Data-driven sciences, or data-intensive research, are said to become the new norm of scientific endeavours. The sheer volume of empirical data combined with the emergence of methods of analysis grounded in machine learning would bring about the “end of theory”, i.e. the advent of a “fourth paradigm“ of science according to which the mere correlation of patterns inductively found in numerical datasets would replace causal models and theories. But behind this bold claim of a renewed and radical empirical inductivism, are data really new in the sciences? What can the history of science tell us about data practices and the many issues they raise? This talk endeavours to delineate a longer history of data in the sciences. From the 17th-century so-called “scientific revolution” and subsequent Enlightenment to 20th-century “Big Science” through 19th-century statistics, this historical overview will help shed new lights on the many epistemological, social, and political issues raised by the supposed turn to data in the sciences.