The notion of redemption was first used by Bachelard when developing his conception of time and consciousness in L'Intuition de l'instant (1932), as he engaged in a polemic with Bergson.1 He used it again two years later in "Idealisme discursif" when reflecting on human being.2 This essay will examine what lies behind this notion and also its consequences, in particular in relation to surprise. It will argue that Bachelard's idea of the redemptive instant is bound up with his epistemology and central to his conception of human existence. At first glance, the idea of redemption formulated by a philosopher of science might well seem suspect, best played down for the sake of academic respectability. If though we conquer our discomfort, we shall find in this idea a guiding thread into important aspects of Bachelard's thinking. He refers to redemption a number of times towards the end of L'Intuition de l'instant, perhaps most strikingly in the following lines: "The full force of time is condensed in the innovatory instant when, by the pool of Siloam, our eyes are opened at the touch of a divine redeemer" (95).3 There is a curious movement here from definition to story. The phrase "the innovatory instant" clearly sums up Bachelard's conception of discontinuous time: why then does he add the story? The context of this book provides the key. L'Intuition de l'instant is ostensibly a study of a work by his friend and colleague Gaston Roupnel, entitled Siloe, hence the reference to Siloam.4 It is not however meant to be an objective study, as Bachelard makes clear in his introduction, in which he refers to his "deformations" of Roupnel's arguments (8), arguments he uses-or bends-in developing his own theory of time against Bergson. Thus, he explicitly takes the idea of redemption from Roupnel, for whom Art-with a capital A-is the way of redemption, redeeming being by creating it, by renewing language, perception, emotion, spirit, and surprise (97-98). At the root of this redemption, Bachelard says, is the instant, the Roupnelian instant being what he terms "an intensity of consciousness," intense because it is the synthesis of contradictory emotions (99). Bachelard holds on to this notion of the redemptive instant, transforming it however, and with it the associated ideas of consciousness and being. For him, it is not art but reason that is redemptive; it is consequently in terms of reason that time, consciousness, and being must be understood. Given that reason, or more precisely the nature of reason in twentieth-century science had preoccupied Bachelard's epistemological work prior to L'Intuition de l'instant, this metaphysical-and metaphorical-turn may well seem baffling.5 I wish to argue however that it is a function of his epistemology, and more specifically of his ideas of rectification and approximate knowledge. In his first work, his Essai sur la connaissance approchee (1928), Bachelard rejects deductive, a priori reason and with it idealism because these can neither explain "the continual movement, the continual progress of scientific knowledge" nor account for "the fundamental incompleteness of knowledge" that he regards as "an epistemological postulate" (13).6 In his attempt to understand how reason works in modern science, we note his emphasis on "the undeniable existence of error," error that "by its very nature cannot be eliminated," so constituting "one of the strongest objections to idealism" and obliging us "to make do with approximations" (13). This notion of ineradicable error is important in Bachelard's thought, underlying the conception he develops here of the interdependence of reason and reality, a conception he will later dramatize by the metaphor of redemption. Why though insist on error when surely accuracy is the aim of science? The answer lies, I suggest, in Bachelard's ideas of newness and surprise. The word "surprise"-etonnement-is used just once in this book, but tellingly.7 Referring to microphysics, Bachelard declares that "The infinitely small is the geometric centre of our surprise. …