This pager uses data on 37 chemical products to test a number of hypotheses about the learning curve and industrial price behavior. The results document a strong and consistent learning effect. Learning is found to be a function of cumulated industry output and cumulated investment rather than calendar time. Standard economies of scale appear significant but small in magnitude relative to the learning effect. Variations in the slope of the learning curve are linked to differences in R&D expenditures and capital intensity. Market concentration is found to strongly influence price flexibility and the timing of learning-related price changes.