In 1849 the Philadelphia physician Samuel George Morton claimed that black people had smaller cranial capacities than white people.1 As Morton included raw data in his publications Stephen Jay Gould was able to reanalyse them 130 years after Morton's death.^1,2 What Gould found was that there were no grounds for Morton's claims: the data had been manipulated to support the investigator's prior hypothesis.

By publishing his data in full Morton was unconsciously engaging in sharing data. While providing access to the full details and results of experiments has long been considered a characteristic ethos of science,3 discussion of the principles and practicalities has been more recent.*RF 2,4-7* Various benefits of sharing data can be identified.^2,7,8 Firstly, as the collection of data generally constitutes the main cost of studies, the use of existing data to answer issues not directly addressed by the primary researchers represents an efficient use of resources. Sharing data can also reduce the burden imposed upon study participants, both for individuals and groups who are at risk of becoming overresearched. Secondly, replicating the findings of one study within other datasets increases their robustness. Thirdly, when an investigation is being planned, analysis of data from earlier studies can help in the formulation of the research question, …