An empty signifier is, strictly speaking, a signifier without a signified. This definition is also, however, the enunciation of a problem. For how would it be possible that a signifier is not attached to any signified and remains, however, an integral part of a system of signification? An empty signifier would be a sequence of sounds, and if the latter are deprived of any signifying function the term ‘signifier’ itself would become excessive. The only possibility for a stream of sounds being detached from any particular signified while still remaining a signifier is if, through the subversion of the sign which the possibility of an empty signifier involves, something is achieved which is internal to signification as such. What is this possibility? Some pseudo-answers can be discarded quite quickly. One would be to argue

that the same signifier can be attached to different signifieds in different contexts (as a result of the arbitrariness of the sign). But it is clear that in that case the signifier would not be empty but equivocal: in each context the function of signification would be fully realised. A second possibility is that the signifier is not equivocal but ambiguous: that either an overdetermination or an underdetermination of signifieds prevents from fully fixing it. Yet this floating of the signifier still does not make of it an empty one. Although the floating takes us one step towards the proper answer to our problem, the terms of the latter are still avoided. We do not have to deal with an excess or deficiency of signification, but with the precise theoretical possibility of something which points, from within the process of signification, to the discursive presence of its own limits.