This article argues that the external history of South African English (SAfE) points towards the merits of conceptualizing SAfE as the product of a three-stage koinéization process, the last stage of which takes place contemporaneously with the establishment of Johannesburg. This is at odds with the standard position, which views SAfE as an early-to-mid 19th-century variety with its characteristic features having been fixed during the earlier colonization of the Cape and Natal. This reconceptualization is, in turn, usefully employed to solve Trudgill's (2004) so-called “South African puzzle'': in essence, the postulation of SAfE as a late 19th-century English explains why START-Backing has occurred in SAfE but not in the closely related Australasian varieties.