In the dim twilight preceding the dawn of Indian literature the historical imagination can perceive the forms of Aryan warriors, the first Western conquerors of Hindustan, issuing from those passes in the north-west through which the tide of invasion has in successive ages rolled to sweep over the plains of India. The earliest poetry of this invading race, whose language and culture ultimately overspread the whole continent, was composed while its tribes still occupied the territories on both sides of the Indus now known as Eastern Kabulistan and the Panjab. That ancient poetry has come down to us in the form of a collection of hymns called the Rigveda. The cause which gathered the poems it contains into a single book was not practical, as in the case of the Sama- and Yajur-veda, but scientific and historical. For its ancient editors were undoubtedly impelled by the motive of guarding this heritage of olden time from change and destruction. The number of hymns comprised in the Rigveda, in the only recension which has been preserved, that of the Çakala school, is 1017, or, if the eleven supplementary hymns (called Valakhilya) which are inserted in the middle of the eighth book are added, 1028. These hymns are grouped in ten books, called ma??alas, or “cycles,” which vary in length, except that the tenth contains the same number of hymns as the first. In bulk the hymns of the Rigveda equal, it has been calculated, the surviving poems of Homer.

The general character of the ten books is not identical in all cases. Six of them (ii.–vii.) are homogeneous. Each of these, in the first place, is the work of a different seer or his descendants according to the ancient tradition, which is borne out by internal evidence. They were doubtless long handed down separately in the families to which they owed their being. Moreover, the hymns contained in these “family books,” as they are usually called, are arranged on a uniform plan differing from that of the rest. The first, eighth, and tenth books are not the productions of a single family of seers respectively, but consist of a number of groups based on identity of authorship. The arrangement of the ninth book is in no way connected with its composers; its unity is due to all its hymns being addressed to the single deity Soma, while its groups depend on identity of metre. The family books also contain groups; but each of these is formed of hymns addressed to one and the same deity.

Turning to the principle on which the entire books of the Rigveda are arranged in relation to one another, we find that Books II.–VII., if allowance is made for later additions, form a series of collections which contain a successively increasing number of hymns. This fact, combined with the uniformity of these books in general character and internal arrangement, renders it probable that they formed the nucleus of the Rigveda, to which the remaining books were successively added. It further seems likely that the nine shorter collections, which form the second part of Book I., as being similarly based on identity of authorship, were subsequently combined and prefixed to the family books, which served as the model for their internal arrangement.

The hymns of the eighth book in general show a mutual affinity hardly less pronounced than that to be found in the family books. For they are connected by numerous repetitions of similar phrases and lines running through the whole book. The latter, however, does not form a parallel to the family books. For though a single family, that of the Ka?vas, at least predominates among its authors, the prevalence in it of the strophic form of composition impresses upon it a character of its own. Moreover, the fact that the eighth book contains fewer hymns than the seventh, in itself shows that the former did not constitute one of the family series.

The first part (1–50) of Book I. has considerable affinities with the eighth, more than half its hymns being attributed to members of the Ka?va family, while in the hymns composed by some of these Ka?vas the favourite strophic metre of the eighth book reappears. There are, moreover, numerous parallel and directly identical passages in the two collections. It is, however, at present impossible to decide which of the two is the earlier, or why it is that, though so nearly related, they should have been separated. Certain it is that they were respectively added at the beginning and the end of a previously existing collection, whether they were divided for chronological reasons or because composed by different branches of the Ka?va family.

As to the ninth book, it cannot be doubted that it came into being as a collection after the first eight books had been combined into a whole. Its formation was in fact the direct result of that combination. The hymns to Soma Pavamana (“the clearly flowing”) are composed by authors of the same families as produced Books II.–VII., a fact, apart from other evidence, sufficiently indicated by their having the characteristic refrains of those families. The Pavamana hymns have affinities to the first and eighth books also. When the hymns of the different families were combined into books, and clearly not till then, all their Pavamana hymns were taken out and gathered into a single collection. This of course does not imply that the Pavamana hymns themselves were of recent origin. On the contrary, though some of them may date from the time when the tenth book came into existence, there is good reason to suppose that the poetry of the Soma hymns, which has many points in common with the Avesta, and deals with a ritual going back to the Indo-Iranian period, reached its conclusion as a whole in early times among the Vedic singers. Differences of age in the hymns of the ninth book have been almost entirely effaced; at any rate, research has as yet hardly succeeded in distinguishing chronological stages in this collection.

