The best documented San religious practice is the healing dance, described at length for the Ju/'hoansi by Lorna Marshall and by others since. Women sit in a circle at the fire, clapping and singing the medicine songs; men dance around them through the night; and in some of the dancers a potency called n/om, held to reside in the lower body, heats and rises. A healer in whom it has risen lays hands on each person present in turn, drawing sickness out, and may cry out, tremble or collapse. This is a description of what has been published about the dance, not an account of how it is done: the knowledge of raising and handling n/om is transmitted person to person within San communities, from an established healer to someone learning, and it is not the atlas's to pass on.
The other body of evidence is the rock art of southern Africa, tens of thousands of painted and engraved sites. Since the 1970s the dominant interpretation, argued most forcefully by David Lewis-Williams, has read much of the imagery as depicting the trance experience of the dance, using the /Xam texts and the Ju/'hoansi ethnography as a key. The reading has been productive and is not universally accepted; critics have argued that it flattens a very long and regionally varied tradition into a single explanation and reads recent ethnography backwards into images made thousands of years earlier by people about whom nothing else is known. The atlas records that argument as open.
What sits behind both is a limit. The religions of the San groups of the southern Cape whose languages ended before anyone wrote them down are not recoverable. The Bleek and Lloyd collection exists because two people happened to begin that work and were able to continue it for fourteen years. Where nobody began, there is nothing.