Studying the history of mathematics, either as the main purpose of a course or as an auxiliary to it , ultimately relies on the reading of historical sources. Moreover, this reading should be considered as an activity per se that raises notoriously difficult pedagogical problems. The latter are related to the time necessary for this reading and to the intrinsic difficulty of the exercise. Taking these two observations as points of departure, I here propose some reflections aiming at showing the huge diversity of the possible ways of reading and at studying the correspondence between them and the particular nature of the sources for history of mathematics.