The evidence at our disposal points to the fact that Java received most of its civilization from Hindu colonists, but who were these colonists and from what part of India did they come? We must not think of any sudden and definite conquest, but rather of a continuous current of immigration starting perhaps from several springs and often merely trickling, but occasionally swelling into a flood. Native traditions collected by Raffles[382] ascribe the introduction of Brahmanism and the Saka era to the sage Tritresta and represent the invaders as coming from Kalinga or from Gujarat.

The difference of locality may be due to the fact that there was a trade route running from Broach to Masulipatam through Tagara (now Ter). People arriving in the Far East by this route might be described as coming either from Kalinga, where they embarked, or from Gujarat, their country of origin. Dubious as is the authority of these legends, they perhaps preserve the facts in outline. The earliest Javanese inscriptions are written in a variety of the Vengi script and the T'ang annals call the island Kaling as well as Java. It is therefore probable that early tradition represented Kalinga as the home of the Hindu invaders. But later immigrants may have come from other parts. Fa-Hsien could find no Buddhists in Java in 418, but Indian forms of Mahayanism indubitably flourished there in later centuries. The Kalasan inscription dated 778 A.D. and engraved in Nagari characters records the erection of a temple to Tara and of a Mahayanist monastery. The change in both alphabet and religion suggests the arrival of new influences from another district and the Javanese traditions about Gujarat are said to find an echo among the bards of western India and in such proverbs as, they who go to Java come not back.[383] In the period of the Hunnish and Arab invasions there may have been many motives for emigration from Gujarat. The land route to Kalinga was probably open and the sea route offers no great difficulties.[384]

Another indication of connection with north-western India is found in the Chinese work Kao Seng Chuan (519 A.D.) or Biographies of Eminent Monks, if the country there called She-p'o can be identified with Java.[385] It is related that Gunavarman, son of the king of Kashmir, became a monk and, declining the throne, went first to Ceylon and then to the kingdom of She-p'o, which he converted to Buddhism. He died at Nanking in 431 B.C.

Taranatha[386] states that Indo-China which he calls the Koki country,[387] was first evangelized in the time of Asoka and that Mahayanism was introduced there by the disciples of Vasubandhu, who probably died about 360 A.D., so that the activity of his followers would take place in the fifth century. He also says that many clergy from the Koki country were in Madhyadesa from the time of Dharmapala (about 800 A.D.) onwards, and these two statements, if they can be accepted, certainly explain the character of Javanese and Cambojan Buddhism. Taranatha is a confused and untrustworthy writer, but his statement about the disciples of Vasubandhu is confirmed by the fact that Dignaga, who was one of them, is the only authority cited in the Kamahayanikan.[388]

The fact that the terms connected with rice cultivation are Javanese and not loan-words indicates that the island had some indigenous civilization when the Hindus first settled there. Doubtless they often came with military strength, but on the whole as colonists and teachers rather than as conquerors. The Javanese kings of whom we know most appear to have been not members of Hindu dynasties but native princes who had adopted Hindu culture and religion. Sanskrit did not oust Javanese as the language of epigraphy, poetry and even religious literature. Javanese Buddhism appears to have preserved its powers of growth and to have developed some special doctrines. But Indian influence penetrated almost all institutions and is visible even to-day. Its existence is still testified to by the alphabet in use, by such titles as Arjo, Radja, Praboe, Dipati (=adhipati), and by various superstitions about lucky days and horoscopes. Communal land tenure of the Indian kind still exists and in former times grants of land were given to priests and, as in India, recorded on copper plates. Offerings to old statues are still made and the Tenggerese[389] are not even nominal Mohammedans. The Balinese still profess a species of Hinduism and employ a Hindu Calendar.

From the tenth century onwards the history of Java becomes a little plainer.

Copper plates dating from about 900 A.D. mention Mataram. A certain Mpoe Sindok was vizier of this kingdom in 919, but ten years later we find him an independent king in east Java. He lived at least twenty-five years longer and his possessions included Pasoeroean, Soerabaja and Kediri. His great-grandson, Er-langga (or Langghya), is an important figure. Er-langga's early life was involved in war, but in 1032 he was able to call himself, though perhaps not with great correctness, king of all Java. His memory has not endured among the Javanese but is still honoured in the traditions of Bali and Javanese literature began in his reign or a little earlier. The poem Arjuna-vivaha is dedicated to him, and one book of the old Javanese prose translation of the Mahabharata bears a date equivalent to 996 A.D.[390]

One of the national heroes of Java is Djajabaja[391] who is supposed to have lived in the ninth century. But tradition must be wrong here, for the free poetic rendering of part of the Mahabharata called Bharata-Yuddha, composed by Mpoe Sedah in 1157 A.D., is dedicated to him, and his reign must therefore be placed later than the traditional date. He is said to have founded the kingdom of Daha in Kediri, but his inscriptions merely indicate that he was a worshipper of Vishnu. Literature and art flourished in east Java at this period for it would seem that the Kawi Ramayana and an ars poetica called Vritta-sancaya[392] were written about 1150 and that the temple of Panataran was built between 1150 and 1175.

In western Java we have an inscription of 1030 found on the river Tjitjatih. It mentions a prince who is styled Lord of the World and native tradition, confirmed by inscriptions, which however give few details, relates that in the twelfth century a kingdom called Padjadjaran was founded in the Soenda country south of Batavia by princes from Toemapel in eastern Java.

There is a gap in Javanese history from the reign of Djajabaja till 1222 at which date the Pararaton,[393] or Book of the Kings of Toemapel and Madjapahit, begins to furnish information. The Sung annals[394] also give some account of the island but it is not clear to what years their description refers. They imply, however, that there was an organized government and that commerce was flourishing. They also state that the inhabitants "pray to the gods and Buddha": that Java was at war with eastern Sumatra: that embassies were sent to China in 992 and 1109 and that in 1129 the Emperor gave the ruler of Java (probably Djajabaja) the title of king.

The Pararaton opens with the fall of Daha in 1222 which made Toemapel, known later as Singasari, the principal kingdom. Five of its kings are enumerated, of whom Vishnuvardhana was buried in the celebrated shrine of Tjandi Djago, where he was represented in the guise of Buddha. His successor Sri Rajasanagara was praised by the poet Prapantja[395] as a zealous Buddhist but was known by the posthumous name of Sivabuddha. He was the first to use the name of Singasari and perhaps founded a new city, but the kingdom of Toemapel came to an end in his reign for he was slain by Djaja Katong,[396] prince of Daha, who restored to that kingdom its previous primacy, but only for a short time, since it was soon supplanted by Madjapahit. The foundation of this state is connected with a Chinese invasion of Java, related at some length in the Yuan annals,[397] so that we are fortunate in possessing a double and fairly consistent account of what occurred.

We learn from these sources that some time after Khubilai Khan had conquered China, he sent missions to neighbouring countries to demand tribute. The Javanese had generally accorded a satisfactory reception to Chinese missions, but on this occasion the king (apparently Djaja Katong) maltreated the envoy and sent him back with his face cut or tattooed. Khubilai could not brook this outrage and in 1292 despatched a punitive expedition. At that time Raden Vidjaja, the son-in-law of Kertanagara, had not submitted to Djaja Katong and held out at Madjapahit, a stronghold which he had founded near the river Brantas. He offered his services to the Chinese and after a two months' campaign Daha was captured and Djaja Katong killed. Raden Vidjaja now found that he no longer needed his Chinese allies. He treacherously massacred some and prepared to fight the rest. But the Mongol generals, seeing the difficulties of campaigning in an unknown country without guides, prudently returned to their master and reported that they had taken Daha and killed the insolent king.

Madjapahit (or Wilwatikta) now became the premier state of Java, and had some permanency. Eleven sovereigns, including three queens, are enumerated by the Pararaton until its collapse in 1468. We learn from the Ming annals and other Chinese documents[398] that it had considerable commercial relations with China and sent frequent missions: also that Palembang was a vassal of Java. But the general impression left by the Pararaton is that during the greater part of its existence Madjapahit was a distracted and troubled kingdom. In 1403, as we know from both Chinese and Javanese sources, there began a great war between the western and eastern kingdoms, that is between Madjapahit and Balambangan in the extreme east, and in the fifteenth century there was twice an interregnum. Art and literature, though not dead, declined and events were clearly tending towards a break-up or revolution. This appears to have been consummated in 1468, when the Pararaton simply says that King Pandansalas III left the Kraton, or royal residence.

It is curious that the native traditions as to the date and circumstances in which Madjapahit fell should be so vague, but perhaps the end of Hindu rule in Java was less sudden and dramatic than we are inclined to think. Islam had been making gradual progress and its last opponents were kings only in title. The Chinese mention the presence of Arabs in the seventh century, and the geography called Ying-yai Sheng-lan (published in 1416), which mentions Grisse, Soerabaja and Madjapahit as the principal towns of Java, divides the inhabitants into three classes: (a) Mohammedans who have come from the west, "their dress and food is clean and proper"; (b) the Chinese, who are also cleanly and many of whom are Mohammedans; (c) the natives who are ugly and uncouth, devil-worshippers, filthy in food and habits. As the Chinese do not generally speak so severely of the hinduized Javanese it would appear that Hinduism lasted longest among the lower and more savage classes, and that the Moslims stood on a higher level. As in other countries, the Arabs attempted to spread Islam from the time of their first appearance. At first they confined their propaganda to their native wives and dependents. Later we hear of veritable apostles of Islam such as Malik Ibrahim, and Raden Rahmat, the ruler of a town called Ampel[399] which became the head quarter of Islam. The princes whose territory lay round Madjapahit were gradually converted and the extinction of the last Hindu kingdom became inevitable.[400]

3

It is remarkable that the great island of Sumatra, which seems to lie in the way of anyone proceeding from India eastwards and is close to the Malay peninsula, should in all ages have proved less accessible to invaders coming from the west than the more distant Java. Neither Hindus, Arabs nor Europeans have been able to establish their influence there in the same thorough manner. The cause is probably to be found in its unhealthy and impenetrable jungles, but even so its relative isolation remains singular.

It does not appear that any prince ever claimed to be king of all Sumatra. For the Hindu period we have no indigenous literature and our scanty knowledge is derived from a few statues and inscriptions and from notices in Chinese writings. The latter do not refer to the island as a whole but to several states such as Indragiri near the Equator and Kandali (afterwards called San-bo-tsai, the Sabaza of the Arabs) near Palembang. The annals of the Liang dynasty say that the customs of Kandali were much the same as those of Camboja and apparently we are to understand that the country was Buddhist, for one king visited the Emperor Wu-ti in a dream, and his son addressed a letter to His Majesty eulogizing his devotion to Buddhism. Kandali is said to have sent three envoys to China between 454 and 519.

