When Kurus Fought Pandavas.

B y H a r it K r is h n a D e b .

That the Great War between the Kurus and the Pandavas which forms the central theme o£ the Hindu Epic, the MahS- bharata, was an actual historical event is acknowledged by most earnest students of Indian history. Attempts have, from time to time, been made by a number of scholars to arrive at a pre­ cise date for this event. But these attempts have only resulted in the development of views so divergent from each other that even an approximate date conforming to the different data can­ not be deduced. Nor is this difficulty an incident of modern scholarship. Erudite scholars from at least the fiftli A.D. have found reasons for variance on the point. Never­ theless, these ancient estimates furnish some chronological clues which it would seem to be w'orth w'litle to investigate. Broadly speaking, there were current, in the 5th- A.D., three views concerning the date of the Great War. Aryabha^ (b. 476 A.D.) places the event in the 32nd century B.C. Varaha Mihira (d. 587 A.D.) assigns it to the 25th century B C. The Puranas which contain an account down, as we shall see, to 425 A.D., date the War in the B.C. We observe here a ver}’ large cleavage of opinion on the question already in the 5th-6th century when, we should think, the Indians had more reliable data to go upon than M'hat we possess to-day. Whence this cleavage ? One thing that strikes us at a first glance is that, while Aryabhata and Varaha Mihira give the dating in their astronomical treatises, reckoning apparently by astronomical data, the Puranas claim to derive the date from the reign-periods of successive monarchs. Why, then, do A ryabha^ and Varaha Mihira differ between them­ selves ? To answer this question, w’e shall have to see what astronomical data they possessed for working out the dates. To begin with Aryabhata. Research has sho^vn that the so-called Kaliyuga era of B.C. 3102 has a purely astro­ nomical origin and, originally, had nothing to do ^ith any actual historicat event.^ Nor is it of very ancient descent. It was devised by Hindu astronomers not long before A.D. 400 when, as a result of Grecian influence, they realized that they required a definite date from which to reckon the movements of the sun, the moon and the planets. In the earlier period, they had been content to calculate, for the purposes of astronomy as

1 JRAS, 1911, pp. 479-96, 675-98. A\’e]l as of astrology, the moveinents of only the sun and the moon, and were not interested in the movements of the planets. Now, however, by calculating backwards upon the basis of the then known mean values of planetary periods, with slight adjustments, they arrived at a certain date, namely, the 18th February, Friday, in B.C. 3102, when, according to these cal­ culations, the sun, the moon and the five planets w^ould have been in conjunction at the beginning of Mesa. Aryabha^ designates the day preceding this day as “ the Bharata Thurs­ day ” which is explained by his commentator as meaning the day on which Yudhisthira and the others laid aside their sover­ eignty and went on theii’ great journej". The underlying assumption is that Yudhisthira laid aside his sovereignt}" just before the sun. the moon and the five planets were believed to have come together on a particular day in B.C. 3102. Such an assumption could only have been made after the advent of Grecian astrology and astronomy. The planetary periods, the signs of the zodiac, the week-days, all of which enter into the calculations, are demonstrably of Grecian origin. The fact that, on the given date B.C. 3102, there was actually no such conjunction, nor even a near approach to such a conjunction, proves the date to have been derived by back- calculation. It is no wonder that such an artificial reckoning did not commend itself even to an astronomer like Varaha Mihira who, coming shortly afterwards, assigned a different date to Yudhis­ thira. Kalhana, writing in the A.D.,, definitely rejects Aryabhata's dating and avows adherence to the dating favoured by Varaha. We however can have no such confidence in Varaha's dating. He places Yudhisthira 2,526 before the Saka era.^ But the begmning of the Saka era does not correspond to the gauge- of Varaha. His gauge-year is 427 Saka and is taken over from the Romaka Siddhanta.^ There is thus an interval of 2,953 years between Varaha’s date for Yudhisthira and the gauge-year of the Romaka which is also his own gauge-year. The number 2,953 is suggestive: it is exactly 100 times the number of days (29 o3) in a lunar synodic month. The author of the Romaka, belonging as he did to the Alexandrian school of astronomy, may be presumed to have knowTi the Hipparchian (and also, Ptolemaic) estimate of precession which works out to 1 daj" in 100 years. Consequently, if there was reason to believe that the equinoxes and the solstices had been occurring in Yudhisthira’s time exactly one lunar month (Or 29’53 days) later than in 427 Saka, it might be inferred that a period of 2,953 years had intervened between Yudhisthira and 427 Saka. It will be show'n that there was reason for such a belief.

