i.e. "particle." It is perhaps to the separation of these categories that the name for grammar, vyakarana, originally referred, rather than to the analysis of words. Even the Brahmanas bear evidence of linguistic investigations, for they mention various grammatical terms, such as "letter" (varna), "masculine" (vrishan), "number"

(vachana), "case-form" (vibhakti).Still more such references are to be found in the Aranyakas, the Upanishads, and the Sutras. But the most important information we have of pre-Paninean grammar is that found in Yaska's work.

Grammatical studies must have been cultivated to a considerable extent before Yaska's time, for he distinguishes a Northern and an Eastern school, besides mentioning nearly twenty predecessors, among whom cakatayana, Gargya, and cakalya are the most important. By the time of Yaska grammarians had learned to distinguish clearly between the stem and the formative elements of words; recognising the personal terminations and the tense affixes of the verb on the one hand, and primary (krit) or secondary (taddhita) nominal suffixes on the other. Yaska has an interesting discussion on the theory of cakatayana, which he himself follows, that nouns are derived from verbs. Gargya and some other grammarians, he shows, admit this theory in a general way, but deny that it is applicable to all nouns. He criticises their objections, and finally dismisses them as untenable. On cakatayana's theory of the verbal origin of nouns the whole system of Panini is founded. The sutra of that grammarian contains hundreds of rules dealing with Vedic forms; but these are of the nature of exceptions to the main body of his rules, which are meant to describe the Sanskrit language. His work almost entirely dominates the subsequent literature. Though belonging to the middle of the Sutra period, it must be regarded as the definite starting-point of the post-Vedic age. Coming to be regarded as an infallible authority, Panini superseded all his predecessors, whose works have consequently perished. Yaska alone survives, and that only because he was not directly a grammarian; for his work represents, and alone represents, the Vedanga "etymology."

Yaska's Nirukta is in reality a Vedic commentary, and is older by some centuries than any other exegetical work preserved in Sanskrit. Its bases are the Nighantus, collections of rare or obscure Vedic words, arranged for the use of teachers. Yaska had before him five such collections. The first three contain groups of synonyms, the fourth specially difficult words, and the fifth a classification of the Vedic gods. These Yaska explained for the most part in the twelve books of his commentary (to which two others were added later). In so doing he adduces as examples a large number of verses, chiefly from the Rigveda, which he interprets with many etymological remarks.

The first book is an introduction, dealing with the principles of grammar and exegesis. The second and third elucidate certain points in the synonymous nighantus; Books IV.-VI. comment on the fourth section, and VII.-XII. on the fifth. The Nirukta, besides being very important from the point of view of exegesis and grammar, is highly interesting as the earliest specimen of Sanskrit prose of the classical type, considerably earlier than Panini himself. Yaska already uses essentially the same grammatical terminology as Panini, employing, for instance, the same words for root (dhatu), primary, and secondary suffixes. But he must have lived a long time before Panini; for a considerable number of important grammarians' names are mentioned between them. Yaska must, therefore, go back to the fifth century, and undoubtedly belongs to the beginning of the Sutra period.

One point of very great importance proved by the Nirukta is that the Rigveda had a very fixed form in Yaska's time, and was essentially identical with our text. His deviations are very insignificant. Thus in one passage (X. 29. I) he reads vayo as one word, against va yo as two words in cakalya's Pada text. Yaska's paraphrases show that he also occasionally differed from the Samhita text, though the quotations themselves from the Rigveda have been corrected so as to agree absolutely with the traditional text. But these slight variations are probably due to mistakes in the Nirukta rather than to varieties of reading in the Rigveda. There are a few insignificant deviations of this kind even in Sayana, but they are always manifestly oversights on the part of the commentator.

To the Sutras is attached a very extensive literature of Paricishtas or "supplements," which seem to have existed in all the Vedic schools. They contain details on matters only touched upon in the Sutras, or supplementary information about subjects not dealt with at all by them. Thus, there is the Acvalayana Grihya-paricishta, in four chapters, connected with the Rigveda. The Gobhila samgraha-paricishta is a compendium of Grihya practices in general, with a special leaning towards magical rites, which came to be attached to the Samaveda. Closely related to, and probably later than this work, is the Karma-pradipa ("lamp of rites"), also variously called sama-grihya- or chhandogyagrihya-paricishta, chhandoga-paricishta, Gobhila-smriti, attributed to the Katyayana of the White Yajurveda or to Gobhila. It deals with the same subjects, though independently, as the Grihya samgraha, with which it occasionally agrees in whole clokas.