The Chinese pilgrim I-Ching[401] visited Sumatra twice, once for two months in 672 and subsequently for some years (about 688-695). He tells us that in the islands of the Southern Sea, "which are more than ten countries," Buddhism flourishes, the school almost universally followed being the Mulasarvastivada, though the Sammitiyas and other schools have a few adherents. He calls the country where he sojourned and to which these statements primarily refer, Bhoja or Sribhoja (Fo-shih or Shih-li-fo-shih), adding that its former name was Malayu. It is conjectured that Shih-li-fo-shih is the place later known as San-bo-tsai[402] and Chinese authors seem to consider that both this place and the earlier Kandali were roughly speaking identical with Palembang. I-Ching tells us that the king of Bhoja favoured Buddhism and that there were more than a thousand priests in the city. Gold was abundant and golden flowers were offered to the Buddha. There was communication by ship with both India and China. The Hinayana, he says, was the form of Buddhism adopted "except in Malayu, where there are a few who belong to the Mahayana." This is a surprising statement, but it is impossible to suppose that an expert like I-Ching can have been wrong about what he actually saw in Sribhoja. So far as his remarks apply to Java they must be based on hearsay and have less authority, but the sculptures of Boroboedoer appear to show the influence of Mulasarvastivadin literature. It must be remembered that this school, though nominally belonging to the Hinayana, came to be something very different from the Theravada of Ceylon.

The Sung annals and subsequent Chinese writers know the same district (the modern Palembang) as San-bo-tsai (which may indicate either mere change of name or the rise of a new city) and say that it sent twenty-one envoys between 960 and 1178. The real object of these missions was to foster trade and there was evidently frequent intercourse between eastern Sumatra, Champa and China. Ultimately the Chinese seem to have thought that the entertainment of Sumatran diplomatists cost more than they were worth, for in 1178 the emperor ordered that they should not come to Court but present themselves in the province of Fu-kien. The Annals state that Sanskrit writing was in use at San-bo-tsai and lead us to suppose that the country was Buddhist. They mention several kings whose names or titles seem to begin with the Sanskrit word Sri.[403] In 1003 the envoys reported that a Buddhist temple had been erected in honour of the emperor and they received a present of bells for it. Another envoy asked for dresses to be worn by Buddhist monks. The Ming annals also record missions from San-bo-tsai up to 1376, shortly after which the region was conquered by Java and the town decayed.[404] In the fourteenth century Chinese writers begin to speak of Su-men-ta-la or Sumatra by which is meant not the whole island but a state in the northern part of it called Samudra and corresponding to Atjeh.[405] It had relations with China and the manners and customs of its inhabitants are said to be the same as in Malacca, which probably means that they were Moslims.

Little light is thrown on the history of Sumatra by indigenous or Javanese monuments. Those found testify, as might be expected, to the existence here and there of both Brahmanism and Buddhism. In 1343 a Sumatran prince named Adityavarman, who was apparently a vassal of Madjapahit, erected an image of Manjusri at Tjandi Djago and in 1375 one of Amoghapasa.

4

The Liang and T'ang annals both speak of a country called Po-li, described as an island lying to the south-east of Canton. Groeneveldt identified it with Sumatra, but the account of its position suggests that it is rather to be found in Borneo, parts of which were undoubtedly known to the Chinese as Po-lo and Pu-ni.[406] The Liang annals state that Po-li sent an embassy to the Emperor Wu-ti in 518 bearing a letter which described the country as devoted to Buddhism and frequented by students of the three vehicles. If the letter is an authentic document the statements in it may still be exaggerations, for the piety of Wu-ti was well known and it is clear that foreign princes who addressed him thought it prudent to represent themselves and their subjects as fervent Buddhists. But there certainly was a Hindu period in Borneo, of which some tradition remains among the natives,[407] although it ended earlier and left fewer permanent traces than in Java and elsewhere.

The most important records of this period are three Sanskrit inscriptions found at Koetei on the east coast of Borneo.[408] They record the donations made to Brahmans by King Mulavarman, son of Asvavarman and grandson of Kundagga. They are not dated, but Kern considers for palaeographical reasons that they are not later than the fifth century. Thus, since three generations are mentioned, it is probable that about 400 A.D. there were Hindu princes in Borneo. The inscriptions testify to the existence of Hinduism there rather than of Buddhism: in fact the statements in the Chinese annals are the only evidence for the latter. But it is most interesting to find that these annals give the family name of the king of Poli as Kaundinya[409] which no doubt corresponds to the Kundagga of the Koetei inscription. At least one if not two of the Hindu invaders of Camboja bore this name, and we can hardly be wrong in supposing that members of the same great family became princes in different parts of the Far East. One explanation of their presence in Borneo would be that they went thither from Camboja, but we have no record of expeditions from Camboja and if adventurers started thence it is not clear why they went to the east coast of Borneo. It would be less strange if Kaundinyas emigrating from Java reached both Camboja and Koetei. It is noticeable that in Java, Koetei, Champa and Camboja alike royal names end in varman.

5

The architectural monuments of Java are remarkable for their size, their number and their beauty. Geographically they fall into two chief groups, the central (Boroboedoer, Prambanan, Dieng plateau, etc.) in or near the kingdom of Mataram and the eastern (Tjandi Djago, Singasari, Panataran, etc.) lying not at the extremity of the island but chiefly to the south of Soerabaja. No relic of antiquity deserving to be called a monument has been found in western Java for the records left by Purnavarman (c. 400 A.D.) are merely rocks bearing inscriptions and two footprints, as a sign that the monarch's triumphal progress is compared to the three steps of Vishnu.

The earliest dated (779 A.D.) monument in mid Java, Tjandi Kalasan, is Buddhist and lies in the plain of Prambanan. It is dedicated to Tara and is of a type common both in Java and Champa, namely a chapel surmounted by a tower. In connection with it was erected the neighbouring building called Tjandi Sari, a two-storied monastery for Mahayanist monks. Not far distant is Tjandi Sevu, which superficially resembles the 450 Pagodas of Mandalay, for it consists of a central cruciform shrine surrounded by about 240 smaller separate chapels, everyone of which, apparently, contained the statue of a Dhyani Buddha. Other Buddhist buildings in the same region are Tjandi Plaosan, and the beautiful chapel known as Tjandi Mendut in which are gigantic seated images of the Buddha, Manjusri and Avalokita. The face of the last named is perhaps the most exquisite piece of work ever wrought by the chisel of a Buddhist artist.

It is not far from Mendut to Boroboedoer, which deserves to be included in any list of the wonders of the world. This celebrated stupa—for in essence it is a highly ornamented stupa with galleries of sculpture rising one above the other on its sides—has been often described and can be described intelligibly only at considerable length. I will therefore not attempt to detail or criticize its beauties but will merely state some points which are important for our purpose.

It is generally agreed that it must have been built about 850 A.D., but obviously the construction lasted a considerable time and there are indications that the architects altered their original plan. The unknown founder must have been a powerful and prosperous king for no one else could have commanded the necessary labour. The stupa shows no sign of Brahmanic influence. It is purely Buddhist and built for purposes of edification. The worshippers performed pradakshina by walking round the galleries, one after the other, and as they did so had an opportunity of inspecting some 2000 reliefs depicting the previous births of Sakyamuni, his life on earth and finally the mysteries of Mahayanist theology. As in Indian pilgrim cities, temple guides were probably ready to explain the pictures.

The selection of reliefs is not due to the artists' fancy but aims at illustrating certain works. Thus the scenes of the Buddha's life reproduce in stone the story of the Lalita Vistara[410] and the Jataka pictures are based on the Divyavadana. It is interesting to find that both these works are connected with the school of the Mulasarvastivadins, which according to I-Ching was the form of Buddhism prevalent in the archipelago. In the third gallery the figure of Maitreya is prominent and often seems to be explaining something to a personage who accompanies him. As Maitreya is said to have revealed five important scriptures to Asanga, and as there is a tradition that the east of Asia was evangelized by the disciples of Asanga or Vasubandhu, it is possible that the delivery and progress of Maitreya's revelation is here depicted. The fourth gallery seems to deal with the five superhuman Buddhas,[411] their paradises and other supra-mundane matters, but the key to this series of sculptures has not yet been found. It is probable that the highest storey proved to be too heavy in its original form and that the central dagoba had to be reduced lest it should break the substructure. But it is not known what image or relic was preserved in this dagoba. Possibly it was dedicated to Vairocana who was regarded as the Supreme Being and All-God by some Javanese Buddhists.[412]

The creed here depicted in stone seems to be a form of Mahayanism. Sakyamuni is abundantly honoured but there is no representation of his death. This may be because the Lalita Vistara treats only of his early career, but still the omission is noteworthy. In spite of the importance of Sakyamuni, a considerable if mysterious part is played by the five superhuman Buddhas, and several Bodhisattvas, especially Maitreya, Avalokita and Manjusri. In the celestial scenes we find numerous Bodhisattvas both male and female, yet the figures are hardly Tantric and there is no sign that any of the personages are Brahmanic deities.

Yet the region was not wholly Buddhist. Not far from Boroboedoer and apparently of about the same age is the Sivaite temple of Banon, and the great temple group of Prambanam is close to Kalasan and to the other Buddhist shrines mentioned above. It consists of eight temples of which four are dedicated to Brahma, Siva, Vishnu and Nandi respectively, the purpose of the others being uncertain. The largest and most decorated is that dedicated to Siva, containing four shrines in which are images of the god as Mahadeva and as Guru, of Ganesa and of Durga. The balustrade is ornamented with a series of reliefs illustrating the Ramayana. These temples, which appear to be entirely Brahmanic, approach in style the architecture of eastern Java and probably date from the tenth century, that is about a century later than the Buddhist monuments. But there is no tradition or other evidence of a religious revolution.

