The Mahabharata War is a critical marker in early Indian history. By popular tradition, the Kali Age started with the death of Krishna, 35 years after the War. The Kali calendar has a beginning of 3102 BC; therefore, it is thought that the Mahabharata War took place in 3137 BC.

The first mention of the Kali calendar is by the astronomer Aryabhata in his treatise on astronomy with an internal date of 500 AD. The earliest epigraphical reference is in the 5th century inscription of King Devasena where it is alluded to indirectly, and in the Aihole inscription of 3735 Kali (634 AD). Because of these late references, some scholars have suggested that the Kali calendar was started at a late period with an assumed conjunction at the beginning of the era for convenience of calculations, and, therefore, the Aihole inscription cannot be taken as proof of the date of the War.

Modern studies using powerful software that can reconstruct the ancient skies indicate that there was actually an approximate conjunction of the planets on Feb 17, 3102 BC as taken by Aryabhata. This may only be a coincidence. Even if the Kali calendar is as old as its starting date, its connections with the Mahabharata War do not appear to be equally ancient. There are also other traditions related to the War. Some of them are old, some new. Of the competing theories, the most prominent ones, in my view, are:

1. The date of 1924 BC. This is from the Puranic genealogies.

2. The date of around 1000 BC. This is the date popularized by Western Indologists as being most “reasonable” based on archaeological data. Repeated in numerous school texts, it has achieved a certain kind of canonicity. This date was first proposed within the framework of the Aryan invasion theory. Although that theory has been discredited, this date has taken independent life of its own.

I shall examine these three different dates while considering the evidence from the Mahabharata, the Puranas, archaeology and astronomy.

The Mahabharata Epic and Archaeology

Is the Mahabharata epic -- the text of 100,000 verses -- which is a source for the events of the War to be taken as history? The epic itself claims to have been originally just 8,800 verses composed by Krishna Dvaipayana Vyasa and called the Jaya. Later, it became 24,000 verses, called the Bharata, when it was recited by Vaishampayana. Finally, it was recited as the 100,000-versed epic (the Mahabharata) by Ugrashravas, the son of Lomaharshana.

Thus, the tradition acknowledges that the Mahabharata grew in stages. The core of the story is very ancient and there is astronomical evidence in it related to the Ashvamedha rite that indicates a period before the 3rd millennium BC (see my book The Ashvamedha for details). The details of the final version may very well include episodes that are poetic exaggerations or imagined material. We see such poetic imagination at work by comparing the Ramayanas of Valmiki and Tulasidas.

Many of the characters of the Mahabharata are mentioned in the Vedic texts that, on account of being considered sacred, have not suffered interpolations and should thus represent historical persons. Krishna, for example, is mentioned in the Chhandogya Upanishad. Other names occurring elsewhere include Vichitravirya, Shantanu, Dhritarashtra, Janamejaya, and Parikshit.

Due to its expansion over several centuries, the Epic includes late material. This means that dating the events of the Epic based on archaeological finds could be misleading. Some scholars have correlated the painted grey ware (PGW) pottery of the period of 1100-900 BC found in Hastinapur (modern Hathipur) to the Kauravas. But there is no basis for such correlation. The Kurukshetra site itself has structures that go back to about 3000 BC.

Panini's grammar (c. 400 BC) knows the Mahabharata. Also, the Epic, in its long descriptions of the religions of the day, does not mention Buddhism, so we can be certain that it was substantially complete prior to 400 or 500 BC. The language of the Epic does not always follow Paninian constructions which also suggests that it is prior to 500 BC.

Dion Chrysostom, Greek Sophist (40-105 AD) mentions that the Indians possess an Iliad of 100,000 verses. Together with its appendix, the Harivamsha, the Epic does add up to this total.

Recent archaeological discoveries indicate that the Sarasvati river dried up around 1900 BC, leading to the collapse of the Harappan civilization that was principally located in the Sarasvati region (accounting for about 70 percent of all the Harappan sites). The Rigveda celebrates the Sarasvati as the greatest river of its day, going from the mountains to the sea (giribhya asamudrat).

There are two schools of thought related to the drying up of the Sarasvati river. According to the first one, the Sarasvati ceased to be a seagoing river about 3000 BC, explaining why the 3rd millennium settlements on the banks of the Sarasvati river end in the Bahawalpur region of the Punjab and do not reach the sea; there was a further shrinking of the river in about 1900 BC due to an earthquake that made its two principal tributaries to be captured by the Sindhu and the Ganga river systems. According to the second view, the Sarasvati flowed to the sea until 1900 BC when it dried up. The first view explains the geographical situation related to the Harappan sites more convincingly.

Given the understanding of the drying up of Sarasvati, with its preeminent status during the Rigvedic times, it follows that the Rigvedic hymns are generally anterior to 1900 BC. If one accepts the theory that the Sarasvati stopped reaching the sea in 3000 BC, then the Rigvedic hymns are prior to 3000 BC. If the tradition that Vyasa was the arranger of the Vedas is correct, the latter explanation would mean that the Mahabharata War could indeed have occurred in 3137 BC.