Of great importance for the understanding of the sacrificial ceremonial are the Prayogas ("Manuals") and Paddhatis ("Guides"), of which a vast number exist in manuscript. These works represent both the crauta and the Grihya ritual according to the various schools. The Prayogas describe the course of each sacrifice and the functions of the different groups of priests, solely from the point of view of practical performance, while the Paddhatis rather follow the systematic accounts of the Sutras and sketch their contents. There are also versified accounts of the ritual called Karikas, which are directly attached to Sutras or to Paddhatis. The oldest of them appears to be the Karika of Kumarila (c. 700 A.D.).

Of a supplementary character are also the class of writings called Anukramanis or Vedic Indices, which give lists of the hymns, the authors, the metres, and the deities in the order in which they occur in the various Samhitas. To the Rigveda belonged seven of these works, all attributed to caunaka, and composed in the mixture of the cloka and trishtubh metre, which is also found in caunaka's Rigveda Praticakhya. There is also a General Index or Sarvanukramani which is attributed to Katyayana, and epitomises in the Sutra style the contents of the metrical indices. Of the metrical indices five have been preserved. The Arshanukramani, containing rather less than 300 clokas, gives a list of the Rishis or authors of the Rigveda. Its present text represents a modernised form of that which was known to the commentator Shadgurucishya in the twelfth century. The Chhandonukramani, which is of almost exactly the same length, enumerates the metres in which the hymns of the Rigveda are composed. It also states for each book the number of verses in each metre as well as the aggregate in all metres. The Anuvakanukramani is a short index containing only about forty verses. It states the initial words of each of the eighty-five anuvakas or lessons into which the Rigveda is divided, and the number of hymns contained in these anuvakas. It further states that the Rigveda contains 1017 hymns (or 1025 according to the Vashkala recension), 10,580-1/2 verses, 153,826 words, 432,000 syllables, besides some other statistical details. The number of verses given does not exactly tally with various calculations that have recently been made, but the differences are only slight, and may be due to the way in which certain repeated verses were counted by the author of the index.

There is another short index, known as yet only in two MSS., called the Padanukramani, or "index of lines" (padas), and composed in the same mixed metre as the others. The Suktanukramani, which has not survived, and is only known by name, probably consisted only of the initial words (pratikas) of the hymns. It probably perished because the Sarvanukramani would have rendered such a work superfluous. No MS. of the Devatanukramani or "Index of gods" exists, but ten quotations from it have been preserved by the commentator Shadgurucishya. It must have been superseded by the Brihaddevata, an index of the "many gods,"

a much more extensive work than any of the other Anukramanis, as it contains about 1200 clokas interspersed with occasional trishtubhs. It is divided into eight adhyayas corresponding to the ashtakas of the Rigveda. Following the order of the Rigveda, its main object is to state the deity for each verse. But as it contains a large number of illustrative myths and legends, it is of great value as an early collection of stories. It is to a considerable extent based on Yaska's Nirukta. Besides Yaska himself and other teachers named by that scholar, it also mentions Bhaguri and Acvalayana as well as the Nidana Sutra, A peculiarity of this work is that it refers to a number of supplementary hymns (khilas) which do not form part of the canonical text of the Rigveda.

Later, at least, than the original form of these metrical Anukramanis, is the Sarvanukramani of Katyayana, which combines the data contained in them within the compass of a single work. Composed in the Sutra style, it is of considerable length, occupying about forty-six pages in the printed edition. For every hymn in the Rigveda it states the initial word or words, the number of its verses, as well as the author, the deity, and the metre, even for single verses. There is an introduction in twelve sections, nine of which form a short treatise on Vedic metres corresponding to the last three sections of the Rigveda Praticakhya. The author begins with the statement that he is going to supply an index of the pratikas and so forth of the Rigveda according to the authorities (yathopadecam), because without such knowledge the crauta and Smarta rites cannot be accomplished. These authorities are doubtless the metrical indices described above. For the text of the Sarvanukramani, which is composed in a concise Sutra style, not only contains some metrical lines (padas), but also a number of passages either directly taken from the Arshanukramani and the Brihaddevata, or with their metrical wording but slightly altered. Another metrical work attributed to caunaka is the Rigvidhana, which describes the magical effects produced by the recitation of hymns or single verses of the Rigveda.