The temples on the Dieng plateau are also purely Brahmanic and probably older, for though we have no record of their foundation, an inscribed stone dated 800 A.D. has been found in this district. The plateau which is 6500 feet high was approached by paved roads or flights of stairs on one of which about 4000 steps still remain. Originally there seem to have been about 40 buildings on the plateau but of these only eight now exist besides several stone foundations which supported wooden structures. The place may have been a temple city analogous to Girnar or Satrunjaya, but it appears to have been deserted in the thirteenth century, perhaps in consequence of volcanic activity. The Dieng temples are named after the heroes of the Mahabharata (Tjandi Ardjuno, Tjandi Bimo, etc.), but these appear to be late designations. They are rectangular towerlike shrines with porches and a single cellule within. Figures of Brahma, Siva and Vishnu have been discovered, as well as spouts to carry off the libation water.

Before leaving mid Java I should perhaps mention the relatively modern (1435-1440 A.D.) temples of Suku. I have not seen these buildings, but they are said to be coarse in execution and to indicate that they were used by a debased sect of Vishnuites. Their interest lies in the extraordinary resemblance which they bear to the temples of Mexico and Yucatan, a resemblance "which no one can fail to observe, though no one has yet suggested any hypothesis to account for it."[413]

The best known and probably the most important monuments of eastern Java are Panataran, Tjandi Djago and Tjandi Singasari.[414]

The first is considered to date from about 1150 A.D. It is practically a three-storied pyramid with a flat top. The sides of the lowest storey are ornamented with a series of reliefs illustrating portions of the Ramayana, local legends and perhaps the exploits of Krishna, but this last point is doubtful.[415] This temple seems to indicate the same stage of belief as Prambanam. It shows no trace of Buddhism and though Siva was probably the principal deity, the scenes represented in its sculptures are chiefly Vishnuite.

Tjandi Djago is in the province of Pasoeroean. According to the Pararaton and the Nagarakretagama,[416] Vishnuvardhana, king of Toemapel, was buried there. As he died in 1272 or 1273 A.D. and the temple was already in existence, we may infer that it dates from at least 1250. He was represented there in the form of Sugata (that is the Buddha) and at Waleri in the form of Siva. Here we have the custom known also in Champa and Camboja of a deceased king being represented by a statue with his own features but the attributes of his tutelary deity. It is strange that a king named after Vishnu should be portrayed in the guise of Siva and Buddha. But in spite of this impartiality, the cult practised at Tjandi Djago seems to have been not a mixture but Buddhism of a late Mahayanist type. It was doubtless held that Buddhas and Bodhisattvas are identical with Brahmanic deities, but the fairly numerous pantheon discovered in or near the ruins consists of superhuman Buddhas and Bodhisattvas with their spouses.[417]

In form Tjandi Djago has somewhat the appearance of a three-storied pyramid but the steps leading up to the top platform are at one end only and the shrine instead of standing in the centre of the platform is at the end opposite to the stairs. The figures in the reliefs are curiously square and clumsy and recall those of Central America.

Tjandi Singasari, also in the province of Pasoeroean, is of a different form. It is erected on a single low platform and consists of a plain rectangular building surmounted by five towers such as are also found in Cambojan temples. There is every reason to believe that it was erected in 1278 A.D. in the reign of Kretanagara, the last king of Toemapel, and that it is the temple known as Siva-buddhalaya in which he was commemorated under the name of Siva-buddha. An inscription found close by relates that in 1351 A.D. a shrine was erected on behalf of the royal family in memory of those who died with the king.[418]

The Nagarakretagama represents this king as a devout Buddhist but his very title Sivabuddha shows how completely Sivaism and Buddhism were fused in his religion. The same work mentions a temple in which the lower storey was dedicated to Siva and the upper to Akshobhya: it also leads us to suppose that the king was honoured as an incarnation of Akshobhya even during his life and was consecrated as a Jina under the name of Srijnanabajresvara.[419] The Singasari temple is less ornamented with reliefs than the others described but has furnished numerous statues of excellent workmanship which illustrate the fusion of the Buddhist and Sivaite pantheons. On the one side we have Prajnaparamita, Manjusri and Tara, on the other Ganesa, the Linga, Siva in various forms (Guru, Nandisvara, Mahakala, etc.), Durga and Brahma. Not only is the Sivaite element predominant but the Buddhist figures are concerned less with the veneration of the Buddha than with accessory mythology.

Javanese architecture and sculpture are no doubt derived from India, but the imported style, whatever it may have been, was modified by local influences and it seems impossible at present to determine whether its origin should be sought on the eastern or western side of India. The theory that the temples on the Dieng plateau are Chalukyan buildings appears to be abandoned but they and many others in Java show a striking resemblance to the shrines found in Champa. Javanese architecture is remarkable for the complete absence not only of radiating arches but of pillars, and consequently of large halls. This feature is no doubt due to the ever present danger of earthquakes. Many reliefs, particularly those of Panataran, show the influence of a style which is not Indian and may be termed, though not very correctly, Polynesian. The great merit of Javanese sculpture lies in the refinement and beauty of the faces. Among figures executed in India it would be hard to find anything equal in purity and delicacy to the Avalokita of Mendut, the Manjusri now in the Berlin Museum or the Prajnaparamita now at Leyden.

6

From the eleventh century until the end of the Hindu period Java can show a considerable body of literature, which is in part theological. It is unfortunate that no books dating from an earlier epoch should be extant. The sculptures of Prambanam and Boroboedoer clearly presuppose an acquaintance with the Ramayana, the Lalita Vistara and other Buddhist works but, as in Camboja, this literature was probably known only in the original Sanskrit and only to the learned. But it is not unlikely that the Javanese adaptations of the Indian epics which have come down to us were preceded by earlier attempts which have disappeared.

The old literary language of Java is commonly known as Basa Kawi or Kawi, that is the language of poetry.[420] It is however simply the predecessor of modern Javanese and many authorities prefer to describe the language of the island as Old Javanese before the Madjapahit period, Middle-Javanese during that period and New Javanese after the fall of Madjapahit. The greater part of this literature consists of free versions of Sanskrit works or of a substratum in Sanskrit accompanied by a Javanese explanation. Only a few Javanese works are original, that is to say not obviously inspired by an Indian prototype, but on the other hand nearly all of them handle their materials with freedom and adapt rather than translate what they borrow.

One of the earliest works preserved appears to be the Tantoe Panggelaran, a treatise on cosmology in which Indian and native ideas are combined. It is supposed to have been written about 1000 A.D. Before the foundation of Madjapahit Javanese literature flourished especially in the reigns of Erlangga and Djajabaja, that is in the eleventh and twelfth centuries respectively. About the time of Erlangga were produced the old prose version of the Mahabharata, in which certain episodes of that poem are rendered with great freedom and the poem called Arjuna-vivaha, or the marriage of Arjuna.

The Bharatayuddha,[421] which states that it was composed by Mpoe Sedah in 1157 by order of Djajabaja, prince of Kediri, is, even more than the prose version mentioned above, a free rendering of parts of the Mahabharata. It is perhaps based on an older translation preserved in Bali.[422] The Kawi Ramayana was in the opinion of Kern composed about 1200 A.D. It follows in essentials the story of the Ramayana, but it was apparently composed by a poet unacquainted with Sanskrit who drew his knowledge from some native source now unknown.[423] He appears to have been a Sivaite. To the eleventh century are also referred the Smaradahana and the treatise on prosody called Vrittasancaya. All this literature is based upon classical Sanskrit models and is not distinctly Buddhist although the prose version of the Mahabharata states that it was written for Brahmans, Sivaites and Buddhists.[424] Many other translations or adaptations of Sanskrit work are mentioned, such as the Nitisastra, the Sarasamuccaya, the Tantri (in several editions), a prose translation of the Brahmandapurana, together with grammars and dictionaries. The absence of dates makes it difficult to use these works for the history of Javanese thought. But it seems clear that during the Madjapahit epoch, or perhaps even before it, a strong current of Buddhism permeated Javanese literature, somewhat in contrast with the tone of the works hitherto cited. Brandes states that the Sutasoma, Vighnotsava, Kunjarakarna, Sang Hyang Kamahayanikan, and Buddhapamutus are purely Buddhist works and that the Tjantakaparva, Arjunavijaya, Nagarakretagama, Wariga and Bubukshah show striking traces of Buddhism.[425] Some of these works are inaccessible to me but two of them deserve examination, the Sang Hyang Kamahayanikan[426] and the story of Kunjarakarna.[427] The first is tentatively assigned to the Madjapahit epoch or earlier, the second with the same caution to the eleventh century. I do not presume to criticize these dates which depend partly on linguistic considerations. The Kamahayanikan is a treatise (or perhaps extracts from treatises) on Mahayanism as understood in Java and presumably on the normal form of Mahayanism. The other work is an edifying legend including an exposition of the faith by no one less than the Buddha Vairocana. In essentials it agrees with the Kamahayanikan but in details it shows either sectarian influence or the idiosyncrasies of the author.

The Kamahayanikan consists of Sanskrit verses explained by a commentary in old Javanese and is partly in the form of questions and answers. The only authority whom it cites is Dignaga. It professes to teach the Mahayana and Mantrayana, which is apparently a misspelling for Mantrayana. The emphasis laid on Bajra (that is vajra or dorje), ghanta, mudra, mandala, mystic syllables, and Devis marks it as an offshoot of Tantrism and it offers many parallels to Nepalese literature. On the other hand it is curious that it uses the form Nibana not Nirvana.[428] Its object is to teach a neophyte, who has to receive initiation, how to become a Buddha.[429] In the second part the pupil is addressed as Jinaputra, that is son of the Buddha or one of the household of faith. He is to be moderate but not ascetic in food and clothing: he is not to cleave to the Puranas and Tantras but to practise the Paramitas. These are defined first as six[430] and then four others are added.[431] Under Prajnaparamita is given a somewhat obscure account of the doctrine of Sunyata. Then follows the exposition of Paramaguhya (the highest secret) and Mahaguhya (the great secret). The latter is defined as being Yoga, the bhavanas, the four noble truths and the ten paramitas. The former explains the embodiment of Bhatara Visesha, that is to say the way in which Buddhas, gods and the world of phenomena are evolved from a primordial principle, called Advaya and apparently equivalent to the Nepalese Adibuddha.[432] Advaya is the father of Buddha and Advayajnana, also called Bharali Prajnaparamita, is his mother, but the Buddha principle at this stage is also called Divarupa. In the next stage this Divarupa takes form as Sakyamuni, who is regarded as a superhuman form of Buddhahood rather than as a human teacher, for he produces from his right and left side respectively Lokesvara and Bajrapani. These beings produce, the first Akshobhya and Ratnasambhava, the second Amitabha and Amoghasiddhi, but Vairocana springs directly from the face of Sakyamuni. The five superhuman Buddhas are thus accounted for. From Vairocana spring Isvara (Siva), Brahma, and Vishnu: from them the elements, the human body and the whole world. A considerable part of the treatise is occupied with connecting these various emanations of the Advaya with mystic syllables and in showing how the five Buddhas correspond to the different skandas, elements, senses, etc. Finally we are told that there are five Devis, or female counterparts corresponding in the same order to the Buddhas named above and called Locana, Mamaki, Pandaravasini, Tara and Dhatvisvari. But it is declared that the first and last of these are the same and therefore there are really only four Devis.

