The so-called Jīvakapustaka is a collection of 93 Āyurvedic prescriptions preserved in the incomplete tenth-century Dunhuang manuscript Ch. ii.003 of the British Library (shelf marks IOL Khot 87110). This manuscript features sections in corrupt Sanskrit verse alongside their Late Khotanese prose renditions and is written in a distinctive form of the Khotanese Documentary Brāhmī Script. Interpreting this manuscript presents multiple challenges, primarily due to the peculiar script type, the complexities in deciphering the Sanskrit text, and the bewildering spelling variants used for writing Late Khotanese. This article aims to provide a new edition, translation, and commentary on the medicated ghee known as Mahāvaideha ghr̥ta (JP 11), which is unique in having a Tocharian B parallel. It also offers a novel methodology for reconstructing the highly corrupt Sanskrit version, by integrating careful observance of its underlying metrical structure, due consideration of the amplified Khotanese rendition, and full awareness of the distorting effect of the spelling habits of the copyist. It will be demonstrated that the Sanskrit version of the Jīvakapustaka owes its “barbarous” nature to a non-native Sanskrit speaker, most likely the Late Khotanese scribe who also wrote the Khotanese section, approximating Sanskrit words to his own writing conventions and phonological system. This study provides the first-ever classification of the graphic, phonological, and grammatical errors made by the Late Khotanese scribe, emphasising their significance in elucidating the phonology of Late Khotanese. Furthermore, it also deals with the identification of the Sanskrit mantra attested at the beginning of the Tocharian manuscript.