With regard to the tenth book, there can be no doubt that its hymns came into being at a time when the first nine already existed. Its composers grew up in the knowledge of the older books, with which they betray their familiarity at every turn. The fact that the author of one of its groups (20–26) begins with the opening words (agnim il?e) of the first stanza of the Rigveda, is probably an indication that Books I.–IX. already existed in his day even as a combined collection. That the tenth book is indeed an aggregate of supplementary hymns is shown by its position after the Soma book, and by the number of its hymns being made up to that of the first book (191). The unity which connects its poetry is chronological; for it is the book of recent groups and recent single hymns. Nevertheless the supplements collected in it appear for the most part to be older than the additions which occur in the earlier books.

There are many criteria, derived from its matter as well as its form, showing the recent origin of the tenth book. With regard to mythology, we find the earlier gods beginning to lose their hold on the imagination of these later singers. Some of them seem to be disappearing, like the goddess of Dawn, while only deities of widely established popularity, such as Indra and Agni, maintain their position. The comprehensive group of the ViÇve devas, or “All gods,” has alone increased in prominence. On the other hand, an altogether new type, the deification of purely abstract ideas, such as “Wrath” and “Faith,” now appears for the first time. Here, too, a number of hymns are found dealing with subjects foreign to the earlier books, such as cosmogony and philosophical speculation, wedding and burial rites, spells and incantations, which give to this book a distinctive character besides indicating its recent origin.

Linguistically, also, the tenth book is clearly distinguished as later than the other books, forming in many respects a transition to the other Vedas. A few examples will here suffice to show this. Vowel contractions occur much more frequently, while the hiatus has grown rarer. The use of the letter l, as compared with r, is, in agreement with later Sanskrit, strikingly on the increase. In inflexion the employment of the Vedic nominative plural in asas is on the decline. With regard to the vocabulary, many old words are going out of use, while others are becoming commoner. Thus the particle sim, occurring fifty times in the rest of the Rigveda, is found only once in the tenth book. A number of words common in the later language are only to be met with in this book; for instance, labh, “to take,” kala, “time,” lakshmi, “fortune,” evam, “thus.” Here, too, a number of conscious archaisms can be pointed out.

Thus the tenth book represents a definitely later stratum of composition in the Rigveda. Individual hymns in the earlier books have also been proved by various recognised criteria to be of later origin than others, and some advance has been made towards assigning them to three or even five literary epochs. Research has, however, not yet arrived at any certain results as to the age of whole groups in the earlier books. For it must be borne in mind that posteriority of collection and incorporation does not necessarily prove a later date of composition.

Some hundreds of years must have been needed for all the hymns found in the Rigveda to come into being. There was also, doubtless, after the separation of the Indians from the Iranians, an intermediate period, though it was probably of no great length. In this transitional age must have been composed the more ancient poems which are lost, and in which the style of the earliest preserved hymns, already composed with much skill, was developed. The poets of the older part of the Rigveda themselves mention predecessors, in whose wise they sing, whose songs they desire to renew, and speak of ancestral hymns produced in days of yore. As far as linguistic evidence is concerned, it affords little help in discriminating periods within the Rigveda except with regard to the tenth book. For throughout the hymns, in spite of the number of authors, essentially the same language prevails. It is quite possible to distinguish differences of thought, style, and poetical ability, but hardly any differences of dialect. Nevertheless, patient and minute linguistic research, combined with the indications derived from arrangement, metre, and subject-matter, is beginning to yield evidence which may lead to the recognition of chronological strata in the older books of the Rigveda.

