This paper investigates the etymology and morphology of the Khotanese verb meaning ‘to kill’, traditionally associated with the stem jsan-, and argues for the reconstruction of an older and synchronically lost verbal stem jsīn-. Although no finite verbal forms of this stem are directly attested, its former existence can be recovered through a small set of nominal and adjectival derivatives (i.e. gerundives). The analysis focuses on the morphophonological behaviour of forms such as jsīna- ‘killing’ and jsīñaa- ‘to be killed, condemned ‘to death’, whose vocalism and palatalisation patterns do not conform to those expected for regular *-Ôa- derivatives of jsan-. These anomalies are shown to be systematic rather than exceptional and point to a distinct historical source. On comparative Indo-Iranian grounds, the paper proposes that the relevant Khotanese forms continue a reduplicated Indo-Iranian verbal base *ǰa-ghn-, ultimately deriving from Proto-Indo-European *gʷhé-gʷhn– ‘to strike repeatedly, kill’. A phonological scenario involving the development of preconsonantal Iranian *ɣ into a glide and its subsequent monophthongisation accounts for both the long vowel ī and the absence of secondary palatalisation effects. The reconstruction of jsīn- sheds new light on the interaction between palatalisation, vowel development, and derivational morphology in Khotanese, and illustrates how obsolete verbal stems may survive exclusively in nominal domains.