The legend of Kunjarakarna relates how a devout Yaksha of that name went to Bodhicitta[433] and asked of Vairocana instruction in the holy law and more especially as to the mysteries of rebirth. Vairocana did not refuse but bade his would-be pupil first visit the realms of Yama, god of the dead. Kunjarakarna did so, saw the punishments of the underworld, including the torments prepared for a friend of his, whom he was able to warn on his return. Yama gave him some explanations respecting the alternation of life and death and he was subsequently privileged to receive a brief but more general exposition of doctrine from Vairocana himself.

This doctrine is essentially a variety of Indian pantheism but peculiar in its terminology inasmuch as Vairocana, like Krishna in the Bhagavad-gita, proclaims himself to be the All-God and not merely the chief of the five Buddhas. He quotes with approval the saying "you are I: I am you" and affirms the identity of Buddhism and Sivaism. Among the monks[434] there are no muktas (i.e. none who have attained liberation) because they all consider as two what is really one. "The Buddhists say, we are Bauddhas, for the Lord Buddha is our highest deity: we are not the same as the Sivaites, for the Lord Siva is for them the highest deity." The Sivaites are represented as saying that the five Kusikas are a development or incarnations of the five Buddhas. "Well, my son" is the conclusion, "These are all one: we are Siva, we are Buddha."

In this curious exposition the author seems to imply that his doctrine is different from that of ordinary Buddhists, and to reprimand them more decidedly than Sivaites. He several times uses the phrase Namo Bhatara, namah Sivaya (Hail, Lord: hail to Siva) yet he can hardly be said to favour the Sivaites on the whole, for his All-God is Vairocana who once (but only once) receives the title of Buddha. The doctrine attributed to the Sivaites that the five Kusikas are identical with the superhuman Buddhas remains obscure.[435] These five personages are said to be often mentioned in old Javanese literature but to be variously enumerated.[436] They are identified with the five Indras, but these again are said to be the five senses (indriyas). Hence we can find a parallel to this doctrine in the teaching of the Kamahayanikan that the five Buddhas correspond to the five senses.

Two other special theses are enounced in the story of Kunjarakarna. The first is Vairocana's analysis of a human being, which makes it consist of five Atmans or souls, called respectively Atman, Cetanatman, Paratman, Niratman and Antaratman, which somehow correspond to the five elements, five senses and five Skandhas. The singular list suggests that the author was imperfectly acquainted with the meaning of the Sanskrit words employed and the whole terminology is strange in a Buddhist writer. Still in the later Upanishads[437] the epithet pancatmaka is applied to the human body, especially in the Garbha Upanishad which, like the passage here under consideration, gives a psychophysiological explanation of the development of an embryo into a human being.

The second thesis is put in the mouth of Yama. He states that when a being has finished his term in purgatory he returns to life in this world first as a worm or insect, then successively as a higher animal and a human being, first diseased or maimed and finally perfect. No parallel has yet been quoted to this account of metempsychosis.

Thus the Kunjarakarna contains peculiar views which are probably sectarian or individual. On the other hand their apparent singularity may be due to our small knowledge of old Javanese literature. Though other writings are not known to extol Vairocana as being Siva and Buddha in one, yet they have no scruple in identifying Buddhist and Brahmanic deities or connecting them by some system of emanations, as we have already seen in the Kamahayanikan. Such an identity is still more definitely proclaimed in the old Javanese version of the Sutasoma Jataka.[438] It is called Purushada-Santa and was composed by Tantular who lived at Madjapahit in the reign of Rajasanagara (1350-1389 A.D.). In the Indian original Sutasoma is one of the previous births of Gotama. But the Javanese writer describes him as an Avatara of the Buddha who is Brahma, Vishnu and Isvara, and he states that "The Lord Buddha is not different from Siva the king of the gods.... They are distinct and they are one. In the Law is no dualism." The superhuman Buddhas are identified with various Hindu gods and also with the five senses. Thus Amitabha is Mahadeva and Amoghasiddhi is Vishnu. This is only a slight variation of the teaching in the Kamahayanikan. There Brahmanic deities emanate from Sakyamuni through various Bodhisattvas and Buddhas: here the Buddha spirit is regarded as equivalent to the Hindu Trimurti and the various aspects of this spirit can be described in either Brahmanic or Buddhistic terminology though in reality all Buddhas, Bodhisattvas and gods are one. But like the other authors quoted, Tantular appears to lean to the Buddhist side of these equations, especially for didactic purposes. For instance he says that meditation should be guided "by Lokesvara's word and Sakyamuni's spirit."

7

Thus it will be seen that if we take Javanese epigraphy, monuments and literature together with Chinese notices, they to some extent confirm one another and enable us to form an outline picture, though with many gaps, of the history of thought and religion in the island. Fa-Hsien tells us that in 418 A.D. Brahmanism flourished (as is testified by the inscriptions of Purnavarman) but that the Buddhists were not worth mentioning. Immediately afterwards, probably in 423, Gunavarman is said to have converted She-po, if that be Java, to Buddhism, and as he came from Kashmir he was probably a Sarvastivadin. Other monks are mentioned as having visited the southern seas.[439] About 690 I-Ching says that Buddhism of the Mulasarvastivadin school was flourishing in Sumatra, which he visited, and in the other islands of the Archipelago. The remarkable series of Buddhist monuments in mid Java extending from about 779 to 900 A.D. confirms his statement. But two questions arise. Firstly, is there any explanation of this sudden efflorescence of Buddhism in the Archipelago, and next, what was its doctrinal character? If, as Taranatha says, the disciples of Vasubandhu evangelized the countries of the East, their influence might well have been productive about the time of I-Ching's visit. But in any case during the sixth and seventh centuries religious travellers must have been continually journeying between India and China, in both directions, and some of them must have landed in the Archipelago. At the beginning of the sixth century Buddhism was not yet decadent in India and was all the fashion in China. It is not therefore surprising if it was planted in the islands lying on the route. It may be, as indicated above, that some specially powerful body of Hindus coming from the region of Gujarat and professing Buddhism founded in Java a new state.

As to the character of this early Javanese Buddhism we have the testimony of I-Ching that it was of the Mulasarvastivadin school and Hinayanist. He wrote of what he had seen in Sumatra but of what he knew only by hearsay in Java and his statement offers some difficulties. Probably Hinayanism was introduced by Gunavarman but was superseded by other teachings which were imported from time to time after they had won for themselves a position in India. For the temple of Kalasan (A.D. 779) is dedicated to Tara and the inscription found there speaks of the Mahayana with veneration. The later Buddhism of Java has literary records which, so far as I know, are unreservedly Mahayanist but probably the sculptures of Boroboedoer are the most definite expression which we shall ever have of its earlier phases. Since they contain images of the five superhuman Buddhas and of numerous Bodhisattvas, they can hardly be called anything but Mahayanist. But on the other hand the personality of Sakyamuni is emphasized; his life and previous births are pictured in a long series of sculptures and Maitreya is duly honoured. Similar collections of pictures and images may be seen in Burma which differ doctrinally from those in Java chiefly by substituting the four human Buddhas[440] and Maitreya for the superhuman Buddhas. But Mahayanist teaching declares that these human Buddhas are reflexes of counterparts of the superhuman Buddhas so that the difference is not great.

Mahayanist Buddhism in Camboja and at a later period in Java itself was inextricably combined with Hinduism, Buddha being either directly identified with Siva or regarded as the primordial spirit from which Siva and all gods spring. But the sculptures of Boroboedoer do not indicate that the artists knew of any such amalgamation nor have inscriptions been found there, as in Camboja, which explain this compound theology. It would seem that Buddhism and Brahmanism co-existed in the same districts but had not yet begun to fuse doctrinally. The same condition seems to have prevailed in western India during the seventh and eighth centuries, for the Buddhist caves of Ellora, though situated in the neighbourhood of Brahmanic buildings and approximating to them in style, contain sculptures which indicate a purely Buddhist cultus and not a mixed pantheon.

Our meagre knowledge of Javanese history makes it difficult to estimate the spheres and relative strength of the two religions. In the plains the Buddhist monuments are more numerous and also more ancient and we might suppose that the temples of Prambanan indicate the beginning of some change in belief. But the temples on the Dieng plateau seem to be of about the same age as the oldest Buddhist monuments. Thus nothing refutes the supposition that Brahmanism existed in Java from the time of the first Hindu colonists and that Buddhism was introduced after 400 A.D. It may be that Boroboedoer and the Dieng plateau represent the religious centres of two different kingdoms. But this supposition is not necessary for in India, whence the Javanese received their ideas, groups of temples are found of the same age but belonging to different sects. Thus in the Khajraho group[441] some shrines are Jain and of the rest some are dedicated to Siva and some to Vishnu.

The earliest records of Javanese Brahmanism, the inscriptions of Purnavarman, are Vishnuite but the Brahmanism which prevailed in the eighth and ninth centuries was in the main Sivaite, though not of a strongly sectarian type. Brahma, Vishnu and Siva were all worshipped both at Prambanan and on the Dieng but Siva together with Ganesa, Durga, and Nandi is evidently the chief deity. An image of Siva in the form of Bhatara Guru or Mahaguru is installed in one of the shrines at Prambanan. This deity is characteristic of Javanese Hinduism and apparently peculiar to it. He is represented as an elderly bearded man wearing a richly ornamented costume. There is something in the pose and drapery which recalls Chinese art and I think the figure is due to Chinese influence, for at the present day many of the images found in the temples of Bali are clearly imitated from Chinese models (or perhaps made by Chinese artists) and this may have happened in earlier times. The Chinese annals record several instances of religious objects being presented by the Emperors to Javanese princes. Though Bhatara Guru is only an aspect of Siva he is a sufficiently distinct personality to have a shrine of his own like Ganesa and Durga, in temples where the principal image of Siva is of another kind.

