Ca during the initial stages of the reaction was due to 0.91% CaCO3 impurity in the hornblende. Incongruence was greatest in the first cycle (0-36 days) with preferential release of Al, Fe, and Mg relative to Si and greater at higher pH. Stoichiometric dissolution was approached more rapidly at lower pH, suggesting that the extent of reaction is an important factor in attaining stoichiometric dissolution. By the third cycle (82-155 days), at pH 3.6, the reaction appeared to be stoichiometric. Nonstoichiometry resulted in the depletion Al, Mg, and Al, relative to Si with calculated mean depths of depletion of 7, 38, and 67 nm, respectively, at pH 4.0. Most of the depletion occurred during the first cycle (0-36 days). The pH dependence increased after the first cycle and order dependence of release of Si, Mg, and Al with respect to (H+) during the second and third cycles was not significantly different from the previously reported value of 0.7. For Fe, however, the order dependence was only 0.36, reflecting the lack of stoichiometry at the higher pH values. Comparison of our second cycle data with data for samples that were dried after an initial weathering treatment showed that drying produced higher dissolution rates and more incongruence. The drying effect was greatest for Fe, with an increase in rate by a factor of 5.6 and least for Si with an increase by a factor of 2.1. Air drying appears to cause the surface of the hornblende to revert to a condition similar to unreacted samples. Comparison of data for dissolution in 0.01 M acetate with data obtained in 0.01 M LiCl suggests acetate increases the rate by about a factor of 3-4. The rate at pH 4.0 for total base release was significantly lower than previous results for experiments conducted for <500 h.