The Puranic Tradition

The Puranas have extensive king-lists together with the years of reign for each king for several parts of the country. The Puranic king lists speak of roughly 100 generations before the Mahabharata War. The Greek historians inform us that the Indians during the time of the Mauryas remembered more than 150 generations of kings spanning over 6,000 years. (We assume that these lists remember the prominent kings only.) These lists come down to the 4th or the 5th century AD and they are quite accurate in their details for the post-Mauryan period for which independent inscriptional evidence is available. One would expect that they would be accurate for the period prior to the Mauryas also. The regnal years are given in the Puranas only for the post-War kings.

The king-list for Magadha has the following dynasties in the post-Bharata War period:

1. Brihadrathas (32 kings)                    967 years
2. Pradyotas of Avanti (5 kings)           173 years
3. Shishunagas (10 kings)                     360 years
4. Nandas (Mahapadma + sons)          100 years
5. Mauryas (9 kings)                            137 years
6. Shungas (10 kings)                           112 years
7. Kanvas (4 kings)                              45 years
8. Andhras (30 kings)                          460 years

One may question the reliability of the earlier parts of this list since the average span of reign for the pre-Nanda kings is more than twice as much for the post-Nanda ones. The explanation appears to be that it was during the imperial Maurya age that comprehensive king-lists were made and, consequently, only the better-known names of the earlier period were included. The centennial counting system, named after the nakshatras, made certain that the count of the dynastic totals was accurate.
During the pre-Nanda period, the list also provides for 24 Aikshavakus, 27 Panchalas, 24 kings of Kashi, 28 Haihayas, 32 Kalingas, 25 Ashmakas, 36 Kurus, 28 Maithilas, and 23 Shurasenas.
We know that Chandragupta Maurya started his reign in 324 BC. Therefore, if we were to accept these periods, the dynastic eras for the post-Bharata age will be:

1. Brihadrathas      1924-957 BC
2. Pradyotas          957-784 BC
3. Shishunagas       784-424 BC
4. Nandas              424-324 BC
5. Mauryas            324-187 BC

and so on.

It is most significant that the Puranic king-lists imply 1924 BC as the epoch of the Mahabharata War. Since this epoch is virtually identical to the rough date of 1900 BC for the catastrophic drying up of the Sarasvati river, it suggests that the two might be linked if they are not the same. The disruption due to the earthquake may have been a contributing factor to the Mahabharata War, or the War could have served as a metaphor for the geological catastrophe.

The Earliest Indian Calendar

The earliest calendar in India was centennial, with a cycle of 2,700 years. Called the Saptarshi calendar, it is still in use in several parts of India. Its current beginning is taken to be 3076 BC. Notices by the Greek historians Pliny and Arrian suggest that, during the Mauryan times, the calendar used in India began in 6676 BC. It is very likely that this was the Saptarshi calendar with a beginning of 6676 BC.

Around 500 CE, a major review of the Indian calendar was attempted. The astronomers Aryabhata, Varahamihira and others used the naksatra references that the Saptarshi were in Magha at the time of the Mahabharata War to determine its epoch. Aryabhata declared the War to have occurred in 3137 BC, and Varahamihira assigned it 2449 BC. This discrepancy arose perhaps from the different assumptions regarding the nakshatras (27 or 28) in the calculations of the two astronomers.

It is likely that the fame of the Kaliyuga era with its beginning assigned to 3102 BCE prompted a change in the beginning of the Saptarshi era to about the same time, that is to 3076 BC.

The Puranic memory of the Mahabharata war having occurred in 1924 BC may represent the transference of a much earlier event to the cataclysmic event at the end of the Harappan period. The memory of the War in popular imagination may represent the conflation of two different actual events.

Coda

The date of 1000 BC or so is just not possible because it is at variance with the astronomical facts related to the period. Furthermore, it is at variance with the Puranic genealogies which, we know, are quite accurate in the post-Mauryan period and are likely to have been accurate earlier as well. Then there are various remembered lines of teachers that show up in various texts. Specifically, the Brihadaranyaka Upanishad remembers a line of 60 teachers. We don't know how many years should be assigned to each teacher but this line could span substantially more than a thousand years. Given that this Upanishad is about 800 BC in the most conservative reckoning, this long list makes it impossible for the Rigvedic period to end in 1000 BC, as required by the War in that epoch.

Recently, astronomical evidence related to the mention of eclipses and the placement of planets in the narrative of the Mahabharata has been examined by several scholars. The date of 3067 BC (35 years after 3102 BC rather than 35 years before it) has provided a nice fit for a large number of data points. But such analysis depends on much subjective interpretation of the stray astronomical references and assumes that the astronomical ideas used in that epoch were the same as in the Classical Surya Siddhanta tradition. Neither can we be sure if the passages used are from the core text or from the much later accretions. Therefore, such an approach will remain highly controversial.

This leaves us with the dates of 1924 BC and 3137 BC. I don't think we have evidence at this time to pick one of these two as the more likely one. If one gives credence to the Puranic genealogies, then 1924 BC would be the time for the War; if, on the other hand, we go by the astronomical evidence related to the Vedas and the subsequent literature, then 3137 BC remains a plausible date. If the pre-Nanda Puranic lists are not accurate for the regnal periods, then the War will have occurred a few centuries later than 1924 BC.