The same type of Brahmanism lasted at least until the erection of Panataran (c. 1150). The temple appears to have been dedicated to Siva but like Prambanan it is ornamented with scenes from the Ramayana and from Vishnuite Puranas.[442] The literature which can be definitely assigned to the reigns of Djajabaja and Erlangga is Brahmanic in tone but both literature and monuments indicate that somewhat later there was a revival of Buddhism. Something similar appears to have happened in other countries. In Camboja the inscriptions of Jayavarman VII (c. 1185 A.D.) are more definitely Buddhist than those of his predecessors and in 1296 Chou Ta-kuan regarded the country as mainly Buddhist. Parakrama Bahu of Ceylon (1153-1186) was zealous for the faith and so were several kings of Siam. I am inclined to think that this movement was a consequence of the flourishing condition of Buddhism at Pagan in Burma from 1050 to 1250. Pagan certainly stimulated religion in both Siam and Ceylon and Siam reacted strongly on Camboja.[443] It is true that the later Buddhism of Java was by no means of the Siamese type, but probably the idea was current that the great kings of the world were pious Buddhists and consequently in most countries the local form of Buddhism, whatever it was, began to be held in esteem. Java had constant communication with Camboja and Champa and a king of Madjapahit married a princess of the latter country. It is also possible that a direct stimulus may have been received from India, for the statement of Taranatha[444] that when Bihar was sacked by the Mohammedans the Buddhist teachers fled to other regions and that some of them went to Camboja is not improbable.

But though the prestige of Buddhism increased in the thirteenth century, no rupture with Brahmanism took place and Pali Buddhism does not appear to have entered Java. The unity of the two religions is proclaimed: Buddha and Siva are one. But the Kamahayanikan while admitting the Trimurti makes it a derivative, and not even a primary derivative, of the original Buddha spirit. It has been stated that the religion of Java in the Madjapahit epoch was Sivaism with a little Buddhism thrown in, on the understanding that it was merely another method of formulating the same doctrine. It is very likely that the bulk of the population worshipped Hindu deities, for they are the gods of this world and dispense its good things. Yet the natives still speak of the old religion as Buddhagama; the old times are "Buddha times" and even the flights of stairs leading up to the Dieng plateau are called Buddha steps. This would hardly be so if in the Madjapahit epoch Buddha had not seemed to be the most striking figure in the non-Mohammedan religion. Also, the majority of religious works which have survived from this period are Buddhist. It is true that we have the Ramayana, the Bharata Yuddha and many other specimens of Brahmanic literature. But these, especially in their Javanese dress, are belles lettres rather than theology, whereas Kamahayanikan and Kunjarakarna are dogmatic treatises. Hence it would appear that the religious life of Madjapahit was rooted in Buddhism, but a most tolerant Buddhism which had no desire to repudiate Brahmanism.

I have already briefly analysed the Sang Hyang Kamahayanikan which seems to be the most authoritative exposition of this creed. The learned editor has collected many parallels from Tibetan and Nepalese works and similar parallels between Javanese and Tibetan iconography have been indicated by Pleyte[445] and others. The explanation must be that the late forms of Buddhist art and doctrine which nourished in Magadha spread to Tibet and Nepal but were also introduced into Java. The Kamahayanikan appears to be a paraphrase of a Sanskrit original, perhaps distorted and mutilated. This original has not been identified with any work known to exist in India but might well be a Mahayanist catechism composed there about the eleventh century. The terminology of the treatise is peculiar, particularly in calling the ultimate principle Advaya and the more personal manifestation of it Divarupa. The former term may be paralleled in Hemacandra and the Amarakosha, which give respectively as synonyms for Buddha, advaya (in whom is no duality) and advayavadin (who preaches no duality), but Divarupa has not been found in any other work.[446] It is also remarkable that the Kamahayanikan does not teach the doctrine of the three bodies of Buddha.[447] It clearly states[448] that the Divarupa is identical with the highest being worshipped by various sects: with Paramasunya, Paramasiva, the Purusha of the followers of Kapila, the Nirguna of the Vishnuites, etc. Many names of sects and doctrines are mentioned which remain obscure, but the desire to represent them all as essentially identical is obvious.

The Kamahayanikan recognizes the theoretical identity of the highest principles in Buddhism and Vishnuism[449] but it does not appear that Vishnu-Buddha was ever a popular conception like Siva-Buddha or that the compound deity called Siva-Vishnu, Hari-Hara, Sankara-Narayana, etc., so well known in Camboja, enjoyed much honour in Java, Vishnu is relegated to a distinctly secondary position and the Javanese version of the Mahabharata is more distinctly Sivaite than the Sanskrit text. Still he has a shrine at Prambanan, the story of the Ramayana is depicted there and at Panataran, and various unedited manuscripts contain allusions to his worship, more especially to his incarnation as Narasimha and to the Garuda on which he rides.[450]

8

At present nearly all the inhabitants of Java profess Islam although the religion of a few tribes, such as the Tenggarese, is still a mixture of Hinduism with indigenous beliefs. But even among nominal Moslims some traces of the older creed survive. On festival days such monuments as Boroboedoer and Prambanan are frequented by crowds who, if they offer no worship, at least take pleasure in examining the ancient statues. Some of these however receive more definite honours: they are painted red and modest offerings of flowers and fruit are laid before them. Yet the respect shown to particular images seems due not to old tradition but to modern and wrongheaded interpretations of their meaning. Thus at Boroboedoer the relief which represents the good tortoise saving a shipwrecked crew receives offerings from women because the small figures on the tortoise's back are supposed to be children. The minor forms of Indian mythology still flourish. All classes believe in the existence of raksasas, boetas (bhutas) and widadaris (vidyadharis), who are regarded as spirits similar to the Jinns of the Arabs. Lakshmi survives in the female genius believed even by rigid Mohammedans to preside over the cultivation of rice and the somewhat disreputable sect known as Santri Birahis are said to adore devas and the forces of nature.[451] Less obvious, but more important as more deeply affecting the national character, is the tendency towards mysticism and asceticism. What is known as ngelmoe[452] plays a considerable part in the religious life of the modern Javanese. The word is simply the Arabic 'ilm (or knowledge) used in the sense of secret science. It sometimes signifies mere magic but the higher forms of it, such as the ngelmoe peling, are said to teach that the contemplative life is the way to the knowledge of God and the attainment of supernatural powers. With such ngelmoe is often connected a belief in metempsychosis, in the illusory nature of the world, and in the efficacy of regulating the breath. Asceticism is still known under the name of tapa and it is said that there are many recluses who live on alms and spend their time in meditation. The affinity of all this to Indian religion is obvious, although the Javanese have no idea that it is in any way incompatible with orthodox Islam.

Indian religion, which in Java is represented merely by the influence of the past on the present, is not dead in Bali[453] where, though much mixed with aboriginal superstitions, it is still a distinct and national faith, able to hold its own against Mohammedanism and Christianity.[454]

The island of Bali is divided from the east coast of Java only by a narrow strait but the inhabitants possess certain characters of their own. They are more robust in build, their language is distinct from Javanese though belonging to the same group, and even the alphabet presents idiosyncrasies. Their laws, social institutions, customs and calendar show many peculiarities, explicable on the supposition that they have preserved the ancient usages of pre-Mohammedan Java. At present the population is divided into the Bali-Agas or aborigines and the Wong Madjapahit who profess to have immigrated from that kingdom. The Chinese references[455] to Bali seem uncertain but, if accepted, indicate that it was known in the middle ages as a religious centre. It was probably a colony and dependency of Madjapahit and when Madjapahit fell it became a refuge for those who were not willing to accept Islam.

Caste is still a social institution in Bali, five classes being recognized, namely Brahmans, Kshatriyas (Satriyas), Vaisyas (Visias), Sudras and Parias. These distinctions are rigidly observed and though intermarriage (which in former times was often punished with death) is now permitted, the offspring are not recognized as belonging to the caste of the superior parent. The bodies of the dead are burned and Sati, which was formerly frequent, is believed still to take place in noble families. Pork is the only meat used and, as in other Hindu countries, oxen are never slaughtered.

An idea of the Balinese religion may perhaps be given most easily by describing some of the temples. These are very abundant: in the neighbourhood of Boeleling (the capital) alone I have seen more than ten of considerable size. As buildings they are not ancient, for the stone used is soft and does not last much more than fifty years. But when the edifices are rebuilt the ancient shape is preserved and what we see in Bali to-day probably represents the style of the middle ages. The temples consist of two or more courts surrounded by high walls. Worship is performed in the open air: there are various pyramids, seats, and small shrines like dovecots but no halls or rooms. The gates are ornamented with the heads of monsters, especially lions with large ears and winglike expansions at the side. The outermost gate has a characteristic shape. It somewhat resembles an Indian gopuram divided into two parts by a sharp, clean cut in the middle and tradition quotes in explanation the story of a king who was refused entrance to heaven but cleft a passage through the portal with his sword.

In the outer court stand various sheds and hollow wooden cylinders which when struck give a sound like bells. Another ornamented doorway leads to the second court where are found some or all of the following objects: (a) Sacred trees, especially Ficus elastica. (b) Sheds with seats for human beings. It is said that on certain occasions these are used by mediums who become inspired by the gods and then give oracles, (c) Seats for the gods, generally under sheds. They are of various kinds. There is usually one conspicuous chair with an ornamental back and a scroll hanging behind it which bears some such inscription as "This is the chair of the Bhatara." Any deity may be invited to take this seat and receive worship. Sometimes a stone linga is placed upon it. In some temples a stone chair, called padmasana, is set apart for Surya. (d) Small shrines two or three feet high, set on posts or pedestals. When well executed they are similar to the cabinets used in Japanese temples as shrines for images but when, as often happens, they are roughly made they are curiously like dovecots. On them are hung strips of dried palm-leaves in bunches like the Japanese gohei. As a rule the shrines contain no image but only a small seat and some objects said to be stones which are wrapped up in a cloth and called Artjeh.[456] In some temples (e.g. the Bale Agoeng at Singaraja) there are erections called Meru, supposed to represent the sacred mountain where the gods reside. They consist of a stout pedestal or basis of brick on which is erected a cabinet shrine as already described. Above this are large round discs made of straw and wood, which may be described as curved roofs or umbrellas. They are from three to five in number and rise one above the other, with slight intervals between them. (e) In many temples (for instance at Sangsit and Sawan) pyramidal erections are found either in addition to the Merus or instead of them. At the end of the second court is a pyramid in four stages or terraces, often with prolongations at the side of the main structure or at right angles to it. It is ascended by several staircases, consisting of about twenty-five steps, and at the top are rows of cabinet shrines.