Though the aid of MSS. for this early period entirely fails, we yet happily possess for the Rigveda an abundant mass of various readings over 2000 years old. These are contained in the other Vedas, which are largely composed of hymns, stanzas, and lines borrowed from the Rigveda. The other Vedas are, in fact, for the criticism of the Rigveda, what manuscripts are for other literary monuments. We are thus enabled to collate with the text of the Rigveda directly handed down, various readings considerably older than even the testimony of Yaska and of the PratiÇakhyas.

The comparison of the various readings supplied by the later Vedas leads to the conclusion that the text of the Rigveda existed, with comparatively few exceptions, in its present form, and not in a possibly different recension, at the time when the text of the Sama-veda, the oldest form of the Yajur-veda, and the Atharva-veda was constituted. The number of cases is infinitesimal in which the Rigveda shows a corruption from which the others are free. Thus it appears that the kernel of Vedic tradition, as represented by the Rigveda, has come down to us, with a high degree of fixity and remarkable care for verbal integrity, from a period which can hardly be less remote than 1000 B.C.

It is only natural that a sacred collection of poetry, historical in its origin, and the heritage of oral tradition before the other Vedas were composed and the details of the later ritual practice were fixed, should have continued to be preserved more accurately than texts formed mainly by borrowing from it hymns which were arbitrarily cut up into groups of verses or into single verses, solely in order to meet new liturgical needs. For those who removed verses of the Rigveda from their context and mixed them up with their own new creations would not feel bound to guard such verses from change as strictly as those who did nothing but continue to hand down, without any break, the ancient text in its connected form. The control of tradition would be wanting where quite a new tradition was being formed.

The criticism of the text of the Rigveda itself is concerned with two periods. The first is that in which it existed alone before the other Vedas came into being; the second is that in which it appears in the phonetically modified form called the Sam?hita text, due to the labours of grammatical editors. Being handed down in the older period exclusively by oral tradition, it was not preserved in quite authentic form down to the time of its final redaction. It did not entirely escape the fate suffered by all works which, coming down from remote antiquity, survive into an age of changed linguistic conditions. Though there are undeniable corruptions in detail belonging to the older period, the text maintained a remarkably high level of authenticity till such modifications as it had undergone reached their conclusion in the Sam?hita text. This text differs in hundreds of places from that of the composers of the hymns; but its actual words are nearly always the same as those used by the ancient seers. Thus there would be no uncertainty as to whether the right word, for instance, was sumnam or dyumnam. The difference lies almost entirely in the phonetic changes which the words have undergone according to the rules of Sandhi prevailing in the classical language. Thus what was formerly pronounced as tuam? hi agne now appears as tvam? hy agne. The modernisation of the text thereby produced is, however, only partial, and is often inconsistently applied. The euphonic combinations introduced in the Sam?hita text have interfered with the metre. Hence by reading according to the latter the older text can be restored. At the same time the Sam?hita text has preserved the smallest minutiÆ of detail most liable to corruption, and the slightest difference in the matter of accent and alternative forms, which might have been removed with the greatest ease. Such points furnish an additional proof that the extreme care with which the verbal integrity of the text was guarded goes back to the earlier period itself. Excepting single mistakes of tradition in the first, and those due to grammatical theories in the second period, the old text of the Rigveda thus shows itself to have been preserved from a very remote antiquity with marvellous accuracy even in the smallest details.

From the explanatory discussions of the Brahma?as in connection with the Rigveda, it results that the text of the latter must have been essentially fixed in their time, and that too in quite a special manner, more, for instance, than the prose formulas of the Yajurveda. For the Çatapatha Brahma?a, while speaking of the possibility of varying some of these formulas, rejects the notion of changing the text of a certain Rigvedic verse, proposed by some teachers, as something not to be thought of. The Brahma?as further often mention the fact that such and such a hymn or liturgical group contains a particular number of verses. All such numerical statements appear to agree with the extant text of the Rigveda. On the other hand, transpositions and omissions of Rigvedic verses are to be found in the Brahma?as. These, however, are only connected with the ritual form of those verses, and in no way show that the text from which they were taken was different from ours.

