This paper revisits the dynamic optimal taxation results of Jones, Manuelli, and Rossi (1993, 1997). They use a growth model with human capital and find that optimal taxes on both capital income and labor income converge to zero in steady state. For one of the models under consideration, I show that the representative household's problem does not have an interior solution. This raises concerns since these corners are inconsistent with aggregate data. Interiority is restored if preferences are modified so that human capital augments the value of leisure time. With this change, the optimal tax problem is analyzed and, reassuringly, the Jones, Manuelli, and Rossi results are confirmed: neither capital income nor labor income should be taxed in steady state.