The notion of duration as consisting of “parts” united by “creative synthesis” is a device, not for explaining how the transition from past to present really takes place (this does not need explaining since, “past” and “present” being mere abstractions, no transition between them actually takes place at all), but for enabling us to employ the abstractions “past” and “present” without constantly being taken in by their logical implications. The notion of “creative synthesis” as what joins “past” and “present” in a process of duration is an antidote to the logical implications of these two distinct terms: creative synthesis, unlike logical relations, is not external to the “parts” which it joins; “parts” united by creative synthesis are not distinct and mutually exclusive. Such a notion as this of creative synthesis contradicts the logical implications contained in the notion of parts. The notion of “parts” united by “creative synthesis” is really a hybrid which attempts to combine the two incompatible notions of logical distinction and duration. The result is self-contradictory and this contradiction acts as a reminder warning us against confusing the actual changing fact with the abstractions in terms of which we describe it and so falling into the mistake of taking it for granted that this changing fact must form a series of distinct stages or things or events or qualities, which can be repeated over and over again. At the same time there is no getting away from the fact that this changing fact lends itself to classification and that explanations in terms of abstractions really do apply to it most successfully. We are therefore faced with the necessity of finding some way of accounting for this, other than by assuming that the facts which we know directly consist of qualities which recur over and over again. CHAPTER III MATTER AND MEMORY We have seen that, according to the theory of change which is fundamental for Bergson’s philosophy, the changing fact which we know directly is described as a process of becoming which does not contain parts nor admit of repetitions. On the other hand this changing fact certainly does lend itself to analysis and classification and explanation and, at first sight at any rate, it is natural to suppose that whatever can be classified and explained must consist of qualities, that is distinct parts which can be repeated on different occasions. The problem for Bergson, if he is to establish his theory of change, is to show that the fact that a changing process can be analysed and classified does not necessarily imply that such a process must consist of distinct qualities which can be repeated. Bergson’s theory of the relation of matter to memory suggests a possible solution of this problem as to how it is possible to analyse and so apply general laws to and explain duration: it becomes necessary, therefore, to give some account of this theory.