The time-gap argument is one of the classic arguments to introduce sense-data. It is based on the fact that we can perceive very distant objets, such as stars, which may have disappeared when we perceive them. Those in favour of the time-gap argument conclude from this that, in such cases, the objet of our perception can not be the extinct star, but sorne entity of a different kind: a sense-datum. They go on to argue that, as there is no relevant difference between such perception and the pereeption of nearby objets, we always perceive sense-data. The usual reply tothis argument is to mantain that there is no contradiction implied in the idea of seeing now that happened in the past. Moltke S. Gram -in his book Direct Realism. A Study of Perception- rejects this solution and proposes an alternative reply. In my opinion Gram' s new reply, in spite of introducing sorne interesting distinctions, is not only inferior to the classic solution but runs the risk of being used to reintroduce the conclusion of the time-gap argument. In this article I seek to show that Gram 's arguments are wrong and to establish the correctness of the classic reply to the time-gap argument.