1 Brhat-Samhita 13,3. 2 JASB, 1884, p. 288. Vaiaha Mihira gives, in his Pancasiddhantika, rules for the calculation of aharganas, according to the Romaka Siddhanta, and states that the initial point for the calculations is Caitra sukla 1 of 427 Saka, ‘‘ when the sun had half set in Yavanapura ” (Alexandria). The last detail shoM's that Alexandrian astrono­ my was at the basis of the fixation of Caitra sukla 1 of 427 Saka as the starting-point of Varaha Mihira as well as of the Komaka. And, as this fixation was bound up with a new determination of the vernal equinox rendered necessary by observed deflection in the equinoctial position, it is legitimate to connect, as I have proposed to do above, this fixation ^^ith the Alexandrian estimate of precession. The statement in the Pancasiddhantika shows further that the spring equinox wa^< held to be occurring on Caitra sukla 1 in 427 Saka. Let us turn now to the state of affairs exhibited in the Mahabharata.^ Bhisma is lying on his bed of arrows, waiting for death. Being a virtuous man, he cannot die except in uttarayan a, that is, the half-year following the winter solstice. He has to wait till the 8th day after the winter solstice—a particularly auspicious day still celebrated in India as BhismastamI and marked as such in Hindu almanacs. On that day, BliTsma perceives that, the lunar month Magha being already on, uttarSyana had arrived, and it was already the eighth day of the lunar month. This stor^% according to Winternitz,“ formed no part of the original account. But he recognizes that the philosopher SaAkara speaks of Bhisma’s death in uttarayana, so that the story of Bhvsma's death as described in our present Mahabharata must have found place there already in the A.D. The fact that Sankara takes it as part of established tradition, and not as a recent incorporation, shows that it was found in the Maha­ bharata which lay before Varaha Mihira as well as the author of the Romaka Siddhanta (6th century). And the story must have indicated to them that, in the days of Bhisma—in the year of the Bharata War—the winter solstice coincided with Magha sukla 1. Comparing the two positions—vernal equinox on Caitra sukla 1 in 427 Saka, and winter solstice on Magha sukla 1 in the year of the battle of Kuruksetra—the inference was drawn that the equinoxes and the solstices had been occurring in Yudhisthira’s time exactly one lunar month (29‘53 days) later than in 427 Saka; and the Hipparcho-Ptolemaic rate of preces­ sion being availed of, the interval was calculated as one of 2,953 years. A dating derived in this way, on the basis of what we know to-day to be a wrong value of precession, can have no claim to oui- acceptance.

