bout the fifth century in which the
great Hinayana texts are conspicuous: then a further series of
improved translations in which the Hinayana falls into the background
and the works of Asanga and Vasubandhu come to the front. This
evidently reflects the condition of Buddhist India about 500-650 A.D.,
just as the translations of the eighth century reflect its later and
tantric phase.

But can Chinese texts be accepted as reasonably faithful reproductions
of the Indian originals whose names they bear, and some of which have
been lost? This question is really double; firstly, did the
translators reproduce with fair accuracy the Indian text before them,
and secondly, since Indian texts often exist in several recensions,
can we assume that the work which the translators knew under a certain
Sanskrit name is the work known to us by that name? In reply it must
be said that most Chinese translators fall short of our standards of
accuracy. In early times when grammars and dictionaries were unknown
the scholarly rendering of foreign books was a difficult business,
for professional interpreters would usually be incapable of
understanding a philosophic treatise. The method often followed was
that an Indian explained the text to a literary Chinese, who recast
the explanation in his own language. The many translations of the more
important texts and the frequent description of the earlier ones as
imperfect indicate a feeling that the results achieved were not
satisfactory. Several so-called translators, especially Kumarajiva,
gave abstracts of the Indian texts.[761] Others, like Dharmaraksha,
who made a Chinese version of Asvaghosha's Buddhacarita, so
amplified and transposed the original that the result can hardly
be called a translation.[762] Others combined different texts in one.
Thus the work called Ta-o-mi-to-ching[763] consists of extracts taken
from four previous translations of the Sukhavativyuha and rearranged
by the author under the inspiration of Avalokita to whom, as he tells
us, he was wont to pray during the execution of his task. Others
again, like Dharmagupta, anticipated a method afterwards used in
Tibet, and gave a word for word rendering of the Sanskrit which is
hardly intelligible to an educated Chinese. The later versions, _e.g._
those of Hsuan Chuang, are more accurate, but still a Chinese
rendering of a lost Indian document cannot be accepted as a faithful
representation of the original without a critical ex