Part 1. On the Original Purāṇa-saṃhitā

The first place that one would look when seeking knowledge of cosmogony from Indian sources is the purāṇas. The purāṇas are traditionally supposed to teach five subjects, the first of which is creation or emanation (sarga). There are reckoned to be eighteen major purāṇas in the Hindu tradition, and some extend over multiple volumes. These sourcebooks of India’s creation stories are among the texts said by H. P. Blavatsky to be derived from the “Book of the Secret Wisdom of the World” that the “Book of Dzyan” is a secret commentary on: “the Purāṇas in India . . . are all derived from that one small parent volume” (The Secret Doctrine, vol. 1, p. xliii). In fact, there is a tradition given in the purāṇas themselves that they come from a single now lost source. This source is described as the original Purāṇa-saṃhitā. It consisted of 4,000 verses, less than in any of the eighteen purāṇas now extant, but not a small book. It would therefore have been an intermediate stage in the derivation “from that one small parent volume” described by Blavatsky, like the “Book of Dzyan” is also said to be.

The idea that all the purāṇas come from a single now lost source, an original Purāṇa-saṃhitā, was also arrived at by Western scholars, independently of Indian tradition. Ludo Rocher writes in his 1986 book, The Purāṇas (part of the series, A History of Indian Literature), p. 45: “The Western concept of a single, original purāṇa, from which all existing purāṇas ultimately derive their origin, resulted from a strict application of the rules of textual criticism, which were the backbone of European, especially German, classical philology. Scholars extended to purāṇas the same rules and principles they would have applied had they been editing Greek or Latin texts. Others, however, came to the same conclusion via a totally different route: the Indian tradition itself suggests that originally there was but one purāṇa.”

This one purāṇa is claimed by the purāṇas to be older than the vedas: “First, of all the scriptures the purāṇa was remembered by Brahmā; and afterwards, the vedas issued forth from his mouths” (Vāyu-purāṇa 1.1.54, Matsya-purāṇa 53.3, etc.). The Secret Doctrine also claims that its teachings are older than the Vedas: “For in the twentieth century of our era scholars will begin to recognize that the Secret Doctrine has neither been invented nor exaggerated, but, on the contrary, simply outlined; and finally, that its teachings antedate the Vedas” (The Secret Doctrine, vol. 1, p. xxxvii).

It would seem that the purāṇas follow what was described by Blavatsky as the “Book of the Secret Wisdom of the World” much more closely than do the other texts that are said to be derived from it. Blavatsky goes on to say there (The Secret Doctrine, vol. 1, p. xliii): “The old book, having described Cosmic Evolution and explained the origin of everything on earth, including physical man, after giving the true history of the races from the First down to the Fifth (our) race, goes no further. It stops short at the beginning of the Kali Yuga just 4989 years ago at the death of Krishna, . . .” Likewise, the purāṇas end their accounts, purporting to give history, at the beginning of the current kali-yuga. Blavatsky continues: “But there exists another book. None of its possessors regard it as very ancient, as it was born with, and is only as old as the Black Age, namely, about 5,000 years. In about nine years hence, the first cycle of the first five millenniums, that began with the great cycle of the Kali-Yuga, will end. And then the last prophecy contained in that book (the first volume of the prophetic record for the Black Age) will be accomplished.” Similarly, in seven of the purāṇas there is an added supplement on the dynasties of the kali-yuga, put in the form of prophecies. This account was carefully edited in Sanskrit by F. E. Pargiter and translated into English in his 1913 book, The Purāṇa Text of the Dynasties of the Kali Age.