To the Paricishtas of the Samaveda belong the two indices called Arsha and Daivata, enumerating respectively the Rishis and deities of the text of the Naigeya branch of the Samaveda. They quote Yaska, caunaka, and Acvalayana among others. There are also two Anukramanis attached to the Black Yajurveda. That of the Atreya school consists of two parts, the first of which is in prose, and the second in clokas. It contains little more than an enumeration of names referring to the contents of its Samhita. The Anukramani of the Charayaniya school of the Kathaka is an index of the authors of the various sections and verses. Its statements regarding passages derived from the Rigveda differ much from those of the Sarvanukramani of the Rigveda, giving a number of totally new names. It claims to be the work of Atri, who communicated it to Laugakshi. The Anukramani of the White Yajurveda in the Madhyamdina recension, attributed to Katyayana, consists of five sections. The first four are an index of authors, deities, and metres. The authors of verses taken from the Rigveda generally agree with those in the Sarvanukramani. There are, however, a good many exceptions, several new names belonging to a later period, some even to that of the catapatha Brahmana. The fifth section gives a summary account of the metres occurring in the text. It is identical with the corresponding portion of the introduction to the Sarvanukramani, which was probably the original position of the section. There are many other Paricishtas of the White Yajurveda, all attributed to Katyayana. Only three of these need be mentioned here. The Nigama-paricishta, a glossary of synonymous words occurring in the White Yajurveda, has a lexicographical interest. The Pravaradhyaya, or "Chapter on Ancestors," is a list of Brahman families drawn up for the purpose of determining the forbidden degrees of relationship in marriage, and of indicating the priests suitable for the performance of sacrifice. The Charana-vyuha, or "Exposition of the Schools"

of the various Vedas, is a very late work of little importance, giving a far less complete enumeration of the Vedic schools than certain sections of the Vishnu- and the Vayu-Purana. There is also a Charana-vyuha among the Paricishtas of the Atharva-veda, which number upwards of seventy. This work makes the statement that the Atharva contains 2000 hymns and 12,380 verses.

In concluding this account of Vedic literature, I cannot omit to say a few words about Sayana, the great mediaeval Vedic scholar, to whom or to whose initiation we owe a number of valuable commentaries on the Rigveda, the Aitareya Brahmana and Aranyaka, as well as the Taittiriya Samhita, Brahmana, and Aranyaka, besides a number of other works. His comments on the two Samhitas would appear to have been only partially composed by himself and to have been completed by his pupils. He died in 1387, having written his works under Bukka I. (1350-79), whose teacher and minister he calls himself, and his successor, Harihara (1379-99). These princes belonged to a family which, throwing off the Muhammadan yoke in the earlier half of the fourteenth century, founded the dynasty of Vijayanagara ("city of victory"), now Hampi, on the Tungabhadra, in the Bellary district. Sayana's elder brother, Madhava, was minister of King Bukka, and died as abbot of the monastery of cringeri, under the name of Vidyaranyasvamin. Not only did he too produce works of his own, but Sayana's commentaries, as composed under his patronage, were dedicated to him as madhaviya, or ("influenced by Madhava"). By an interesting coincidence Professor Max Muller's second edition of the Rigveda, with the commentary of Sayana, was brought out under the auspices of a Maharaja of Vijayanagara. The latter city has, however, nothing to do with that from which King Bukka derived his title.

CHAPTER X

THE EPICS

(Circa 500-50 B.C.)

In turning from the Vedic to the Sanskrit period, we are confronted with a literature which is essentially different from that of the earlier age in matter, spirit, and form. Vedic literature is essentially religious; Sanskrit literature, abundantly developed in every other direction, is profane. But, doubtless as a result of the speculative tendencies of the Upanishads, a moralising spirit at the same time breathes through it as a whole. The religion itself which now prevails is very different from that of the Vedic age. For in the new period the three great gods, Brahma, Vishnu, and civa are the chief objects of worship. The important deities of the Veda have sunk to a subordinate position, though Indra is still relatively prominent as the chief of a warrior's heaven. Some new gods of lesser rank have arisen, such as Kubera, god of wealth; Ganeca, god of learning; Karttikeya, god of war; cri or Lakshmi, goddess of beauty and fortune; Durga or Parvati, the terrible spouse of civa; besides the serpent deities and several classes of demigods and demons.

