This paper argues that the hypothesis that there are facts is ungrounded. I first introduce a series of important theoretical distinctions to say what facts are not – and to avoid misunderstandings as to what I take to be facts, states of affairs and relations. Then I present the so-called problem of the glue, which is linked to Bradley’s regress. Finally, I propose a stronger version of the problem of the glue, which I call the problem of directional glue, with the aim of giving additional evidence that the problem of the glue is not solvable by the theories of facts which fall prey to it.