^ AnuSasana-parva, ch. 67. “ Geschichte der Indischen Literatur, vol. I, p. 307, There now remains for our consideration the Puranic dating which has the effect of placing the Great War in the 15th cen­ tury B.C. To be more exact, the Puranas posit a period of 1,015 or 1,050 years between the birth of Pariksit and the coronation-year of Mahapadma, I have elsewhere ^ given reasons for assigning this latter event to B.C 413. Briefly, my argument is as follows. The dynastic account in the Matsya, Vayu and Brahmanda Puranas is brought down just to the end of the 836th year after Mahapadma’s coronation ; and, im­ mediately afterwards is added an exposition of the Saptarshi reckoning. It is reasonable to suppose that the object of this juxtaposition was'to indicate that the Saptarshi era had been used here—that, in fact, the 836th j^ear after Mahapadma’s coronation fell in, or coincided with, the last year of a Saptarshi centennium. That the use of some era was necessary to make the Puranic chronology intelligible could not fail to have been recognized by the Puranic compilers who mention the Andhra dynasty as having gone by and allude to eighteen successive Saka rulers, thus betraying their late age. The allusion to the number of the Saka rulers irnplies that tlie Saka dynasty also had come to an end. This Saka dynast}- could only have been that founded by Cas^na which came to an end about 400 A.D.; and w'e actually find, from dated coins and inscriptions, that there flere, in this line. 18 mahaksaira'pas who issued coins dated, of course, in the Saka era.^ On the other hand, the Matsya version, which is earlier tlian the Vayu, ends the dynastic enumeration with a bare reference to the Vakataka djmasty of Berar which flourished between about 300 and 500 A.D., without, however, any allusion to the number of its rulers, sliowing that the dynasty had not yet come to an end. The iMatsya compilation should therefore be assigned to the A.D. No compiler of such a late age could ignore the necessity of mentioning an era to which his “ 836th year after Mahapadma" was intended to correspond. It W'as a year well known to him ; for, even by a rough estimate, the 836th year after Mahapadma falls somew'bere about the 5th century A.D. Of the various eras then in use, the Vikrama, the Saka, the Kalachuri and the Gupta eras had political associations which rendered them unsuitable foi' adoption in historical chronology. Moreover, the Vikrama era had been for four largely superseded, in popular use, by the Saka era; and, the Saka dynasty being recently overthrown, the use of the Saka era could not readily be thought of. Nor had the Gupta or the Kalachuri era met with general acceptance. The Mandasor inscription of

1 Trans. First Oriental Conjerence (Poona, I9I9), Poona, 1922, pp. 351ff. The date should be B.C. 412. -_R&pson, Cat. oj Goins of Andhras Y>-o&i ■ I exclude Nahapana and I^varadatta, neither of whom belonged to Caatana’s dynasty, and Nos. XVn and XXI of whom no coins are known. Bandhuvarman mentions Kumaragupta as reigning but employs the Vikrama era to express the dates (437 and 473 A.D.) The Kaliyuga era of B.C. 3102 was also known: but, as it was,in its origin, connected with thebeginning of the Kaliyuga, its use, if it were availed of b^' the Puranic compilers, would be in connexion with the period of Pariksit, not in connexion with the 836th year after Mahapadma. But it could not be so availed of, even. The Puranic compilers could not possibly harmonize their own chronology with such an early date for Yudhisthira w4iich would have the effect of placing such recent events as, for instance, the fall of the Andhra djTiasty several centuries B.C. Besides, it is not unlikely that the Puranic compilers were aware of the artificial character of the Kaliyuga era and could not on that account subscribe to the view that Yudhisthira flourished any­ where about the 32nd century B.C. The Saptarshi era, on the other hand, is also known as the sastra-samvatsara or the la II kika-sam vatsam. Th ese names indicate that the era must have been employed in ^astras requiring the use of an era and was well known to common people {laulcika). And the only sastras which required imperatively the use of an era are Sastras of the Puranic order, setting forth historical chronology. The Puranas, also, are literature of the laukika type, being meant for the masses. Kalhana, writing his Rajatarangiul in the 12th century A.D., uses the Saptarshi era in preference to the Saka era^ as soon as he is able to give, no doubt from older records at his disposal, any definite dates for definite historical events. This shows that the use of the Saptarshi era in historical chronology had already become traditional long before the age of Kalhana; and the only way in which it could have attained that status would be by its exclusive use in literature of the Puranic order. Its use. therefore, in Puranic compilations of the 5th and 6th centuries A.D. is just what we may expect from what we know of its k ter history. The 836th year after Mahapadma, which is the last definite date mentioned in the Matsya and the Vayu Puranas, thus fell in, or coincided with, the last year of a Saptarshi centennium.^ It is easy to see which centennium is meant: it corresponds to 325-425 A.D. An earlier centennium or a later one would con­ flict with Graeco-Indian synchronisms furnished by Asoka’s inscriptions. Moreover, as I have shown above, the Matsya compilation, where this date is given first, belongs to the 5th centui-y A.D. ; so that, it would naturally be to the centennium ending in 425 A.D. that the last date in the compilation would refer. A second set of data confirms this conclusion. The dynastic account in the Puranas pertains to the three Ages—the Treta-,