While the spirit of Vedic literature, at least in its earlier phase, is optimistic, Sanskrit poetry is pervaded by Weltschmerz, resulting from the now universally accepted doctrine of transmigration. To that doctrine, according to which beings pass by gradations from Brahma through men and animals to the lowest forms of existence, is doubtless also largely due the fantastic element characteristic of this later poetry. Here, for instance, we read of Vishnu coming down to earth in the shape of animals, of sages and saints wandering between heaven and earth, of human kings visiting Indra in heaven.

Hand in hand with this fondness for introducing the marvellous and supernatural into the description of human events goes a tendency to exaggeration. Thus King Vicvamitra, we are told, practised penance for thousands of years in succession; and the power of asceticism is described as so great as to cause even the worlds and the gods to tremble. The very bulk of the Mahabharata, consisting as it does of more than 200,000 lines, is a concrete illustration of this defective sense of proportion.

As regards the form in which it is presented to us, Sanskrit literature contrasts with that of both the earlier and the later Vedic period. While prose was employed in the Yajurvedas and the Brahmanas, and finally attained to a certain degree of development, it almost disappears in Sanskrit, nearly every branch of literature being treated in verse, often much to the detriment of the subject, as in the case of law. The only departments almost entirely restricted to the use of prose are grammar and philosophy, but the cramped and enigmatical style in which these subjects are treated hardly deserves the name of prose at all. Literary prose is found only in fables, fairy tales, romances, and partially in the drama. In consequence of this neglect, the prose of the later period compares unfavourably with that of the Brahmanas. Even the style of the romances or prose kavyas, subject as it is to the strict rules of poetics, is as clumsy as that of the grammatical commentaries; for the use of immense compounds, like those of the Sutras, is one of its essential characteristics.

Sanskrit literature, then, resembles that of the earlier Vedic age in being almost entirely metrical. But the metres in which it is written, though nearly all based on those of the Veda, are different. The bulk of the literature is composed in the cloka, a development of the Vedic anushtubh stanza of four octosyllabic lines; but while all four lines ended iambically in the prototype, the first and third line have in the cloka acquired a trochaic rhythm. The numerous other metres employed in the classical poetry have become much more elaborate than their Vedic originals by having the quantity of every syllable in the line strictly determined.

The style, too, excepting the two old epics, is in Sanskrit poetry made more artificial by the frequent use of long compounds, as well as by the application of the elaborate rules of poetics, while the language is regulated by the grammar of Panini. Thus classical Sanskrit literature, teeming as it does with fantastic and exaggerated ideas, while bound by the strictest rules of form, is like a tropical garden full of luxuriant and rank growth, in which, however, many a fair flower of true poetry may be culled.

It is impossible even for the Sanskrit scholar who has not lived in India to appreciate fully the merits of this later poetry, much more so for those who can only become acquainted with it in translations. For, in the first place, the metres, artificial and elaborate though they are, have a beauty of their own which cannot be reproduced in other languages. Again, to understand it thoroughly, the reader must have seen the tropical plains and forests of Hindustan steeped in intense sunshine or bathed in brilliant moonlight; he must have viewed the silent ascetic seated at the foot of the sacred fig-tree; he must have experienced the feelings inspired by the approach of the monsoon; he must have watched beast and bird disporting themselves in tank and river; he must know the varying aspects of Nature in the different seasons; in short, he must be acquainted with all the sights and sounds of an Indian landscape, the mere allusion to one of which may call up some familiar scene or touch some chord of sentiment. Otherwise, for instance, the mango-tree, the red Acoka, the orange Kadamba, the various creepers, the different kinds of lotus, the mention of each of which should convey a vivid picture, are but empty names. Without a knowledge, moreover, of the habits, modes of thought, and traditions of the people, much must remain meaningless. But those who are properly equipped can see many beauties in classical Sanskrit poetry which are entirely lost to others. Thus a distinguished scholar known to the present writer has entered so fully into the spirit of that poetry, that he is unable to derive pleasure from any other.