Daily worship is not performed in these temples but offerings are laid before the shrines from time to time by those who need the help of the gods and there are several annual festivals. The object of the ritual is not to honour any image or object habitually kept in the temple but to induce the gods, who are supposed to be hovering round like birds, to seat themselves in the chair provided or to enter into some sacred object, and then receive homage and offerings. Thus both the ideas and ceremonial are different from those which prevail in Hindu temples and have more affinity with Polynesian beliefs. The deities are called Dewa, but many of them are indigenous nature spirits (especially mountain spirits) such as Dewa Gunung Agung, who are sometimes identified with Indian gods.

Somewhat different are the Durga temples. These are dedicated to the spirits of the dead but the images of Durga and her attendant Kaliki receive veneration in them, much as in Hindu temples. But on the whole the Malay or Polynesian element seemed to me to be in practice stronger than Hinduism in the religion of the Balinese and this is borne out by the fact that the Pemangku or priest of the indigenous gods ranks higher than the Pedanda or Brahman priest. But by talking to Balinese one may obtain a different impression, for they are proud of their connection with Madjapahit and Hinduism: they willingly speak of such subjects and Hindu deities are constantly represented in works of art. Ganesa, Indra, Vishnu, Krishna, Surya, Garuda and Siva, as well as the heroes of the Mahabharata, are well known but I have not heard of worship being offered to any of them except Durga and Siva under the form of the linga. Figures of Vishnu riding on Garuda are very common and a certain class of artificers are able to produce images of all well known Indian gods for those who care to order them. Many Indian works such as the Veda, Mahabharata, Ramayana, Brahmapurana and Nitisastra are known by name and are said to exist not in the original Sanskrit but in Kawi. I fancy that they are rarely read by the present generation, but any knowledge of them is much respected. The Balinese though confused in their theology are greatly attached to their religion and believe it is the ancient faith of Madjapahit.

I was unable to discover in the neighbourhood of Singaraja even such faint traces of Buddhism as have been reported by previous authors,[457] but they may exist elsewhere. The expression Siva-Buddha was known to the Pedandas but seemed to have no living significance, and perhaps certain families have a traditional and purely nominal connection with Buddhism. In Durga temples however I have seen figures described as Pusa, the Chinese equivalent of Bodhisattva, and it seems that Chinese artists have reintroduced into this miscellaneous pantheon an element of corrupt Buddhism, though the natives do not recognize it as such.

The art of Bali is more fantastic than that of ancient Java. The carved work, whether in stone or wood, is generally polychromatic. Figures are piled one on the top of another as in the sculptures of Central America and there is a marked tendency to emphasize projections. Leaves and flowers are very deeply carved and such features as ears, tongues and teeth are monstrously prolonged. Thus Balinese statues and reliefs have a curiously bristling and scaly appearance and are apt to seem barbaric, especially if taken separately.[458] Yet the general aspect of the temples is not unpleasing. The brilliant colours and fantastic outlines harmonize with the tropical vegetation which surrounds them and suggest that the guardian deities take shape as gorgeous insects. Such bizarre figures are not unknown in Indian mythology but in Balinese art Chinese influence is perhaps stronger than Indian. The Chinese probably frequented the island as early as the Hindus and are now found there in abundance. Besides the statues called Pusa already mentioned, Chinese landscapes are often painted behind the seats of the Devas and in the temple on the Volcano Batoer, where a special place is assigned to all the Balinese tribes, the Chinese have their own shrine. It is said that the temples in southern Bali which are older and larger than those in the north show even more decided signs of Chinese influence and are surrounded by stone figures of Chinese as guardians.

FOOTNOTES:

[Footnote 369: I have not been able to find anything more than casual and second-hand statements to the effect that Indian antiquities have been found in these islands.]

[Footnote 370: There is no lack of scholarly and scientific works about Java, but they are mostly written in Dutch and dissertations on special points are more numerous than general surveys of Javanese history, literature and architecture. Perhaps the best general account of the Hindu period in Java will be found in the chapter contributed by Kern to the publication called Neerlands Indie (Amsterdam, 1911, chap. VI. II. pp. 219-242). The abundant publications of the Bataviaasch Genootschap van Kunsten en Wetenschappen comprise Verhandelingen, Notulen, and the Tijdschrift voor Indische Taal-, Land-, en Volkenkunde (cited here as Tijdschrift), all of which contain numerous and important articles on history, philology, religion and archaeology. The last is treated specially in the publications called Archaeologisch Onderzoek op Java en Madura. Veth's Java, vols. I. and IV. and various articles in the Encyclopaedie van Nederlandsch-Indie may also be consulted. I have endeavoured to mention the more important editions of Javanese books as well as works dealing specially with the old religion in the notes to these chapters.

Although Dutch orthography is neither convenient nor familiar to most readers I have thought it better to preserve it in transcribing Javanese. In this system of transcription j=y; tj=ch; dj=j; sj=sh; w=v; oe=u.]

[Footnote 371: Ram. IV. 40. 30. Yavadvipam saptarajyopasobhitam Suvarnarupyakadvipam suvarnakaramanditam.]

[Footnote 372: Ptolemy's Geography, VII. 2. 29 (see also VIII. 27, 10). [Greek: Iabadiou (e Sabadiou), ho semainei krithes, nesos. Euphorotate de legetai he nesos einai kai eti pleiston chruson poiein, echein te metropolin onoma Arguren epi tois dusmikois perasin.]]

[Footnote 373: The Milinda Panha of doubtful but not very late date also mentions voyages to China.]

[Footnote 374: Groeneveldt, Notes on the Malay Archipelago compiled from Chinese sources, 1876 (cited below as Groeneveldt), p. 10. Confirmed by the statement in the Ming annals book 324 that in 1432 the Javanese said their kingdom had been founded 1376 years before.]

[Footnote 375: Kern in Versl. en Med. K. Ak. v. W. Afd. Lett. 3 Rks. I. 1884, pp. 5-12.]

[Footnote 376: Chap. XL. Legge, p. 113, and Groeneveldt, pp. 6-9.]

[Footnote 377: He perhaps landed in the present district of Rembang "where according to native tradition the first Hindu settlement was situated at that time" (Groeneveldt, p. 9).]

[Footnote 378: Groeneveldt, p. 9. The transcriptions of Chinese characters given in the following pages do not represent the modern sound but seem justified (though they cannot be regarded as certain) by the instances collected in Julien's Methode pour dechiffrer et transcrire les noms sanscrits. Possibly the syllables Do-a-lo-pa-mo are partly corrupt and somehow or other represent Purnavarman.]

[Footnote 379: Kern in Versl. en Meded, Afd. Lett. 2 R. XI. D. 1882.]

[Footnote 380: Groeneveldt, pp. 12, 13.]

[Footnote 381: Groeneveldt, p. 14.]

[Footnote 382: History of Java, vol. II. chap. X.]

[Footnote 383: Jackson, Java and Cambodja. App. IV. in Bombay Gazetteer, vol. I. part 1, 1896.]

[Footnote 384: It is also possible that when the Javanese traditions speak of Kaling they mean the Malay Peninsula. Indians in those regions were commonly known as Kaling because they came from Kalinga and in time the parts of the Peninsula where they were numerous were also called Kaling.]

[Footnote 385: See for this question Pelliot in B.E.F.E.O. 1904, pp. 274 ff. Also Schlegel in T'oung Pao, 1899, p. 247, and Chavannes, ib. 1904, p. 192.]

[Footnote 386: Chap. xxxix. Schiefner, p. 262.]

[Footnote 387: Though he expressly includes Camboja and Champa in Koki, it is only right to say that he mentions Nas-gling (=Yava-dvipa) separately in another enumeration together with Ceylon. But if Buddhists passed in any numbers from India to Camboja and vice versa, they probably appeared in Java about the same time, or rather later.]

[Footnote 388: See Kamaha. pp. 9, 10, and Watters, Yuan Chwang, II. pp. 209-214.]

[Footnote 389: They preserve to some extent the old civilization of Madjapahit. See the article "Tengereezen" in Encyclopaedie van Nederlandsch-Indie.]

[Footnote 390: See Kern, Kawi-studien Arjuna-vivaha, I. and II. 1871. Juynboll, Drie Boeken van het oudjavaansche Mahabharata, 1893, and id. Wirataparwwa, 1912. This last is dated Saka 918 = 996 A.D.]

[Footnote 391: Or Jayabaya.]

[Footnote 392: See Ramayana. Oudjavaansche Heldendicht, edited Kern, 1900, and Wrtta Sancaya, edited and translated by the same, 1875.]

[Footnote 393: Composed in 1613 A.D.]

[Footnote 394: Groeneveldt, p. 14.]

[Footnote 395: In the work commonly called "Nagarakretagama" (ed. Brandes, Verhand. Bataav. Genootschap. LIV. 1902), but it is stated that its real name is "Decawarnnana." See Tijdschrift, LVI. 1914, p. 194.]

[Footnote 396: Or Jayakatong.]

[Footnote 397: Groeneveldt, pp. 20-34.]

[Footnote 398: Groeneveldt, pp. 34-53.]

[Footnote 399: Near Soerabaja. It is said that he married a daughter of the king of Champa, and that the king of Madjapahit married her sister. For the connection between the royal families of Java and Champa at this period see Maspero in T'oung Pao, 1911, pp. 595 ff., and the references to Champa in Nagarakretagama, 15, 1, and 83, 4.]

[Footnote 400: See Raffles, chap, X, for Javanese traditions respecting the decline and fall of Madjapahit.]

[Footnote 401: See Takakusu, A record of the Buddhist religion, especially pp. xl to xlvi.]

[Footnote 402: In another pronunciation the characters are read San-fo-chai. The meaning appears to be The Three Buddhas.]

[Footnote 403: E.g. Si-li-ma-ha-la-sha (Srimaharaja) Si-li-tieh-hwa (perhapsSrideva).]

[Footnote 404: The conquest however was incomplete and about 1400 a Chinese adventurer ruled there some time. The name was changed to Ku-Kang, which is said to be still the Chinese name for Palembang.]

[Footnote 405: The Ming annals expressly state that the name was changed to Atjeh about 1600.]

[Footnote 406: For the identification of Po-li see Groeneveldt, p. 80, and Hose and McDougall, Pagan Tribes of Borneo, chap. II. It might be identified with Bali, but it is doubtful if Hindu civilization had spread to that island or even to east Java in the sixth century.]