The Sutras also contain altered forms of Rigvedic verses, but these are, as in the case of the Brahma?as, to be explained not from an older recension of the text, but from the necessity of adapting them to new ritual technicalities. On the other hand, they contain many statements which confirm our present text. Thus all that the Sutra of Çankhayana says about the position occupied by verses in a hymn, or the total number of verses contained in groups of hymns, appears invariably to agree with our text.

We have yet to answer the question as to when the Sam?hita text, which finally fixed the canonical form of the Rigveda, was constituted. Now the Brahma?as contain a number of direct statements as to the number of syllables in a word or a group of words, which are at variance with the Sam?hita text owing to the vowel contractions made in the latter. Moreover, the old part of the Brahma?a literature shows hardly any traces of speculations about phonetic questions connected with the Vedic text. The conclusion may therefore be drawn that the Sam?hita text did not come into existence till after the completion of the Brahma?as. With regard to the Ara?yakas and Upanishads, which form supplements to the Brahma?as, the case is different. These works not only mention technical grammatical terms for certain groups of letters, but contain detailed doctrines about the phonetic treatment of the Vedic text. Here, too, occur for the first time the names of certain theological grammarians, headed by Çakalya and Ma??ukeya, who are also recognised as authorities in the PratiÇakhyas. The Ara?yakas and Upanishads accordingly form a transition, with reference to the treatment of grammatical questions, between the age of the Brahma?as and that of Yaska and the PratiÇakhyas. The Sam?hita text must have been created in this intermediate period, say about 600 B.C.

This work being completed, extraordinary precautions soon began to be taken to guard the canonical text thus fixed against the possibility of any change or loss. The result has been its preservation with a faithfulness unique in literary history. The first step taken in this direction was the constitution of the Pada, or “word” text, which being an analysis of the Sam?hita, gives each separate word in its independent form, and thus to a considerable extent restores the Sam?hita text to an older stage. That the Pada text was not quite contemporaneous in origin with the other is shown by its containing some undoubted misinterpretations and misunderstandings. Its composition can, however, only be separated by a short interval from that of the Sam?hita, for it appears to have been known to the writer of the Aitareya Ara?yaka, while its author, Çakalya, is older than both Yaska, who quotes him, and Çaunaka, composer of the Rigveda PratiÇakhya, which is based on the Pada text.

The importance of the latter as a criterion of the authenticity of verses in the Rigveda is indicated by the following fact. There are six verses in the Rigveda1 not analysed in the Pada text, but only given there over again in the Sam?hita form. This shows that Çakalya did not acknowledge them as truly Rigvedic, a view justified by internal evidence. This group of six, which is doubtless exhaustive, stands midway between old additions which Çakalya recognised as canonical, and the new appendages called Khilas, which never gained admission into the Pada text in any form.

A further measure for preserving the sacred text from alteration with still greater certainty was soon taken in the form of the Krama-pa?ha, or “step-text.” This is old, for it, like the Pada-pa?ha, is already known to the author of the Aitareya Ara?yaka. Here every word of the Pada text occurs twice, being connected both with that which precedes and that which follows. Thus the first four words, if represented by a, b, c, d, would be read as ab, bc, cd. The Ja?a-pa?ha, or “woven-text,” in its turn based on the Krama-pa?ha, states each of its combinations three times, the second time in reversed order (ab, ba, ab; bc, cb, bc). The climax of complication is reached in the Ghana-pa?ha, in which the order is ab, ba, abc, cba, abc; bc, cb, bcd, &c.

The PratiÇakhyas may also be regarded as safeguards of the text, having been composed for the purpose of exhibiting exactly all the changes necessary for turning the Pada into the Sam?hita text.

Finally, the class of supplementary works called Anukrama?is, or “Indices” aimed at preserving the Rigveda intact by registering its contents from various points of view, besides furnishing calculations of the number of hymns, verses, words, and even syllables, contained in the sacred book.

The text of the Rigveda has come down to us in a single recension only; but is there any evidence that other recensions of it existed in former times?