It would be a mistake to suppose that Sanskrit literature came into being only at the close of the Vedic period, or that it merely forms its continuation and development. As a profane literature, it must, in its earliest phases, which are lost, have been contemporaneous with the religious literature of the Vedas. Beside the productions of the latest Vedic period, that of the Upanishads and Sutras, there grew up, on the one hand, the rich Pali literature of Buddhism, and, on the other, the earliest form of Sanskrit poetry in the shape of epic tales. We have seen that even the Rigveda contains some hymns of a narrative character. Later we find in the Brahmanas a number of short legends, mostly in prose, but sometimes partly metrical, as the story of cunahcepa in the Aitareya. Again, the Nirukta, which must date from the fifth century B.C., contains many prose tales, and the oldest existing collection of Vedic legend, the metrical Brihaddevata, cannot belong to a much later time.

Sanskrit epic poetry falls into two main classes. That which comprises old stories goes by the name of Itihasa, "legend," Akhyana, "narrative," or Purana, "ancient tale," while the other is called Kavya or artificial epic. The Mahabharata is the chief and oldest representative of the former group, the Ramayana of the latter. Both these great epics are composed in the same form of the cloka metre as that employed in classical Sanskrit poetry. The Mahabharata, however, also contains, as remnants of an older phase, archaic verses in the upajati and vamcastha (developments of the Vedic trishtubh and jagati) metres, besides preserving some old prose stories in what is otherwise an entirely metrical work. It further differs from the sister epic in introducing speeches with words, such as "Brihadacva spake," which do not form part of the verse, and which may be survivals of prose narrative connecting old epic songs. The Ramayana, again, is, in the main, the work of a single poet, homogeneous in plan and execution, composed in the east of India. The Mahabharata, arising in the western half of the country, is a congeries of parts, the only unity about which is the connectedness of the epic cycle with which they deal; its epic kernel, moreover, which forms only about one-fifth of the whole work, has become so overgrown with didactic matter, that in its final shape it is not an epic at all, but an encyclopaedia of moral teaching.

The Mahabharata, which in its present form consists of over 100,000 clokas, equal to about eight times as much as the Iliad and Odyssey put together, is by far the longest poem known to literary history. It is a conglomerate of epic and didactic matter divided into eighteen books called parvan, with a nineteenth, the Harivamca, as a supplement. The books vary very considerably in length, the twelfth being the longest, with nearly 14,000, the seventeenth the shortest, with only 312 clokas. All the eighteen books, excepting the eighth and the last three, are divided into subordinate parvans; each book is also cut up into chapters (adhyayas).

No European edition of the whole epic has yet been undertaken. This remains one of the great tasks reserved for the future of Sanskrit philology, and can only be accomplished by the collaboration of several scholars. There are complete MSS. of the Mahabharata in London, Oxford, Paris, and Berlin, besides many others in different parts of India; while the number of MSS. containing only parts of the poem can hardly be counted.

Three main editions of the epic have appeared in India. The editio princeps, including the Harivamca, but without any commentary, was published in four volumes at Calcutta in 1834-39. Another and better edition, which has subsequently been reproduced several times, was printed at Bombay in 1863. This edition, though not including the supplementary book, contains the commentary of Nilakantha. These two editions do not on the whole differ considerably. Being derived from a common source, they represent one and the same recension. The Bombay edition, however, generally has the better readings. It contains about 200 clokas more than the Calcutta edition, but these additions are of no importance.

A third edition, printed in Telugu characters, was published in four volumes at Madras in 1855-60. It includes the Harivamca and extracts from Nilakantha's commentary. This edition represents a distinct South Indian recension, which seems to differ from that of the North about as much as the three recensions of the Ramayana do from one another. Both recensions are of about equal length, omissions in the first being compensated by others in the second. Sometimes one has the better text, sometimes the other.

