This book examines several epistemological regimes in studies of numerical data over the last four centuries. It distinguishes these regimes and mobilises questions present in the philosophy of science, sociology and historical works throughout the 20th century.