This article argues that one may use textual variation as a means of achieving a closer appreciation of the text studied, an understanding of the processes that shaped the textual tradition, and a discrimination of which variants are likely to be the author's own composition, rather than introduced by scribes. Central to this process is a determination of exactly which variants derive from the author. As well as traditional literary judgeinent, one may use analysis of the whole textual tradition (employing computer-assisted methods to gather and analyse all data of textual variation) to create a hypothesis of textual relations througliout the tradition, and hence a view of what manuscripts (and what combinations of manuscripts) are most likely to preserve readings archetypal to the whole tradition, and so most likely to be of the author's own composition.