Sanskrit
The complete Sanskrit text is available in three successive editions, made by Mitra (1888), Wogihara (1932–35), and Vaidya (1960), from Nepalese manuscripts.5 The manuscripts themselves are of relatively recent date but are based on recensions dated to the eleventh or twelfth century.
Despite this late dating, the Sanskrit matches closely the fragmentary third-century manuscripts from Bamiyan mentioned above, even if the latter show more linguistic features characteristic of Buddhist Hybrid Sanskrit. It therefore appears to have been relatively stable over many centuries.
There are suggestions in the Tibetan accounts of the various stages by which the successive Tibetan translations reached their present form that the numerous Sanskrit manuscripts consulted by successive translators may have not only been slightly different but also fell into two broad groups: those made before and after a revision of the text to conform to The Ornament of Clear Realization.6 However, the extant Sanskrit versions show no evidence of belonging to more than one principal lineage.
The extant Sanskrit text was used as the principal basis for this English translation, for reasons explained below. It corresponds quite closely to the Tibetan translation.