Unlike the vedas, which had to be preserved unchanged, the purāṇas were expected to evolve and expand and be augmented (upabṛṃhaṇa) with new material. The five subjects that a purāṇa is traditionally supposed to teach are: (1) sarga, creation or emanation; (2) pratisarga, dissolution and re-creation; (3) vaṃśa, lineage or race, the genealogies or dynasties of kings, sages, and gods; (4) manvantara, the time period of a manu or humanity; (5) vaṃśānucarita, accounts of the individual kings, sages, and gods that comprise the genealogical listings. However, some of the purāṇas as we now have them include very little of these five subjects, and instead consist almost entirely of stories, praises of gods and goddesses, instructions for worship, descriptions or glorifications of sacred places, and various other subjects. As new material was added and old material was left out, the purāṇas evolved until in some cases there was almost nothing left in them of the one original Purāṇa-saṃhitā. How far is it possible to recover the original Purāṇa-saṃhitā from the extant purāṇas, we must wonder.

Pargiter’s in-depth work on the dynasties of the kali-yuga, the first ever critical edition of a purāṇa text, brought out some important facts. He established his text on the basis of the several printed editions then available plus sixty-three manuscripts. Of the seven purāṇas that have this account, he noted (op. cit., p. vi): “The versions of the Matsya, Vāyu, and Brahmānda present a remarkable similarity.” The Viṣṇu and Bhāgavata are condensations of this account, and the Garuḍa is a further condensation. The Bhaviṣya as we now have it “shows all the ancient matter utterly corrupted” (p. xxviii), even though the original Bhaviṣya is the source from which the Matsya, Vāyu, and Brahmāṇḍa say they took their account. Pargiter also found that (p. x): “There are clear indications that the Sanskrit account as it exists in the Matsya, Vāyu, and Brahmāṇḍa was originally in Prakrit, or, more accurately, that it is a Sanskritized version of older Prakrit ślokas. . . . The above conclusion holds good for the whole of the text of the Matsya, Vāyu, and Brahmānda; their verses are older Prakrit ślokas Sanskritized. It also holds good for such portions of the Viṣṇu and Bhāgavata as have preserved the old verses; but the main portions of these two Purāṇas are condensed redactions composed directly in Sanskrit.” So according to this, the Matsya, Vāyu, and Brahmāṇḍa are the oldest of the extant purāṇas.

Meanwhile, in 1910 S. P. L. Narasimhaswami had begun a comparative study of the purāṇas that would eventually lead to his reconstruction of the original Purāṇa-saṃhitā in 4,000 verses, unfortunately never published. He independently also concluded that the Matsya, Vāyu, and Brahmāṇḍa are the oldest of the extant purāṇas, and added to these the Harivaṃśa, a purāṇa-like supplement to the Mahābhārata. In his article, “Purana Samhita” (Journal of Sri Venkatesvara Oriental Institute, Tirupati, vol. 6, 1945, pp. 54-71, attached), he writes (p. 59): “Keeping these facts in mind, I began to examine the ślokas which were repeated in different Purāṇas. Staunch sectarian Purāṇas, like Padma, Kūrma, Liṅga, etc. do not contain these stanzas. Those like Vishṇu, Mārkaṇḍeya, etc. contain very few of them. Matsya and Harivaṃśa (although the latter is not a Purāṇa) contain hundreds of stanzas in common with Vāyu and Brahmāṇḍa, while these last Purāṇas have thousands of stanzas in common though they are not in a continuous line.” After preparing a parallel text of the account of the Yādava dynasty in the Brahmāṇḍa, Vāyu, Matsya, and Harivaṃśa, he concluded: “When I made sufficient progress in the formation of the parallel text, I was convinced that the common portion was the Purāṇa-saṃhitā.”

Of these texts, we see that the Vāyu and Brahmāṇḍa purāṇas have thousands of these old verses in common. As now extant, the Vāyu Purāṇa has 10,714 verses in the Bibliotheca Indica edition, or 10,991 verses in the Ānandāśrama edition, while the Brahmāṇḍa Purāṇa has 14,286 verses in the Veṅkaṭeśvara edition (the only one published). According to Narasimhaswami (ibid.), they have 7,557 verses in common, and there are two lacunae in the Brahmāṇḍa Purāṇa that would add 826 verses to this. So the total of 8,383 verses would have to be reduced by about half to get to the 4,000 verse extent of the original Purāṇa-saṃhitā. Because the extant Vāyu Purāṇa is shorter than the extant Brahmāṇḍa Purāṇa, most researchers regard the Vāyu Purāṇa as being the oldest purāṇa we have, and the closest to the original Purāṇa-saṃhitā.