1 Tlie Saptarshi era is reckoned by centuries. A century began in A.D. 1925, and the next century will begin in A.D. 2025. the Dvapara- and the Kali-yugas—occupying altogether a period of {3,600 + 2,400+ 1,200 or) 7,200 years. The beginning of hif^tory is identified with the commencement of Treta ; and it is explicitly stated, at the end of the dynastic account, that the Kaliyuga (of the 28th caturyuga) had “ gone by.” The 836th year after Mahapadma thus fell in or coincided with the last year of the Puranic Kaliyuga ; consequently if that year be equated with 425 A.D , the commencement of Treta or the beginning of tVie historical period would have to be dated 6776 B.C. We possess the evidence of Megasthenes, thanks to Pliny and Soiinus. that this was also the date assigned in Puranic literature of the B.C. to the beginning of Indian history.^ That is a striking corroboration. Yet another confirmation is available. At the close of the dynastic account, just before the state­ ment regarding the interval between Mahapadma and Pariksit, there is a couplet in the Vayu, Brahmanda, Visnu and Bhagavata Puranas, indicating the use of the Jupiter cycle as helping to define the closing year of the dynastic account. The couplet runs thus; yada candrasca suryasca tatha tisya-brhaspati ekara.s'au samesyanti tada krtayugam bhavet Translation. ''' When the moon, the sun, and likewise Tisya and Brhaspati, will be together in the same sign {ra&i), then will Krtayuga be.” Fleet, commenting on the verse, which occurs also in substantially the same form in the Mahabharata {Vanaparva, ch. 190), remarks: “ The verse does not exactly assert what is technically known as a conjunction: it only says that the sun. the moon, Jupiter and Tisya ‘ will come together or will be (together), in one sign.’ But a conjunction is obviously imphed ; because otherwdse the occurrence will be too common. Jupiter spends nearly one year out of every t\^ elve in Karka ; and, on each occasion while he is there, he will be in conjunc­ tion with Tisya, and the sun and the moon will be in conjunc­ tion with each other in that same sign once if not twice ; but it is only at very long intervals that all the four will be in oon|unetion." It is unnecessary, however, to travel, with Fleet, beyond what is exactly asserted in the verse, the object of which is not to give an absolute date for the beginning of the Krta^mga but to define more precisely the year iv, which this Krtayuga was to begin. As Fleet himself was puzzled to observe, it must have been only about the beginning of the Hindu summer solstice that the state of things described in the verse (viz. the sun in the ‘ sign ’ Karka) would belong ; whereas the beginning of Krtayuga, as indeed of evers^ one of

1 Indian Antiquary, VI, p, 250. Arrian’s figure.s are inutilisable. the four Yugas, is required to coincide with a vernal equinox. The contradiction does not exist ; for what the verse assertb is that Krtayuga will begin—of course, at venial equinox—^during a Tisya-samvatsara of a Jupiter cycle, the samvatsara commencing with an amavasya. The use of the word rasi {‘'sig n ’") in the expression eka-raMu shows that the Twelve-year cycle of the mean-sign system is here employed. And A.D. 424 (Nov. 20)-425 (Kov. 16) conforjus to this requirement. This use of the Jupiter cycle for the purpose of indicating a particular year in the 5th century A.D. is not surprising, because the cycle is known to have been much in vogue in the 5th and 6th centuries AD., as inscriptions and astronomical books testify.^ It is worth noting that, in the inscriptions, it is seldom used alone but almost always appears joined to a date expressed in some other era ; so that its introduction M^as made not so much to give an absolute date as to provide a clue to the determination of the era expressing the date. In fact, it was just because, in some inscriptions, the Jupiter cycle had been used in conjunction with dates expressed in the Gupta era, that Fjeet was able to demonstrate the initial point ot that era. One other point needs notice in this connexion. How comes it that the Jupiter cycle used here belongs to the mean sign system, and not to the heliacal rising system ? The latter system is found to occur in the several Northern Indian inscriptions hitherto known ; and it is also the older system, having a larger number of votaries among the early astronomers. For an explanation, we may point to the occurrence of the mean- sign system alone in the astronomical treatises of Aryabhata of Kusumapura (Patna) and Brahmagupta of Bhillamalla (Bhin- mal), and also to the fact that, in Marwad and among Marwadi settlers in other parts of India, the mean sign system obtains. Moreover, the years of the heliacal rising system are f)f unequal duratioi, not convenient for employment in historical chrono­ logy ; and its use in inscriptions was probably commended by the circumstance that its years, commencing from actual risings of Jupiter, had, in the eyes of the people, an air of reality which the mean sign system lacked There can thus remain no doubt that the 836th year after Mahapadma’s coronation ended in the Tisya-sarhvatsara of the Jupiter cycle corresponding to 424-5 A.D. which overlapped the last year of a Saptarshi centennium and the terminal year of the Puranic Kaliyuga. We are thus entitled to assume that the year of Pariksit’s birth, that is, the year of the Great Battle of Kuruksetra, is, according to the Puranas, 1427 or 1462 B.C. (1,015 or 1,050 years before Mahapadma). These intervals, of so many years before and so many years after Mahapadma,

