Often, however, the translator, whatever his weaknesses may have been, had before him a text differing in bulk and arrangement from the Pali and Sanskrit texts which we possess. Thus, there are four Chinese translations of works bearing some relation to the Dhammapada of the Pali Canon. All of these describe the original text as the compilation of Dharmatrâta, to whom is also ascribed the compilation of the Tibetan Udânavarga[765]. His name is not mentioned in connection with the Pali text, yet two of the Chinese translations are closely related to that text. The Fa-chü-ching[766] is a collection of verses translated in 224 A.D. and said to correspond with the Pali except that it has nine additional chapters and some additional stanzas. The Fa-chü-p'i-yü-ching[767] represents another edition of the same verses, illustrated by a collection of parables. It was translated between 290 and 306. The Ch'u-yao-ching[768], translated in 399, is a similar collection of verses and parables, but founded on another Indian work of much greater length. A revised translation containing only the verses was made between 980 and 1001[769]. They are said to be the same as the Tibetan Udâna, and the characteristics of this book, going back apparently to a Sanskrit original, are that it is divided into thirty-three chapters, and that though it contains about 300 verses found in Pali, yet it is not merely the Pali text plus additions, but an anthology arranged on a different principle and only partly identical in substance[770].
There can be little doubt that the Pali Dhammapada is one among several collections of verses, with or without an explanatory commentary of stories. In all these collections there was much common matter, both prose and verse, but some were longer, some shorter, some were in Pali and some in Sanskrit. Whereas the Chinese Dhammapada is longer than the Indian texts, the Chinese version of Milinda's Questions[771] is much shorter and omits books iv-vii. It was made between 317 and 420 A.D. and the inference is that the original Indian text received later additions.
A more important problem is this: what is the relation to the Pali Canon of the Chinese texts bearing titles corresponding to Dîrgha, Madhyama, Samyukta and Ekottara? These collections of sûtras do not call themselves Nikâya but A-han or Agama: the titles are translated as Ch'ang (long), Chung (medium), Tsa (miscellaneous) and Tseng-i, representing Ekottara rather than Anguttara[772]. There is hence prima facie reason to suppose that these works represent not the Pali Canon, but a somewhat similar Sanskrit collection. That one or many Sanskrit works may have coexisted with a somewhat similar Pali work is clearly shown by the Vinaya texts, for here we have the Pali Canon and Chinese translations of five Sanskrit versions, belonging to different schools, but apparently covering the same ground and partly identical. For the Sûtra Pitaka no such body of evidence is forthcoming, but the Sanskrit fragments of the Samyuktâgama found near Turfan contain parts of six sûtras which are arranged in the same order as in the Chinese translation and are apparently the original from which it was made. It is noticeable that three of the four great Agamas were translated by monks who came from Tukhara or Kabul. Guṇabhadra, however, the translator of the Samyuktâgama, came from Central India and the text which he translated was brought from Ceylon by Fa-Hsien. It apparently belonged to the Abhayagiri monastery and not to the Mahâvihâra. Nanjio[773], however, states that about half of it is repeated in the Chinese versions of the Madhyama and Ekottara Agamas. It is also certain that though the Chinese Agamas and Pali Nikâyas contain much common matter, it is differently distributed[774].
There was in India a copious collection of sûtras, existing primarily as oral tradition and varying in diction and arrangement, but codified from time to time in a written form. One of such codifications is represented by the Pali Canon, at least one other by the Sanskrit text which was rendered into Chinese. With rare exceptions the Chinese translations were from the Sanskrit[775]. The Sanskrit codification of the sûtra literature, while differing from the Pali in language and arrangement, is identical in doctrine and almost identical in substance. It is clearly the product of the same or similar schools, but is it earlier or later than the Pali or contemporary with it? The Chinese translations merely fix the latest possible date. A portion of the Samyuktâgama (Nanjio, No. 547) was translated by an unknown author between 220 and 280. This is probably an extract from the complete work which was translated about 440, but it would be difficult to prove that the Indian original was not augmented or rearranged between these dates. The earliest translation of a complete Agama is that of the Ekottarâgama, 384 A.D. But the evidence of inscriptions[776] shows that works known as Nikâyas existed in the third century B.C. The Sanskrit of the Agamas, so far as it is known from the fragments found in Central Asia, does not suggest that they belong to this epoch, but is compatible with the theory that they date from the time of Kanishka of which if we know little, we can at least say that it produced much Buddhist Sanskrit literature. M. Sylvain Lévi has suggested that the later appearance of the complete Vinaya in Chinese is due to the late compilation of the Sanskrit original[777]. It seems to me that other explanations are possible. The early translators were clearly shy of extensive works and until there was a considerable body of Chinese monks, to what public would these theological libraries appeal? Still, if any indication were forthcoming from India or Central Asia that the Sanskrit Agamas were arranged or rearranged in the early centuries of our era, the late date of the Chinese translations would certainly support it. But I am inclined to think that the Nikâyas were rewritten in Sanskrit about the beginning of our era, when it was felt that works claiming a certain position ought to be composed in what had become the general literary language of India[778]. Perhaps those who wrote them in Sanskrit were hardly conscious of making a translation in our sense, but simply wished to publish them in the best literary form.
