The current era of "big data" and "data-driven science" is a result, not only of technological innovations, but also of a number of deep epistemological, social, cultural, and political transformation. In explaining the rise of "big data," standard narratives focus mainly, however, on the growing technological capacity, since the late twentieth century, to produce, store, and transmit data. In the experimental sciences, high-throughput technologies applied to genomics, proteomics, or crystallography, have indeed augmented the rate of data production. But for data to accumulate in public databases and thus become visible, it also has to be considered scientifically valuable and belonging to the public sphere. In this paper Bruno Strasser explores the shifting meaning of "data" in protein crystallography and its consequences on data accumulation. He argues that labeling something as "data" produces a number of obligations, most importantly, to make it public in case it has served as a basis for granting authorship. In crystallography, like elsewhere in science and beyond, the category of data has expanded, include information which were previously considered private, thus contributing to the accumulation of public data and to the making of the "big data" era.