S. P. L. Narasimhaswami concluded that the Vāyu Purāṇa is the oldest purāṇa in another statement, naming additional purāṇas, in his only other published article that I know of, “Aikṣvāku Dynasty” (Bhāratīya Vidyā, vol. 4, 1943, pp. 217-220, attached), where he writes (p. 219): “In the light of the Purāṇa-saṃhitā, the value of the different Purāṇas has to be assessed differently. Some Purāṇas, like the Agni, the Garuḍa, the Bhaviṣya and the Brahmavaivarta, have no historical matter in them and are only Purāṇas in name. . . . Others like the Viṣṇu, the Bhāgavata, the Mārkaṇḍeya, and the Vāmana are cognizant of the Saṃhitā and incorporate it partly in them. The rest which are very old, like the Vāyu, the Brahmāṇḍa, and the Matsya contain the Saṃhitā in them, either wholly or partially. It is these Purāṇas that helped me in the task of recovering the Saṃhitā. Of these the Vāyu-purāṇa is the oldest and most valuable.”

Despite regarding the account of the dynasties of the kali-yuga found in the Matsya Purāṇa as slightly older in his 1913 book (p. xiv), F. E. Pargiter had come to the conclusion that the Vāyu and Brahmāṇḍa were the oldest purāṇas we have in his 1922 book, Ancient Indian Historical Tradition (p. 78): “These two appear to be the oldest of the Puranas that we possess now, and are on the whole the most valuable in all matters of traditional history.” He had then come to regard them as originally one purāṇa (p. 77): “The Vāyu and Brahmāṇḍa have the best text of the genealogies. Their accounts agree closely, so that they are really only two versions of the same text. They have a great part of their contents in common, generally almost verbatim, and it appears they were originally one Purana.” That they were originally one, incidentally, is also the conclusion that I had reached before seeing his book, and for the very same reason that he there gives (pp. 77-78). This is as follows:

The lists of the eighteen purāṇas given in the majority of the purāṇas omit the Vāyu Purāṇa. In a minority of the lists, the Vāyu Purāṇa is given in place of the Śiva Purāṇa. But both of these are major purāṇas, and we cannot have nineteen. Pargiter notes that only two of the lists have both the Vāyu and the Brahmāṇḍa, and one of these two lists is from the Vāyu itself as we now have it (the other is from the Garuḍa). The obvious implication is that the Vāyu was not separate from the Brahmāṇḍa until quite late. They are the same purāṇa. The majority of the lists, which omit the Vāyu, are correct, since the Vāyu is there as the Brahmāṇḍa.

To demonstrate that the two are one, the close parallel contents of the extant Vāyu and Brahmāṇḍa purāṇas were laid out in detail in a chart prepared by Willibald Kirfel. He did this at the beginning of his introduction to his major 1927 study, Das Purāṇa Pañcalakṣaṇa. In this 598-page book, Kirfel gathered together from the various purāṇas all the passages on the five subjects that a purāṇa is traditionally supposed to teach, the purāṇa-pañca-lakṣaṇa, the “five defining characteristics of a purāṇa.” So the book is entirely in Sanskrit. It is prefaced by a 40-page introduction in German. Kirfel’s German introduction was translated into English by P. V. Ramanujasvami, at the request of his brother, S. P. L. Narasimhaswami, and published in Journal of Sri Venkatesvara Oriental Institute, Tirupati, vol. 7, 1946, pp. 81-101; vol. 8, 1947, pp. 9-33 (attached as Purāṇa Pañcalakṣaṇa Introduction). Ludo Rocher notes that this translation “should be used with extreme caution” (The Purāṇas, p. 44, fn. 12). Nonetheless, it affords us some access to Kirfel’s German in English. About the Brahmāṇḍa and Vāyu purāṇas, Kirfel writes (English translation, p. 83): “The first result of the Purāṇic text-comparison is the perception that the Bḍ. [Brahmāṇḍa] and Vā. [Vāyu] must have originally formed a single Purāṇa.”

