Ever since the history of science emerged as a discipline at the heart of the Age of Enlightenment in the eighteenth century, Arabic science 1 —or at least certain sectors of it—have constantly been cited by the philosophers and historians of science. For the former, such as Condorcet, it was a guarantee of the continued progress of enlightenment during a period dominated by ‘superstitions and darkness’; for the latter, notably Montucla, Arabic science was necessary not for the sketching of a historical picture, but in order to establish the facts of the history of the mathematical disciplines. But philosophers and historians alike had received only the echoes of Arabic science, which had reached them through ancient Latin translations. We must, of course, beware of over-generalization or errors of perspective, and bear in mind that the sciences do not all maintain the same connection with their history; thus, of the mathematical sciences at least, astronomy is the one most firmly linked with its history, if only on account of the values of the observations that were recorded in books over the course of time and consulted by successors. Consequently Arabic astronomy assumed a privileged position, fairly rapidly attracting the attention of historians such as Caussin de Perceval, Delambre and, above all, J.-J. Sédillot—to name but French scholars—at the beginning of the nineteenth century.
Later in the course of the same century, the image of Arabic science began to change and to become shrouded with nuance. German Romantic philosophy, and the German school of philology which it engendered, had given considerable impetus to the philological and historical disciplines. The history of Arabic science gained from this rapid expansion, before becoming its victim: the study of Greek or Latin scientific texts could no longer eschew the Arabic works; 2 but the snare of history through languages—which we have stressed elsewhere 3 —enmeshed the history of Arabic science and bore it into retreat. De jure, therefore, it lost its right to exist, while de facto it was indispensable to historians, who referred to it increasingly.
This paradox, which is apparent not only in second-order studies but permeates a major work like Le Système du Monde by Pierre Duhem, is in