The Chara?a-vyuha, or “Exposition of Schools,” a supplementary work of the Sutra period, mentions as the five Çakhas or “branches” of the Rigveda, the Çakalas, the Vashkalas, the AÇvalayanas, the Çankhayanas, and the Ma??ukeyas. The third and fourth of these schools, however, do not represent different recensions of the text, the sole distinction between them and the Çakalas having been that the AÇvalayanas recognised as canonical the group of the eleven Valakhilya or supplementary hymns, and the Çankhayanas admitted the same group, diminished only by a few verses. Hence the tradition of the Pura?as, or later legendary works, mentions only the three schools of Çakalas, Vashkalas, and Ma??ukas. If the latter ever possessed a recension of an independent character, all traces of it were lost at an early period in ancient India, for no information of any kind about it has been preserved. Thus only the two schools of the Çakalas and the Vashkalas come into consideration. The subsidiary Vedic writings contain sufficient evidence to show that the text of the Vashkalas differed from that of the Çakalas only in admitting eight additional hymns, and in assigning another position to a group of the first book. But in these respects it compares unfavourably with the extant text. Thus it is evident that the Çakalas not only possessed the best tradition of the text of the Rigveda, but handed down the only recension, in the true sense, which, as far as we can tell, ever existed.

The text of the Rigveda, like that of the other Sam?hitas, as well as of two of the Brahma?as (the Çatapatha and the Taittiriya, together with its Ara?yaka), has come down to us in an accented form. The peculiarly sacred character of the text rendered the accent very important for correct and efficacious recitation. Analogously the accent was marked by the Greeks in learned and model editions only. The nature of the Vedic accent was musical, depending on the pitch of the voice, like that of the ancient Greeks. This remained the character of the Sanskrit accent till later than the time of Pa?ini. But just as the old Greek musical accent, after the beginning of our era, was transformed into a stress accent, so by the seventh century A.D. (and probably long before) the Sanskrit accent had undergone a similar change. While, however, in modern Greek the stress accent has remained, owing to the high pitch of the old acute, on the same syllable as bore the musical accent in the ancient language, the modern pronunciation of Sanskrit has no connection with the Vedic accent, but is dependent on the quantity of the last two or three syllables, much the same as in Latin. Thus the penultimate, if long, is accented, e.g. Kalidasa, or the antepenultimate, if long and followed by a short syllable, e.g. brahma?a or Himalaya (“abode of snow”). This change of accent in Sanskrit was brought about by the influence of Prakrit, in which, as there is evidence to show, the stress accent is very old, going back several centuries before the beginning of our era.

There are three accents in the Rigveda as well as the other sacred texts. The most important of these is the rising accent, called ud-atta (“raised”), which corresponds to the Greek acute. Comparative philology shows that in Sanskrit it rests on the same syllable as bore it in the proto-Aryan language. In Greek it is generally on the same syllable as in Sanskrit, except when interfered with by the specifically Greek law restricting the accent to one of the last three syllables. Thus the Greek heptÁ corresponds to the Vedic saptÁ, “seven.” The low-pitch accent, which precedes the acute, is called the anudatta (“not raised”). The third is the falling accent, which usually follows the acute, and is called svarita (“sounded”).

Of the four different systems of marking the accent in Vedic texts, that of the Rigveda is most commonly employed. Here the acute is not marked at all, while the low-pitch anudatta is indicated by a horizontal stroke below the syllable bearing it, and the svarita by a vertical stroke above. Thus yajnasyÀ (“of sacrifice”) would mean that the second syllable has the acute and the third the svarita (yajnÁsyÀ). The reason why the acute is not marked is because it is regarded as the middle tone between the other two.2

The hymns of the Rigveda consist of stanzas ranging in number from three to fifty-eight, but usually not exceeding ten or twelve. These stanzas (often loosely called verses) are composed in some fifteen different metres, only seven of which, however, are at all frequent. Three of them are by far the commonest, claiming together about four-fifths of the total number of stanzas in the Rigveda.