However, Kirfel did not regard the Brahmāṇḍa/Vāyu Purāṇa as the oldest, as did Pargiter and Narasimhaswami. Kirfel’s approach was to gather together the passages from the various purāṇas on each of the five subjects of a purāṇa (although he took the first two closely related subjects together, sarga and pratisarga, creation and dissolution), then to place them into text groups having matching accounts, and lastly to arrange these text groups as much as possible into what he regarded as their chronological order. Thus, on the subject of creation or emanation and dissolution followed by re-creation, his first text group consists of the Brahma Purāṇa, the Harivaṃśa, and the Śiva Purāṇa, with partial support from the Agni Purāṇa. His second text group was divided into two sub-groups. Group 2A consists of the Padma Purāṇa and the Viṣṇu Purāṇa, with a little support from the Garuḍa Purāṇa. Group 2B consists of the Brahmāṇḍa Purāṇa, the Vāyu Purāṇa, the Kūrma Purāṇa, the Liṅga Purāṇa, and the Mārkaṇḍeya Purāṇa. His third text group consists solely of the Matsya Purāṇa. (He did not use the Nārada, Brahma-vaivarta, Skanda, or Vāmana purāṇas in his book, and for this subject he did not find or give anything from the Bhāgavata, Bhaviṣya, or Varāha purāṇas.) As may be seen from this, he regarded the account of creation and dissolution from the four purāṇas in the first text group to be older than that from the Brahmāṇḍa/Vāyu Purāṇa, found in his text group 2B.

Part of Kirfel’s reasoning for this is that the account from the first text group is much briefer, and hence presumably less expanded. By contrast, on the subject of the dynasties of the kali-yuga, Pargiter saw the briefer accounts in the Viṣṇu, Bhāgavata, and Garuḍa purāṇas as condensations of the accounts in the Vāyu, Brahmāṇḍa, and Matsya purāṇas. Kirfel also used the criterion of whether the accounts contained Sāṃkhya ideas. This is based on the assumption that Sāṃkhya philosophy is a later development, and thus was added to the purāṇas later. By contrast, Indian tradition regards Sāṃkhya as the oldest philosophy, so that it would naturally be in the purāṇas from early on. Narasimhaswami disregarded both of these criteria used by Kirfel, and focused instead on parallel old verses. Kirfel perceived the hand of a reviser in the Brahmāṇḍa/Vāyu verses by comparing similar material from other purāṇas. But how do we know which direction this revising went in? What was convincing evidence to Kirfel was not convincing to others. Of course, the usefulness of Kirfel’s book is not dependent on accepting his chronological views. The value of his compilation for comparing the accounts of the various purāṇas on the five subjects is very great indeed. He concluded (English translation, pp. 28-29): “Apart from the abridgement in A. [Agni] and Ga. [Garuḍa] as well as the prose paraphrase of Vi. [Viṣṇu], we find in the Purāṇas only three complete compositions of this text [the pañca-lakṣaṇa], namely that of the Br. [Brahma] and H. [Harivaṃśa], that of the Bḍ.-Vā. [Brahmāṇḍa-Vāyu] and that of the Mt. [Matsya]; all others contain only smaller or greater parts of the same.” He, too, was trying to ascertain the contents of an original or “Ur-purāṇa.”

The “Original Purāṇa Saṃhitā,” by V. S. Agrawala (Purāṇa, vol. 8, 1966, pp. 232-245, attached), summarizes the information we have on this, and accepts the extant Vāyu Purāṇa as the oldest and closest to the original Purāṇa-saṃhitā. As we have seen, the extant Vāyu Purāṇa has about 11,000 verses, while the original Purāṇa-saṃhitā is reported to have had 4,000 verses. So Agrawala here (pp. 242-244) provides a listing of what portions of the extant Vāyu Purāṇa making up about 7,000 verses should be discarded, and what portions making up about 4,000 verses should be retained as constituting the original Purāṇa-saṃhitā. This may be compared to Narasimhaswami’s detailed listing of what chapters, and how many verses in each, made up the original Purāṇa-saṃhitā that he reconstructed (“Purana Samhita,” pp. 63-69). While Narasimhaswami and Pargiter were interested in recovering history from the original Purāṇa-saṃhitā, Agrawala was interested in recovering the ancient knowledge called Purāṇa-Vidyā.

