This paper examines a monopoly facing an uncertain demand curve. In a dynamic setting this uncertainty may lead the firm to experiment with the quantity supplied in the market. Adjusting initial supply away from its myopically optimal level eventually increases the informativeness of observed market outcomes and hence increases future profits. Conditions for optimal experimentation are discussed within a two period model. Finally, the model is extended to an infinite time horizon. It is shown that the optimal policy entails experimentation and that learning via experimentation leads to complete learning in the long run.