1 Fleet, Gupta hiscrs., App. Ill (by S. B. Dikshit). Cf. Fleet's Art. ‘Hindu Chronology ’ in Enc. Britt. (11th ed.). profess to be based, as already observed, upon reign-periods of successive monarchs. And, just as astronomical eras—at first the Saptarshi era, and later, the Jovian era—are referred to in connexion with the closing j^ear of the account {424-5 A.D.), so an astronomical detail, also connected with the Saptarshis. or stars of the Great Bear, is furnished with refer­ ence to “ the period of ParikBit.” The detail is thus expressed ; saptarsinarh tu yau purvau drsyete uditau nisi tayor madhye tu naksatram drsyate yat samam divi . tena saptarsayo yukta jfieya vyomni satam samah naksatr.inahi rsinarii ca yogasy-aitan nidarsanam saptarsayo Maghayukta kale Pariksite satarii brahmanas tu caturvirnse bhavisyanti satam samah ^ Translation. “ Those two of the Seven Rishis which are the first (among them) to rise at night—the nakshatra which is seen to lie between them equally in the heavens—with that (nakehatra) are the Seven Rishis reckoned to be conjoined for a hundred years. Here is an illustration of the conjunction of nakshatras with Rishis : the Seven Rishis were joined to Maghas for 100 years in Pariksit’s time ; and the Brahmanas (Rishis) will be for 100 years in the 24th (nakshatra).” Then follows the statement that thenceforward this whole world will be in trouble,” introducing details of miseries reserved for the sandhyamsa period—or tlie last two centuries— of the Kaliyuga (225-425 A.D.). Sridhara Svaml, commenting on the Bhagavata version of this passage (which omits 11. 4 and 6), remarks ; “ that the constellation, consisting of seven stars, is in the form of a wheeled carriage. Marichi, he observes, is at the extremity and next to him, Vasistha in the arched, part of the yoke ; and bej^ond him Angiras: next to whom are four stars in a quadrangle : Atri at the north-east corner south of him Pulastya ; next to whom is Puiaha : and Cratu is north of the last. Such being their relative position, the tw'o stars, which rise first, are Puiaha and Cratu ; and whichever asterism is' in a line south from the middle of those stars, is that with which the Seven Rishis are united ; and they so remain for 100 years. A similar passage is found in the Visnu Purana, and a similar exposition of it is given by the coinmentator Ratnagarbha.” ® The attached diagram (Plate 2) illustrates the position depicted by the commentators for an observer in northern latitudes. A similar stellar disposition is also described by Varaha Mihira Br. S., ch. 13): there is only a slight diflEerence,