In the Preface to his 1963 book, Matsya Purāṇa—A Study (An Exposition of the Ancient Purāṇa-Vidyā), Agrawala explains (p. ix): “Purāṇa-Vidyā—The point of view which has inspired the present study of the Matsya Purāṇa is an investigation not of chronology or of canons of authorship but of the real secrets of what once was known as the Purāṇa-Vidyā. Like other Vidyās as Vyākaraṇa [grammar], Jyotisha [astronomy/astrology], Nirukta [etymology] etc., Purāṇa also was a subject of intensive purposive study in which serious teachers and pupils were engaged. What that purpose was is often stated in the Purāṇas themselves. The objective was to present, amplify and preserve the meaning of the Vedic Sṛishṭi-Vidyā or the science of cosmogony.” The ancient Purāṇa-Vidyā is apparently the key that Blavatsky refers to in this statement from The Secret Doctrine (vol. 1, p. 423): “But there was a time when the Puranas were esoteric works, and so they are still for the Initiates who can read them with the key that is in their possession.”

Pargiter had found that, for the account of the dynasties of the kali-yuga he edited, the verses from the Matsya, Vāyu, and Brahmāṇḍa purāṇas were originally in Prakrit, being Sanskritized versions of older Prakrit ślokas. At that time, the so-called “Buddhist Hybrid Sanskrit” had not yet been identified or studied. In Buddhist texts, old verses are found that use Prakrit-type words and inflections, words and inflections that could not be changed into classical Sanskrit without spoiling the meter. Even a few old prose texts were found written in this dialect, dubbed “Buddhist Hybrid Sanskrit” by Franklin Edgerton, who published a grammar and dictionary of it in 1953. We can now see that these old purāṇa verses in Sanskritized Prakrit are like the “Buddhist Hybrid Sanskrit” verses in Buddhist texts, where the process of changing them into classical Sanskrit is more visible. While it is possible to regard these old verses as going back to a vernacular Prakrit form of these early writings, it is also possible to regard them as remnants of an older pre-classical form of sacred Sanskrit, closer to the esoteric Senzar. Senzar is the name given to the language of “that one small parent volume” from which the purāṇas are said to be derived.

In summary, Indian tradition speaks of an original Purāṇa-saṃhitā, no longer available, that is the source of the eighteen purāṇas now known. The idea that the purāṇas come from a single now lost source was arrived at independently by Western scholars through their own researches. The idea that the purāṇas come from a single now lost source was also stated by H. P. Blavatsky in The Secret Doctrine, and this source is the book that the “Book of Dzyan” is a commentary on. This source is said to describe cosmic evolution up to the beginning of the present kali-yuga. The purāṇas also describe cosmic evolution and end with the beginning of the present kali-yuga. The Secret Doctrine speaks of another book that gives the prophecies of the kali-yuga. Seven of the purāṇas also have a supplement that gives in the form of prophecies an account of the dynasties of the kali-yuga. The original Purāṇa-saṃhitā is said to consist of 4,000 verses. This would be an intermediate text between the “one small parent volume” and the eighteen known purāṇas. Attempts to recover the original Purāṇa-saṃhitā indicate that many, if not most, of its 4,000 verses may be found in the extant Vāyu Purāṇa and its twin Brahmāṇḍa Purāṇa, supplemented by the Matsya Purāṇa and the Harivaṃśa. Research showed that these verses were Sanskritized from an earlier language, a language that may have been intermediate between Senzar and classical Sanskrit. Attempts have also been made to recover the ancient knowledge called Purāṇa-Vidyā, which would provide the key to the meaning of the original Purāṇa-saṃhitā.