We owe the effectiveness of the scientic enterprise in large part to the social contract under which sci- entists publish their ndings in such a way that they may be con rmed or refuted and receive credit for their work in return. Because of the limitations of the printed page, data have been largely left out of this arrange- ment. We have grown accustomed to reading papers in which tables, figures, and statistics summarize the underlying data, but the data themselves are unavailable. There are exceptions, such as DNA sequences, for which there exist specialized public repositories that authors are required to use. But the vast majority of data types do not have such repositories.