There is an essential difference between Greek and Vedic prosody. Whereas the metrical unit of the former system is the foot, in the latter it is the line (or verse), feet not being distinguished. Curiously enough, however, the Vedic metrical unit is also called pada, or “foot,” but for a very different reason; for the word has here really the figurative sense of “quarter” (from the foot of a quadruped), Because the most usual kind of stanza has four lines. The ordinary padas consist of eight, eleven, or twelve syllables. A stanza or ?ich is generally formed of three or four lines of the same kind. Four or five of the rarer types of stanza are, however, made up of a combination of different lines.

It is to be noted that the Vedic metres have a certain elasticity to which we are unaccustomed in Greek prosody, and which recalls the irregularities of the Latin Saturnian verse. Only the rhythm of the last four or five syllables is determined, the first part of the line not being subject to rule. Regarded in their historical connection, the Vedic metres, which are the foundation of the entire prosody of the later literature, occupy a position midway between the system of the Indo-Iranian period and that of classical Sanskrit. For the evidence of the Avesta, with its eight and eleven syllable lines, which ignore quantity, but are combined into stanzas otherwise the same as those of the Rigveda, indicates that the metrical practice of the period when Persians and Indians were still one people, depended on no other principle than the counting of syllables. In the Sanskrit period, on the other hand, the quantity of every syllable in the line was determined in all metres, with the sole exception of the loose measure (called Çloka) employed in epic poetry. The metrical regulation of the line, starting from its end, thus finally extended to the whole. The fixed rhythm at the end of the Vedic line is called v?itta, literally “turn” (from v?it, Lat. vert-ere), which corresponds etymologically to the Latin versus.

The eight-syllable line usually ends in two iambics, the first four syllables, though not exactly determined, having a tendency to be iambic also. This verse is therefore the almost exact equivalent of the Greek iambic dimeter.

Three of these lines combine to form the gayatri metre, in which nearly one-fourth (2450) of the total number of stanzas in the Rigveda is composed. An example of it is the first stanza of the Rigveda, which runs as follows:—

It may be closely rendered thus in lines imitating the rhythm of the original:—

Four of these eight-syllable lines combine to form the anush?ubh stanza, in which the first two and the last two are more closely connected. In the Rigveda the number of stanzas in this measure amounts to only about one-third of those in the gayatri. This relation is gradually reversed, till we reach the post-Vedic period, when the gayatri is found to have disappeared, and the anush?ubh (now generally called Çloka) to have become the predominant measure of Sanskrit poetry. A development in the character of this metre may be observed within the Rigveda itself. All its verses in the oldest hymns are the same, being iambic in rhythm. In later hymns, however, a tendency to differentiate the first and third from the second and fourth lines, by making the former non-iambic, begins to show itself. Finally, in the latest hymns of the tenth book the prevalence of the iambic rhythm disappears in the odd lines. Here every possible combination of quantity in the last four syllables is found, but the commonest variation, nearly equalling the iambic in frequency, is [short][long][long][shortlong]. The latter is the regular ending of the first and third line in the post-Vedic Çloka.

The twelve-syllable line ends thus: [long][short][long][short][short]. Four of these together form the jagati stanza. The trish?ubh stanza consists of four lines of eleven syllables, which are practically catalectic jagatis, as they end [long][short][long][shortlong]. These two verses being so closely allied and having the same cadence, are often found mixed in the same stanza. The trish?ubh is by far the commonest metre, about two-fifths of the Rigveda being composed in it.

Speaking generally, a hymn of the Rigveda consists entirely of stanzas in the same metre. The regular and typical deviation from this rule is to conclude a hymn with a single stanza in a metre different from that of the rest, this being a natural method of distinctly marking its close.

A certain number of hymns of the Rigveda consist not merely of a succession of single stanzas, but of equal groups of stanzas. The group consists either of three stanzas in the same simple metre, generally gayatri, or of the combination of two stanzas in different mixed metres. The latter strophic type goes by the name of Pragatha, and is found chiefly in the eighth book of the Rigveda.