The epic kernel of the Mahabharata or the "Great Battle of the descendants of Bharata," consisting of about 20,000 clokas, describes the eighteen days' fight between Duryodhana, leader of the Kurus, and Yudhishthira, chief of the Pandus, who were cousins, both descended from King Bharata, son of cakuntala. Within this narrative frame has come to be included a vast number of old legends about gods, kings, and sages; accounts of cosmogony and theogony; disquisitions on philosophy, law, religion, and the duties of the military caste. These lengthy and heterogeneous interpolations render it very difficult to follow the thread of the narrative. Entire works are sometimes inserted to illustrate a particular statement. Thus, while the two armies are drawn up prepared for battle, a whole philosophical poem, in eighteen cantos, the Bhagavadgita is recited to the hero Arjuna, who hesitates to advance and fight against his kin. Hence the Mahabharata claims to be not only a heroic poem (kavya), but a compendium teaching, in accordance with the Veda, the fourfold end of human existence (spiritual merit, wealth, pleasure, and salvation), a smriti or work of sacred tradition, which expounds the whole duty of man, and is intended for the religious instruction of all Hindus. Thus, in one (I. lxii. 35) of many similar passages, it makes the statement about itself that "this collection of all sacred texts, in which the greatness of cows and Brahmans is exalted, must be listened to by virtuous-minded men." Its title, Karshna Veda, or "Veda of Krishna" (a form of Vishnu), the occurrence of a famous invocation of Narayana and Nara (names of Vishnu) and Sarasvati (Vishnu's wife) at the beginning of each of its larger sections, and the prevalence of Vishnuite doctrines throughout the work, prove it to have been a smriti of the ancient Vishnuite sect of the Bhagavatas.

Thus it is clear that the Mahabharata in its present shape contains an epic nucleus, that it favours the worship of Vishnu, and that it has become a comprehensive didactic work. We further find in Book I. the direct statements that the poem at one time contained 24,000 clokas before the episodes (upakhyana) were added, that it originally consisted of only 8800 clokas, and that it has three beginnings. These data render it probable that the epic underwent three stages of development from the time it first assumed definite shape; and this conclusion is corroborated by various internal and external arguments.

There can be little doubt that the original kernel of the epic has as a historical background an ancient conflict between the two neighbouring tribes of the Kurus and Panchalas, who finally coalesced into a single people. In the Yajurvedas these two tribes already appear united, and in the Kathaka King Dhritarashtra Vaichitravirya, one of the chief figures of the Mahabharata, is mentioned as a well-known person. Hence the historical germ of the great epic is to be traced to a very early period, which cannot well be later than the tenth century B.C. Old songs about the ancient feud and the heroes who played a part in it, must have been handed down by word of mouth and recited in popular assemblies or at great public sacrifices.

These disconnected battle-songs were, we must assume, worked up by some poetic genius into a comparatively short epic, describing the tragic fate of the Kuru race, who, with justice and virtue on their side, perished through the treachery of the victorious sons of Pandu, with Krishna at their head. To the period of this original epic doubtless belong the traces the Mahabharata has preserved unchanged of the heroic spirit and the customs of ancient times, so different from the later state of things which the Mahabharata as a whole reflects. To this period also belongs the figure of Brahma as the highest god. The evidence of Pali literature shows that Brahma already occupied that position in Buddha's time. We may, then, perhaps assume that the original form of our epic came into being about the fifth century B.C. The oldest evidence we have for the existence of the Mahabharata in some shape or other is to be found in Acvalayana's Grihya Sutra, where a Bharata and Mahabharata are mentioned. This would also point to about the fifth century B.C.

To the next stage, in which the epic, handed down by rhapsodists, swelled to a length of about 20,000 clokas, belongs the representation of the victorious Pandus in a favourable light, and the introduction on a level with Brahma of the two other great gods, civa, and especially Vishnu, of whom Krishna appears as an incarnation.

We gather from the account of Megasthenes that about 300 B.C., these two gods were already prominent, and the people were divided into civaites and Vishnuites. Moreover, the Yavanas or Greeks are mentioned in the Mahabharata as allies of the Kurus, and even the cakas (Scythians) and Pahlavas (Parthians) are named along with them; Hindu temples are also referred to as well as Buddhist relic mounds. Thus an extension of the original epic must have taken place after 300 B.C. and by the beginning of our era.

The Brahmans knew how to utilise the great influence of the old epic tradition by gradually incorporating didactic matter calculated to impress upon the people, and especially on kings, the doctrines of the priestly caste. It thus at last assumed the character of a vast treatise on duty (dharma), in which the divine origin and immutability of Brahman institutions, the eternity of the caste system, and the subordination of all to the priests, are laid down. When the Mahabharata attributes its origin to Vyasa, it implies a belief in a final redaction, for the name simply means "Arranger." Dahlmann has recently put forward the theory that the great epic was a didactic work from the very outset; this view, however, appears to be quite irreconcilable with the data of the poem, and is not likely to find any support among scholars.

