The idea that when any important matter is committed to
writing it should be expressed in a literary dialect not too
intelligible to the vulgar is prevalent from Morocco to China. The
language of Bengal illustrates what may have happened to the Buddhist
scriptures. It is said that at the beginning of the nineteenth century
ninety per cent, of the vocabulary of Bengali was Sanskrit, and the
grammatical construction sanskritized as well. Though the literary
language now-a-days is less artificial, it still differs widely from the
vernacular. Similarly the spoken word of the Buddha was forced into
conformity with one literary standard or another and ecclesiastical Pali
became as artificial as Sanskrit. The same incidents may be found worked
up in both languages. Thus the Sanskrit version of the story of Purna in
the Divyava-dana repeats what is found in Pali in the
Samyutta-Nikaya[648] and reappears in Sanskrit in the Vinaya of the
Mulasarvastivadin school.

The Chinese Tripitaka has been catalogued and we possess some
information respecting the books which it contains, though none of them
have been edited in Europe. Thus we know something[649] of the
Sarvastivadin recension of the Abhidhamma. Like the Pali version it
consists of seven books of which one, the Jnana-prasthana by
Katyayaniputra, is regarded as the principal, the rest being
supplementary. All the books are attributed to human authors, and though
some of these bear the names of the Buddha's immediate disciples,
tradition connects Katyayaniputra with Kanishka's council. This is not a
very certain date, but still the inference is that about the time of the
Christian era the contents of the Abhidhamma-Pitaka were not rigidly
defined and a new recension was possible.

The Sanskrit manuscripts discovered in Central Asia include Sutras from
the Samyukta and Ekottara Agamas (equivalent to the Samyutta and
Anguttara Nikayas), a considerable part of the Dharmapada, fragments of
the Sutta-Nipata and the Pratimoksha of the Sarvastivadin school. These
correspond fairly well with the Pali text but represent another
recension and a somewhat different arrangement. We have therefore here
fragments of a Sanskrit version which must have been imported to Central
Asia from northern India and covers, so far as the fragments permit us
to judge, the same ground as the Vinaya and Suttas of the Pali Canon.
Far from displaying the diffuse and inflated style which characterizes
the Ma