It seems probable that the Hinayanist portion of the Chinese Tripitaka is in the main a translation of the Canon of the Sarvastivâdins which must have consisted of:
(1) Four Agamas or Nikâyas only, for the Dhammapada is placed outside the Sutta Pitaka.
(2) A voluminous Vinaya covering the same ground as the Pali recension but more copious in legend and anecdote.
(3) An Abhidharma entirely different from the Pali works bearing this name.
It might seem to follow from this that the whole Pali Abhidharma and some important works such as the Thera-Therîgâthâ were unknown to the Hinayanists of Central Asia and Northern India in the early centuries of our era. But caution is necessary in drawing such inferences, for until recently it might have been said that the Sutta Nipâta also was unknown, whereas fragments of it in a Sanskrit version have now been discovered in Eastern Turkestan[779]. The Chinese editors draw a clear distinction between Hinayanist and Mahayanist scriptures. They exclude from the latter works analogous to the Pali Nikâyas and Vinaya, and also the Abhidharma of the Sarvâstivâdins. But the labours of Hsüan Chuang and I-Ching show that this does not imply the rejection of all these works by Mahayanists.
5
Buddhist literary activity has an interesting side aspect, namely the expedients used to transliterate Indian words, which almost provided the Chinese with an alphabet. To some extent Indian names, particularly proper names possessing an obvious meaning, are translated. Thus Asoka becomes Wu-yu, without sorrow: Aśvaghosha, Ma-ming or horse-voice, and Udyâna simply Yüan or park[780]. But many proper names did not lend themselves to such renderings and it was a delicate business to translate theological terms like Nirvâṇa and Samâdhi. The Buddhists did not perhaps invent the idea of using the Chinese characters so as to spell with moderate precision[781], but they had greater need of this procedure than other writers and they used it extensively[782] and with such variety of detail that though they invented some fifteen different syllabaries, none of them obtained general acceptance and Julien[783] enumerates 3000 Chinese characters used to represent the sounds indicated by 47 Indian letters. Still, they gave currency[784] to the system known as fan-ch'ieh which renders a syllable phonetically by two characters, the final of the first and the initial of the second not being pronounced. Thus, in order to indicate the sound Chung, a Chinese dictionary will use the two characters chu yung, which are to be read together as Ch ung.
The transcriptions of Indian words vary in exactitude and the later are naturally better. Hsüan Chuang was a notable reformer and probably after his time Indian words were rendered in Chinese characters as accurately as Chinese words are now transcribed in Latin letters. It is true that modern pronunciation makes such renderings as Fo seem a strange distortion of the original. But it is an abbreviation of Fo-t'o and these syllables were probably once pronounced something like Vut-tha[785]. Similarly Wên-shu-shih-li[786] seems a parody of Manjuśri. But the evidence of modern dialects shows that the first two syllables may have been pronounced as Man-ju. The pupil was probably taught to eliminate the obscure vowel of shih, and li was taken as the nearest equivalent of ri, just as European authors write chih and tzŭ without pretending that they are more than conventional signs for Chinese sounds unknown to our languages. It was certainly possible to transcribe not only names but Sanskrit prayers and formulæ in Chinese characters, and though many writers sneer at the gibberish chanted by Buddhist priests yet I doubt if this ecclesiastical pronunciation, which has changed with that of the spoken language, is further removed from its original than the Latin of Oxford from the speech of Augustus.