[Footnote 407: See Hose and McDougall, l.c. p. 12.]

[Footnote 408: See Kern, "Over de Opschriften uit Koetei" in Verslagen Meded. Afd. Lett. 2 R. XI. D. Another inscription apparently written in debased Indian characters but not yet deciphered has been found in Sanggau, south-west Borneo.]

[Footnote 409: Groeneveldt, p. 81. The characters may be read Kau-di-nya according to Julien's method. The reference is to Liang annals, book 54.]

[Footnote 410: See Pleyte, Die Buddhalegende in den Sculpturen von Borobudur. But he points out that the version of the Lalita Vistara followed by the artist is not quite the same as the one that we possess.]

[Footnote 411: Amitabha, Amoghasiddhi, Ratnasambhava, Akshobhya, Vairocana, sometimes called Dhyani Buddhas, but it does not seem that this name was in common use in Java or elsewhere. The Kamahayanikan calls them the Five Tathagatas.]

[Footnote 412: So in the Kunjarakarna, for which see below. The Kamahayanikan teaches an elaborate system of Buddha emanations but for purposes of worship it is not quite clear which should be adored as the highest.]

[Footnote 413: Fergusson, History of Indian and Eastern Architecture, ed. 1910, vol. II. p. 439.]

[Footnote 414: See Archaeologisch Onderzoek op Java en Madura, I. "Tjandi Djago," 1904; II. "Tj. Singasari en Panataran," 1909.]

[Footnote 415: See Knebel in Tijds. voor Indische T., L. en Volkenkunde, 41, 1909, p. 27.]

[Footnote 416: See passages quoted in Archaeol. Onderzoek, I. pp. 96-97.]

[Footnote 417: Hayagriva however may be regarded as a Brahmanic god adopted by the Buddhists.]

[Footnote 418: See for reasons and references Archaeol. Onderzoek, II. pp. 36-40. The principal members of the king's household probably committed suicide during the funeral ceremonies.]

[Footnote 419: Kern in Tijds. voor T., L. en Volkenkunde, Deel LII. 1910, p. 107. Similarly in Burma Alompra was popularly regarded as a Bodhisattva.]

[Footnote 420: Sanskrit Kavi, a poet. See for Javanese literature Van der Tuuk in J.R.A.S. XIII. 1881, p. 42, and Hinloopen Labberton, ib. 1913, p. 1. Also the article "Litteratuur" in the Encyc. van Nederlandsch-Indie, and many notices in the writings of Kern and Veth.]

[Footnote 421: Edited by Gunning, 1903.]

[Footnote 422: A fragment of it is printed in Notulen. Batav. Gen. LII. 1914, 108.]

[Footnote 423: Episodes of the Indian epics have also been used as the subjects of Javanese dramas. See Juynboll, Indonesische en achterindische tooneelvoorstellingen uit het Ramayana, and Hinloopen Labberton, Pepakem Sapanti Sakoentala, 1912.]

[Footnote 424: Juynboll, Drie Boeken van het Oudjavaansche Mahabharata, p. 28.]

[Footnote 425: Archaeol. Onderzoek, I. p. 98. This statement is abundantly confirmed by Krom's index of the proper names in the Nagarakretagama in Tijdschrift, LVI. 1914, pp. 495 ff.]

[Footnote 426: Edited with transl. and notes by J. Kat, 's Gravenhage, 1910.]

[Footnote 427: Edited with transl. by H. Kern in Verh. der K. Akademie van Wetenschappen te Amsterdam. Afd. Lett. N.R. III. 3. 1901.]

[Footnote 428: But this probably represents nizbana and is not a Pali form. Cf. Bajra, Bayu for Vajra, Vayu.]

[Footnote 429: Adyabhishiktayushmanta, p. 30. Praptam buddhatvam bhavadbhir, ib. and Esha marga varah sriman mahayana mahodayah Yena yuyam gamishyanto bhavishyatha Tathagatah.]

[Footnote 430: Dana, sila, kshanti, virya, dhyana, prajna.]

[Footnote 431: Maitri, karuna, mudita, upeksha.]

[Footnote 432: The Karandavyuha teaches a somewhat similar doctrine of creative emanations. Avalokita, Brahma, Siva, Vishnu and others all are evolved from the original Buddha spirit and proceed to evolve the world.]

[Footnote 433: The use of this word, as a name for the residence of Vairocana, seems to be peculiar to our author.]

[Footnote 434: This term may include Sivaite ascetics as well as Buddhist monks.]

[Footnote 435: See further discussion in Kern's edition, p. 16.]

[Footnote 436: As are the Panchpirs in modern India.]

[Footnote 437: Garbha. Up. 1 and 3, especially the phrase asmin pancatmake sarire. Pinda Up. 2. Bhinne pancatmake dehe. Maha Nar. Up. 23. Sa va esha purushah pancadha pancatma.]

[Footnote 438: See Kern, "Over de Vermenging van Civaisme en Buddhisme op Java" in Vers. en Meded. der Kon. Akad. van Wet. Afd. Lett. 3 R. 5 Deel, 1888.

For the Sutasomajataka see Speyer's translation of the Jatakamala, pp. 291-313, with his notes and references. It is No. 537 in the Pali Collection of Jatakas.]

[Footnote 439: See Nanjio Cat. Nos. 137, 138.]

[Footnote 440: Gotama, Kassapa, Konagamana and Kakusandha.]

[Footnote 441: About 950-1050 A.D. Fergusson, Hist. of Indian Architecture, II. p. 141.]

[Footnote 442: See Knebel, "Recherches preparatoires concernant Krishna et les bas reliefs des temples de Java" in Tijdschrift, LI. 1909, pp. 97-174.]

[Footnote 443: In Camboja the result seems to have been double. Pali Buddhism entered from Siam and ultimately conquered all other forms of religion, but for some time Mahayanist Buddhism, which was older in Camboja, revived and received Court patronage.]

[Footnote 444: Chap. 37.]

[Footnote 445: "Bijdrage tot de Kennis van het Mahayana opJava" in Bijd. tot de Taal Lund en Volkenkunde van Nederlandsch-Indie, 1901 and 1902.]

[Footnote 446: This use of advaya and advayavadin strengthens the suspicion that the origins of the Advaita philosophy are to be sought in Buddhism.]

[Footnote 447: It uses the word trikaya but expressly defines it as meaning Kaya, vak and citta.]

[Footnote 448: In a passage which is not translated from the Sanskrit and may therefore reflect the religious condition of Java.]

[Footnote 449: So too in the Sutasoma Jataka Amoghasiddhi is said to be Vishnu.]

[Footnote 450: See Juynboll in Bijdragen tot de Taal Land en Volkenkunde van Ned.-Indie, 1908, pp. 412-420.]

[Footnote 451: Veth, Java, vol. IV. p. 154. The whole chapter contains much information about the Hindu elements in modern Javanese religion.]

[Footnote 452: See Veth, l.c. and ngelmoe in Encycl. van Nederlandsch-Indie. ]

[Footnote 453: Also to some extent in Lombok. The Balinese were formerly the ruling class in this island and are still found there in considerable numbers.]

[Footnote 454: It has even been suggested that hinduized Malays carried some faint traces of Indian religion to Madagascar. See T'oung Pao 1906, p. 93, where Zanahari is explained as Yang ( = God in Malay) Hari.]

[Footnote 455: Groeneveldt, pp. 19, 58, 59.]

[Footnote 456: This word appears to be the Sanskrit area, an image for worship.]

[Footnote 457: E.g. Van Eerde, "Hindu Javaansche en Balische Eeredienst" in Bijd. T.L. en Volkenkunde van Nederlandsch-Indie, 1910. I visited Bali in 1911.]

[Footnote 458: See Pleyte, Indonesian Art, 1901, especially the seven-headed figure in plate XVI said to be Krishna.]



CHAPTER XLI

CENTRAL ASIA

1

The term Central Asia is here used to denote the Tarim basin, without rigidly excluding neighbouring countries such as the Oxus region and Badakshan. This basin is a depression surrounded on three sides by high mountains: only on the east is the barrier dividing it from China relatively low. The water of the whole area discharges through the many branched Tarim river into Lake Lobnor. This so-called lake is now merely a flooded morass and the basin is a desert with occasional oases lying chiefly near its edges. The fertile portions were formerly more considerable but a quarter of a century ago this remote and lonely region interested no one but a few sportsmen and geographers. The results of recent exploration have been important and surprising. The arid sands have yielded not only ruins, statues and frescoes but whole libraries written in a dozen languages. The value of such discoveries for the general history of Asia is clear and they are of capital importance for our special subject, since during many centuries the Tarim region and its neighbouring lands were centres and highways for Buddhism and possibly the scene of many changes whose origin is now obscure. But I am unfortunate in having to discuss Central Asian Buddhism before scholars have had time to publish or even catalogue completely the store of material collected and the reader must remember that the statements in this chapter are at best tentative and incomplete. They will certainly be supplemented and probably corrected as year by year new documents and works of art are made known.

Tarim, in watery metaphor, is not so much a basin as a pool in a tidal river flowing alternately to and from the sea. We can imagine that in such a pool creatures of very different provenance might be found together. So currents both from east to west and from west to east passed through the Tarim, leaving behind whatever could live there: Chinese administration and civilization from the east: Iranians from the west, bearing with them in the stream fragments that had drifted from Asia Minor and Byzantium, while still other currents brought Hindus and Tibetans from the south.

One feature of special interest in the history of the Tarim is that it was in touch with Bactria and the regions conquered by Alexander and through them with western art and thought. Another is that its inhabitants included not only Iranian tribes but the speakers of an Aryan language hitherto unknown, whose presence so far east may oblige us to revise our views about the history of the Aryan race. A third characteristic is that from the dawn of history to the middle ages warlike nomads were continually passing through the country. All these people, whether we call them Iranians, Turks or Mongols had the same peculiarity: they had little culture of their own but they picked up and transported the ideas of others. The most remarkable example of this is the introduction of Islam into Europe and India. Nothing quite so striking happened in earlier ages, yet tribes similar to the Turks brought Manichaeism and Nestorian Christianity into China and played no small part in the introduction of Buddhism.