What evidence have we as to when the Mahabharata attained to the form in which we possess it? There is an inscription in a land grant dating from 462 A.D. or at the latest 532 A.D., which proves incontrovertibly that the epic about 500 A.D. was practically of exactly the same length as it is stated to have in the survey of contents (anukramanika) given in Book I., and as it actually has now; for it contains the following words: "It has been declared in the Mahabharata, the compilation embracing 100,000 verses, by the highest sage, Vyasa, the Vyasa of the Vedas, the son of Paracara." This quotation at the same time proves that the epic at that date included the very long 12th and 13th, as well as the extensive supplementary book, the Harivamca, without any one of which it would have been impossible to speak even approximately of 100,000 verses. There are also several land grants, dated between 450 and 500 A.D., and found in various parts of India, which quote the Mahabharata as an authority teaching the rewards of pious donors and the punishments of impious despoilers. This shows that in the middle of the fifth century it already possessed the same character as at present, that of a Smriti or Dharmacastra. It is only reasonable to suppose that it had acquired this character at least a century earlier, or by about 350 A.D. Further research in the writings of the Northern Buddhists and their dated Chinese translations will probably enable us to put this date back by some centuries. We are already justified in considering it likely that the great epic had become a didactic compendium before the beginning of our era. In any case, the present state of our knowledge entirely disproves the suggestions put forward by Prof. Holtzmann in his work on the Mahabharata, that the epic was turned into a Dharmacastra by the Brahmans after 900 A.D., and that whole books were added at this late period.

The literary evidence of Sanskrit authors from about 600 to 1100 A.D. supplies us with a considerable amount of information as to the state of the great epic during those five centuries. An examination of the works of Bana, and of his predecessor Subandhu, shows that these authors, who belong to the beginning of the seventh century, not only studied and made use of legends from every one of the eighteen books of the Mahabharata for the poetical embellishment of their works, but were even acquainted with the Harivamca. We also know that in Bana's time the Bhagavadgita was included in the great epic. The same writer mentions that the Mahabharata was recited in the temple of Mahakala at Ujjain. That such recitation was already a widespread practice at that time is corroborated by an inscription of about 600 A.D. from the remote Indian colony of Kamboja, which states that copies of the Mahabharata, as well as of the Ramayana and of an unnamed Purana, were presented to a temple there, and that the donor had made arrangements to ensure their daily recitation in perpetuity. This evidence shows that the Mahabharata cannot have been a mere heroic poem, but must have borne the character of a Smriti work of long-established authority. Even at the present day both public and private recitations of the Epics and Puranas are common in India, and are always instituted for the edification and religious instruction of worshippers in temples or of members of the family. As a rule, the Sanskrit texts are not only declaimed, but also explained in the vernacular tongue for the benefit both of women, and of such males as belong to classes unacquainted with the learned language of the Brahmans.

We next come to the eminent Mimamsa philosopher Kumarila, who has been proved to have flourished in the first half of the eighth century A.D. In the small portion of his great commentary, entitled Tantra-varttika, which has been examined, no fewer than ten of the eighteen books of the Mahabharata are named, quoted, or referred to. It is clear that the epic as known to him not only included the first book (adiparvan), but that that book in his time closely resembled the form of its text which we possess. It even appears to have contained the first section, called anukramanika or "Survey of contents," and the second, entitled parva-samgraha or "Synopsis of sections." Kumarila also knew Books XII. and XIII., which have frequently been pronounced to be of late origin, as well as XIX. It is evident from his treatment of the epic that he regarded it as a work of sacred tradition and of great antiquity, intended from the beginning for the instruction of all the four castes. To him it is not an account of the great war between the Kauravas and Pandus; the descriptions of battles were only used for the purpose of rousing the martial instincts of the warrior caste.

The great Vedantist philosopher cankaracharya, who wrote his commentary in 804 A.D., often quotes the Mahabharata as a Smriti, and in discussing a verse from Book XII. expressly states that the Mahabharata was intended for the religious instruction of those classes who by their position are debarred from studying the Vedas and the Vedanta.

