It is divided into two parts, mDo (Sutra) and rGyud (Tantra), besides a volume of hymns and an index.  The same method of division is really applicable to the Kanjur, for the Tibetan Dulva is little more than a combination of Sutras and Jatakas and sections two, three, four and six of the Kanjur are collections of special sutras.  In both compilations the tantric section appears to consist of later books expounding ideas which are further from the teaching of Gotama than the Mahayanist sutras.

To the great majority of works in both collections is prefixed a title which gives the Sanskrit name first in transcription and then in translation, for instance “In Sanskrit Citralakshana:  in Tibetan Ri-moi-mthsan-nid."[992] Hence there is usually no doubt as to what the Tibetan translations profess to be.  Sometimes however the headings are regrettably brief.  The Vinaya for instance appears to be introduced with that simple superscription and with no indication of the school or locality to which the text belonged.

Although the titles of books are given in Sanskrit, yet all Indian proper names which have a meaning (as most have) are translated.  Thus the name Drona (signifying a measure and roughly equivalent to such an English name as Dr. Bushell) is rendered by Bre-bo, a similar measure in Tibetan.  This habit greatly increases the difficulty of reading Tibetan texts.  The translators apparently desired to give a Tibetan equivalent for every word and even for every part of a word, so as to make clear the etymology as well as the meaning of the sacred original.  The learned language thus produced must have varied greatly from the vernacular of every period but its slavish fidelity makes it possible to reconstruct the original Sanskrit with tolerable certainty.

I have already mentioned the presence of translations from the Pali.  There are also a few from the Chinese[993] which appear to be of no special importance.  One work is translated from the Bruza language which was perhaps spoken in the modern Gilgit[994] and another from the language of Khotan.[995] Some works in the Kanjur have no Sanskrit titles and are perhaps original compositions in Tibetan.  The Tanjur appears to contain many such.

But the Kanjur and Tanjur as a whole represent the literature approved by the late Buddhism of Bengal and certain resemblances to the arrangement of the Chinese Tripitaka suggest that not only new sutras but new classifications of sutras had replaced the old Pitakas and Agamas.  The Tibetan Canon being later than the Chinese has lost the Abhidharma and added a large section of Tantras.  But both canons recognize the divisions known as Prajna-paramita, Ratnakuta, Avatamsaka, and Mahaparinirvana as separate sections.  The Ratnakuta is clearly a collection of sutras equivalent to a small Nikaya.[996] This is probably also true of the voluminous Prajna-paramita in its various editions, but the divisions are not commonly treated as separate works except the Vajracchedika.  The imperfectly known Avatamsaka Sutra appears to be a similar collection, since it is described as discourses of the Buddha pronounced at eight assemblies.  The Mahaparinirvana Sutra though not nominally a collection of sutras (at least in its Pali form) is unique both in subject and structure, and it is easy to understand why it was put in a class by itself.