Collective memory, social memory, professional memory: although these notions are in current use when we name the shared (or assumed to be shared) representations of the past, they are very ambiguous. The point at issue is to show how memories can become common to some or to all members of a group . In this paper, I shall base my arguments on the simplest situation imaginable: The sharing of a memory of an olfactory experience by two individuals, namely one of (...) my informants – a gravedigger – and myself. We can try to explain this shared olfactory memory, or the imagination of this sharing, by taking two facets into consideration. First of all, we must learn to take seriously those few seconds when a shared experience of the sensory world took shape between the anthropologist and his informant. Secondly, we must attempt to answer both of the following questions: what is the nature of this sharing? And what conditions make it possible? In the first part of this text, I examine the issue of the nature of sharing by weighing up its aspects as I consider protomemory, memory and metamemory. In the second part, I outline one of the essential conditions of sharing: the existence of what I call sociotransmitters. For the most part, my attention remains focused on that particular moment when the gravedigger said, “Smell this!” , and when what was at stake was to save the interaction. (shrink)