From the middle of the eleventh century A.D. we have the oldest known abstract of the Mahabharata, the work of the Kashmirian poet Kshemendra, entitled Bharata-Manjari. This condensation is specially important, because it enables the scholar to determine the state of the text in detail at that time. Professor Buhler's careful comparison of the MSS. of this work with the great epic has led him to the conclusion that Kshemendra's original did not differ from the Mahabharata as we have it at present in any other way than two classes of MSS. differ from each other. This poetical epitome shows several omissions, but these are on the whole of such a nature as is to be expected in any similar abridgment. It is, however, likely that twelve chapters (342-353) of Book XII., treating of Narayana, which the abbreviator passes over, did not exist in the original known to him. There can, moreover, be no doubt that the forms of several proper names found in the Manjari are better and older than those given by the editions of the Mahabharata. Though the division of the original into eighteen books is found in the abridgment also, it is made up by turning the third section (gada-parvan) of Book IX. (calya-parvan) into a separate book, while combining Books XII. and XIII. into a single one. This variation probably represents an old division, as it occurs in many MSS. of the Mahabharata.

Another work of importance in determining the state of the Mahabharata is a Javanese translation of the epic, also dating from the eleventh century.

The best-known commentator of the Mahabharata is Nilakantha, who lived at Kurpara, to the west of the Godavari, in Maharashtra, and, according to Burnell, belongs to the sixteenth century. Older than Nilakantha, who quotes him, is Arjuna Micra, whose commentary, along with that of Nilakantha, appears in an edition of the Mahabharata begun at Calcutta in 1875. The earliest extant commentator of the great epic is Sarvajna Narayana, large fragments of whose notes have been preserved, and who cannot have written later than in the second half of the fourteenth century, but may be somewhat older.

The main story of the Mahabharata in the briefest possible outline is as follows: In the country of the Bharatas, which, from the name of the ruling race, had come to be called Kurukshetra, or "Land of the Kurus," there lived at Hastinapura, fifty-seven miles north-east of the modern Delhi, two princes named Dhritarashtra and Pandu. The elder of these brothers being blind, Pandu succeeded to the throne and reigned gloriously. He had five sons called Pandavas, the chief of whom were Yudhishthira, Bhima, and Arjuna. Dhritarashtra had a hundred sons, usually called Kauravas, or Kuru princes, the most prominent of whom was Duryodhana. On the premature death of Pandu, Dhritarashtra took over the reins of government, and receiving his five nephews into his palace, had them brought up with his own sons. As the Pandus distinguished themselves greatly in feats of arms and helped him to victory, the king appointed his eldest nephew, Yudhishthira, to be heir-apparent. The Pandu princes, however, soon found it necessary to escape from the plots their cousins now began to set on foot against them. They made their way to the king of Panchala, whose daughter Draupadi was won, in a contest between many kings and heroes, by Arjuna, who alone was able to bend the king's great bow and to hit a certain mark. In order to avoid strife, Draupadi consented to become the common wife of the five princes. At Draupadi's svayamvara (public choice of a husband) the Pandus made acquaintance with Krishna, the hero of the Yadavas, who from this time onward became their fast friend and adviser. Dhritarashtra, thinking it best to conciliate the Pandavas in view of their double alliance with the Panchalas and Yadavas, now divided his kingdom, giving Hastinapura to his sons, and to his nephews a district where they built the city of Indraprastha, the modern Delhi (i.).

Here the Pandavas ruled wisely and prospered greatly. Duryodhana's jealousy being aroused, he resolved to ruin his cousins, with the aid of his uncle cakuni, a skilful gamester. Dhritarashtra was accordingly induced to invite the Pandus to Hastinapura. Here Yudhishthira, accepting the challenge to play at dice with Duryodhana, lost everything, his kingdom, his wealth, his army, his brothers, and finally Draupadi. In the end a compromise was made by which the Pandavas agreed to go into banishment for twelve years, and to remain incognito for a thirteenth, after which they might return and regain their kingdom (ii.).

With Draupadi they accordingly departed to the Kamyaka forest on the Sarasvati. The account of their twelve years' life here, and the many legends told to console them in their exile, constitute the vana-parvan or "Forest book," one of the longest in the poem (iii.).

The thirteenth year they spent in disguise as servants of Virata, king of the Matsyas. At this time the Kurus, in alliance with another king, invaded the country of the Matsyas, causing much distress. Then the Pandus arose, put the enemy to flight, and restored the king. They now made themselves known, and entered into an alliance with the king (iv.).