This paper presents the empirical claim that sequences of conditionals with special properties known as Sobel Sequences (Lewis, 1973) can actually be divided into two distinct phenomena with distinct properties. One of these phenomena can be described in much the way Lewis originally does, though perhaps not exactly; I continue to term these Sobel Sequences. However, I show that one crucial property that has been claimed by many authors to be exhibited by Sobel Sequences, namely Unidirectionality, does not hold of true Sobel Sequences. I call the discourses that do show this property Lewis Sequences. Since Lewis Sequences have been included among the empirical quarry of recent work on conditionals in discourse, they have muddled the analysis of true Sobel Sequences. By unmuddling, I suggest that a more conservative semantic approach has a better chance to account for Sobel Sequences; furthermore I sketch a fully pragmatic account of Lewis Sequences which collapses their analysis with that of several other phenomena outside the terrain of conditionals.