A brief catalogue of the languages represented in the manuscripts and inscriptions discovered will give a safe if only provisional idea of the many influences at work in Central Asia and its importance as a receiving and distributing centre. The number of tongues simultaneously in use for popular or learned purposes was remarkably large. To say nothing of great polyglot libraries like Tun-huang, a small collection at Toyog is reported as containing Indian, Manichaean, Syriac, Sogdian, Uigur and Chinese books. The writing materials employed were various like the idioms and include imported palm leaves, birch bark, plates of wood or bamboo, leather and paper, which last was in use from the first century A.D. onwards. In this dry atmosphere all enjoyed singular longevity.

Numerous Sanskrit writings have been found, all dealing with religious or quasi religious subjects, as medicine and grammar were then considered to be. Relatively modern Mahayanist literature is abundant but greater interest attaches to portions of an otherwise lost Sanskrit canon which agree in substance though not verbally with the corresponding passages in the Pali Canon and are apparently the original text from which much of the Chinese Tripitaka was translated. The manuscripts hitherto published include Sutras from the Samyukta and Ekottara Agamas, a considerable part of the Dharmapada, and the Pratimoksha of the Sarvastivadin school. Fa-Hsien states that the monks of Central Asia were all students of the language of India and even in the seventh century Hsuan Chuang tells us the same of Kucha. Portions of a Sanskrit grammar have been found near Turfan and in the earlier period at any rate Sanskrit was probably understood in polite and learned society. Some palm leaves from Ming-Oi contain fragments of two Buddhist religious dramas, one of which is the Sariputra-prakarana of Asvaghosha. The handwriting is believed to date from the epoch of Kanishka so that we have here the oldest known Sanskrit manuscripts, as well as the oldest specimens of Indian dramatic art.[459] They are written like the Indian classical dramas in Sanskrit and various forms of Prakrit. The latter represent hitherto unknown stages in the development of Indian dialects and some of them are closely allied to the language of Asoka's inscriptions. Another Prakrit text is the version of the Dharmapada written in Kharoshthi characters and discovered by the Dutreuil de Rhins mission near Khotan,[460] and numerous official documents in this language and alphabet have been brought home by Stein from the same region. It is probable that they are approximately coeval with the Kushan dynasty in India and the use of an Indian vernacular as well as of Sanskrit in Central Asia shows that the connection between the two countries was not due merely to the introduction of Buddhism.

Besides these hitherto unknown forms of Prakrit, Central Asia has astonished the learned world with two new languages, both written in a special variety of the Brahmi alphabet called Central Asian Gupta. One is sometimes called Nordarisch and is regarded by some authorities as the language of the Sakas whose incursions into India appear to have begun about the second century B.C. and by others as the language of the Kushans and of Kanishka's Empire. It is stated that the basis of the language is Iranian but strongly influenced by Indian idioms.[461] Many translations of Mahayanist literature (for instance the Suvarnaprabhasa, Vajracchedika and Aparimitayus Sutras) were made into it and it appears to have been spoken principally in the southern part of the Tarim basin.[462] The other new language was spoken principally on its northern edge and has been called Tokharian, which name implies that it was the tongue of the Tokhars or Indoscyths.[463] But there is no proof of this and it is safer to speak of it as the language of Kucha or Kuchanese. It exists in two different dialects known as A and B whose geographical distribution is uncertain but numerous official documents dated in the first half of the seventh century show that it was the ordinary speech of Kucha and Turfan. It was also a literary language and among the many translations discovered are versions in it of the Dharmapada and Vinaya. It is extremely interesting to find that this language spoken by the early and perhaps original inhabitants of Kucha not only belongs to the Aryan family but is related more nearly to the western than the eastern branch. It cannot be classed in the Indo-Iranian group but shows perplexing affinities to Latin, Greek, Keltic, Slavonic and Armenian.[464] It is possible that it influenced Chinese Buddhist literature.[465]

Besides the "Nordarisch" mentioned above which was written in Brahmi, three other Iranian languages have left literary remains in Central Asia, all written in an alphabet of Aramaic origin. Two of them apparently represent the speech of south-western Persia under the Sassanids, and of north-western Persia under the Arsacids. The texts preserved in both are Manichaean but the third Iranian language, or Sogdian, has a more varied literary content and offers Buddhist, Manichaean and Christian texts, apparently in that chronological order. It was originally the language of the region round Samarkand but acquired an international character for it was used by merchants throughout the Tarim basin and spread even to China. Some Christian texts in Syriac have also been found.

The Orkhon inscriptions exhibit an old Turkish dialect written in the characters commonly called Runes and this Runic alphabet is used in manuscripts found at Tun-huang and Miran but those hitherto published are not Buddhist. But another Turkish dialect written in the Uigur alphabet, which is derived from the Syriac, was (like Sogdian) extensively used for Buddhist, Manichaean and Christian literature. The name Uigur is perhaps more correctly applied to the alphabet than the language[466] which appears to have been the literary form of the various Turkish idioms spoken north and south of the Tien-shan. The use of this dialect for Buddhist literature spread considerably when the Uigurs broke the power of Tibet in the Tarim basin about 860 and founded a kingdom themselves: it extended into China and lasted long, for Sutras in Uigur were printed at Peking in 1330 and Uigur manuscripts copied in the reign of K'ang Hsi (1662-1723) are reported from a monastery near Suchow.[467] I am informed that a variety of this alphabet written in vertical columns is still used in some parts of Kansu where a Turkish dialect is spoken. Though Turkish was used by Buddhists in both the east and west of the Tarim basin, it appears to have been introduced into Khotan only after the Moslim conquest. Another Semitic script, hitherto unknown and found only in a fragmentary form, is believed to be the writing of the White Huns or Hephthalites.

As the Tibetans were the predominant power in the Tarim basin from at least the middle of the eighth until the middle of the ninth century, it is not surprising that great stores of Tibetan manuscripts have been found in the regions of Khotan, Miran and Tun-huang. In Turfan, as lying more to the north, traces of Tibetan influence, though not absent, are fewer. The documents discovered must be anterior to the ninth century and comprise numerous official and business papers as well as Buddhist translations.[468] They are of great importance for the history of the Tibetan language and also indicate that at the period when they were written Buddhism at most shared with the Bon religion the allegiance of the Tibetans. No Manichaean or Christian translations in Tibetan have yet been discovered.

Vast numbers of Chinese texts both religious and secular are preserved in all the principal centres and offer many points of interest among which two may be noticed. Firstly the posts on the old military frontier near Tun-huang have furnished a series of dated documents ranging from 98 B.C. to 153 A.D.[469] There is therefore no difficulty in admitting that there was intercourse between China and Central Asia at this period. Secondly, some documents of the T'ang dynasty are Manichaean, with an admixture of Buddhist and Taoist ideas.[470]

The religious monuments of Central Asia comprise stupas, caves and covered buildings used as temples or viharas. Buddhist, Manichaean and Christian edifices have been discovered but apparently no shrines of the Zoroastrian religion, though it had many adherents in these regions, and though representations of Hindu deities have been found, Hinduism is not known to have existed apart from Buddhism.[471] Caves decorated for Buddhist worship are found not only in the Tarim basin but at Tun-huang on the frontier of China proper, near Ta-t'ung-fu in northern Shensi, and in the defile of Lung-men in the province of Ho-nan. The general scheme and style of these caves are similar, but while in the last two, as in most Indian caves, the figures and ornaments are true sculpture, in the caves of Tun-huang and the Tarim not only is the wall prepared for frescoes, but even the figures are executed in stucco. This form of decoration was congenial to Central Asia for the images which embellished the temple walls were moulded in the same fashion. Temples and caves were sometimes combined, for instance at Bazaklik where many edifices were erected on a terrace in front of a series of caves excavated in a mountain corner. Few roofed buildings are well preserved but it seems certain that some were high quadrilateral structures, crowned by a dome of a shape found in Persia, and that others had barrel-shaped roofs, apparently resembling the chaityas of Ter and Chezarla.[472] Le Coq states that this type of architecture is also found in Persia.[473] The commonest type of temple was a hall having at its further end a cella, with a passage behind to allow of circumambulation. Such halls were frequently enlarged by the addition of side rooms and sometimes a shrine was enclosed by several rectangular courts.[474]

Many stupas have been found either by themselves or in combination with other buildings. The one which is best preserved (or at any rate reproduced in greatest detail)[475] is the Stupa of Rawak. It is set in a quadrangle bounded by a wall which was ornamented on both its inner and outer face by a series of gigantic statues in coloured stucco. The dome is set upon a rectangular base disposed in three stories and this arrangement is said to characterize all the stupas of Turkestan as well as those of the Kabul valley and adjacent regions.

This architecture appears to owe nothing to China but to include both Indian (especially Gandharan) and Persian elements. Many of its remarkable features, if not common elsewhere, are at least widely scattered. Thus some of the caves at Ming-Oi have dome-like roofs ornamented with a pattern composed of squares within squares, set at an angle with each other. A similar ornamentation is reported from Pandrenthan in Kashmir and from Bamian.[476]

The antiquities of Central Asia include frescoes executed on the walls of caves and buildings, and paintings on silk paper.[477] The origin and affinities of this art are still the subject of investigation and any discussion of them would lead me too far from my immediate subject. But a few statements can be made with some confidence. The influence of Gandhara is plain in architecture, sculpture, and painting. The oldest works may be described as simply Gandharan but this early style is followed by another which shows a development both in technique and in mythology. It doubtless represents Indian Buddhist art as modified by local painters and sculptors. Thus in the Turfan frescoes the drapery and composition are Indian but the faces are eastern asiatic. Sometimes however they represent a race with red hair and blue eyes.

On the whole the paintings testify to the invasion of Far Eastern art by the ideas and designs of Indian Buddhism rather than to an equal combination of Indian and Chinese influence but in some forms of decoration, particularly that employed in the Khan's palace at Idiqutshahri,[478] Chinese style is predominant. It may be too that the early pre-buddhist styles of painting in China and Central Asia were similar. In the seventh century a Khotan artist called Wei-ch'ih Po-chih-na migrated to China, where both he and his son Wei-ch'ih I-seng acquired considerable fame.

Persian influence also is manifest in many paintings. A striking instance may be seen in two plates published by Stein[479] apparently representing the same Boddhisattva. In one he is of the familiar Indian type: the other seems at first sight a miniature of some Persian prince, black-bearded and high-booted, but the figure has four arms. As might be expected, it is the Manichaean paintings which are least Indian in character. They represent a "lost late antique school"[480] which often recalls Byzantine art and was perhaps the parent of mediaeval Persian miniature painting.