1 Pargiter, Dynasties of the K ali Age, pp. 60-61. The Matsya version is cited here. ^ Colebrooke, Essays, II. 357. the names Atri and Pulastya being interchanged, which, how­ ever, does not affect the stellar disposition. Varaha Mihira^ citing from Garga, speaks of the Great Bear rising in the north east, Marlci being the easterninost star, with Vasistha to its west. In both accounts, an evening aspect is contem­ plated, since both speak of the ‘rising’ (udaya) oi the stars; and Sridhara makes the meaning explicit, so far at least as the Puranic account is concerned, when he speaks of Pulaha and Kratu ‘ rising first.’ Moreover, it is obvious that the evening or sunset of a particular day in the year must be in­ tended—a particular day, besides, which was so well known to astronomers as not to require specific mention. In other words, the observation was meant to apply to the moment of sunset of an equinoctial or a solstitial daj^ That the spring equinox only can be intended follows from the fact that the heavens, at sunsetj conld have presented such an appearance onlj^ about spring­ time. We may consider, also, in this connexion, the fact that astronomers of that period reckoned only from vernal equinox as the starting-point of every yuga ; and the Puranas are con­ cerned here with the starting-point of its 29th caturyuga, ( = the ending-point of the 28th mturyuga) which, as I have explained above, was held to coincide with vernal equinox in 425 A.D. I ’inally, the years of the Saptarshi era which has been used here are Caitradi lunar and consequently start from the vernal equinox. It follows from all these considerations that the disposition of the Great Bear as described in the Puranas pertains to the moment of sunset on the day of the spring equinox. On that day the sun rises due east and sets due west, and the solstitial colure cuts the observer’s horizon at the north and the south points. What, therefore, the Puranas understand by the statement that the Seven Rishis were joined to Maghas in the time of Pariksit ” is that, in the days of Pariksit, the solstitial colure passed midway between the stars Pulaha and Kratu {a and ^ Ursa Majoris) and intersected the Maghas naksatra (of which the principal star is Regulus). That is wh}' Sridhai'a speaks of Maghas having been, at that time, on a line drawn south from the middle of these two stars. Calculating with the now well-ascei'tained rate of pre­ cession (60'2" a year), we can realize (see Plate 2) that this was actually the position about 1400 B.C.—approximately the period to which Pariksit is assigned by the Puranic chrono­ logy based upon dj^nastic lists and reign-periods. It might seem that this is no confirmation, but a coincidence due to the Puranic chroniclers themselves having worked back at a known rate of precession. But such cannot have been the case, since the modem value of precession was not then available and is indeed not found in any Hindu book. The closest approximation to such a value, viz. 54" a year, is met with for the first time in an interpolated chapter of the Surya Siddhanta {c. 1000 A.D.) aud is unknown to Aryabha^, Vara ha Mihira and Brahmagupta. Varaha Mihira notes, it is true, ^he fact of precession. Indeed, he gives prominence to the fact by men­ tioning it at what we may regard as substantially the commence­ ment of his treatise, the Brhat Sarhhita (ch. 2). But he omits to give the rate of precession as anything like 54" a year. On the other hand, as I have shown above, he works back himself or accepts from the Romaka a working back to the age of Yudhisthira at the Hipparcho-Ptolemaic rate of 36" a 3-ear. Supposing even that the rate of 54" a year was known to the Puranic chroniclers of the 5th century A.D., a back-calculation on that basis would yield a different position for the solstitial colure for the date they assigned to Pariksit, the difference amounting to something like 2 degrees ; so that, with such a ■calculation, the colure would be represented to have passed, not midway between a and ^ Ursa Majoris but considerably nearer to the latter. We are thus obliged to concede that the position of the colure as defined in the Puranas for " the po iod of Pariksit ” is not the result of a back-calculation adopted with the object of providing anj" specious asti’onomical confirmation for their historical chronology, but is based upon genuine tradition of actual observation made about 1400 B.C. when the colure actually passed midway between a and fS Ursa Majoris.’^ And, as the Puranic date of Pariksit, based upon an altogether different set of data, namely, the reign-periods of successive monarchs, tallies with the astronomical evidence, we may safely infer that the Kurus fought the Pandavas about 1400 B.C.'^ ■ ■

1 The position described (‘ mid-way between Pulaha and Kratu implies great accuracy of observation. Adapted, to a certain degree of approximation, from Plate I, Art. C o n s t e l ­ l a t i o n in the Encyclopcedia Brittanica (llth edition). Drawn by Kumar Suhrid Sinha. The seven stars of Ursa Major, counting from a, p, onwards, are named: Pulaha, Kratu, Atri, Pulastya, Angiras, Vasistha and Marld. According to Varaha Mihira, the star y is Pulastya, and the star 8 is Atri.