Dating the Mahabharata war – A tale of eclipses, archaeology, and genealogies
Using a combination of data from ancient texts, contemporary astronomy and radiocarbon dating of tsunamis, the author of this essay calculates that the Mahabharata war was fought this fortnight 2922 years ago
Is it possible to find a precise date for what was perhaps the most important battle in our early history, the Kurukshetra war? I believe this is entirely possible using clues from a specific episode of the war.
The fourteenth day of the Mahabharata war was a crucial one for both sides. On the evening of the thirteenth day, Arjuna, devastated by the death of his young son, Abhimanyu, had vowed that he would kill Jayadratha – the man he held responsible – by the sunset of the next day. If he could not, he would give up his own life. Krishna immediately sent a spy to the Kaurava camp. The spy reported that the Kauravas were planning on having six maharathis, or lead warriors (including Karna and Ashvatthama) form a protective cordon around Jayadratha all day. They hoped to frustrate Arjuna’s efforts and thereby compel him to commit suicide – practically guaranteeing the outcome of the war. Learning of this, Krishna retired to his room, brooding and tense. However, he seemed fine the next morning.
Things were not going well for the Pandavas on the fourteenth day. It was past midday, but Arjuna had not yet succeeded in launching a sustained attack on Jayadratha. The six warriors kept blocking all his attempts. It was early winter, so sunsets anyway happened fairly early, but today, the sky seemed to be darkening even faster than usual.
Krishna tells Arjuna that it would shortly look as if the sun has indeed set; the sun’s disc would be obscured, and the sky would darken. While everyone else would think that this was a “genuine” sunset, Arjuna should know that it was actually “fake” and that the sun would reappear after some time. He explains that everyone on the Kaurava side would become careless as soon as the sun was obscured and would begin rejoicing, thinking their troubles over and expecting Arjuna to die. He should then take advantage of the confusion to kill Jayadratha. Arjuna agrees immediately and things go according to plan. Jayadratha, who had hidden himself, comes out and stares at the sun, his head thrown back. Arjuna immediately attacks him, also startling the other warriors who have let go of their weapons and are relaxing. They can’t see well and are bewildered that Arjuna is fighting instead of committing suicide.
Finally, Jayadratha is killed, and the sun reappears very soon after. The Kauravas realise that the “sunset” they had seen was an “illusion”, while the Pandavas cheer and kill many more people before the (real) sunset. All this is described in detail in chapters 147 and 148 of book 7 (Drona Parva) of the Sanskrit Mahabharata as well as in KM Ganguli’s English translation (7.145).
Discounting supernatural explanations, this seems a clear instance of a solar eclipse that occurred after midday and ended before sunset. If we could find the date of such an eclipse precisely, we might be able to date the Mahabharata war. Vyasa provides some other clues about possible dates in Book 6.3, in which he visits Dhritarashtra before the battle is to commence. While discussing a series of ill omens that he has been seeing recently, Vyasa laments the fact that the moonlight is dim even on “Kartiki pournamasi”, that is, on the full moon night of the month of Kartika (corresponding roughly to mid October-mid November). This points us to the fact that we are now looking for a solar eclipse that occurred in (or even slightly after) the month of Kartika, and that there would have been a full moon very shortly before the war started. Also, we are looking for a solar eclipse that definitely ended before sunset, and that did not start too early in the day.
To narrow down possible time intervals in which to look for a solar eclipse that satisfies all these conditions, I rely on research independently conducted by different experts. First, noted archaeologist BB Lal conducted some exciting research on Hastinapur (the capital of the Kauravas, won by the Pandavas after the war). In 1950-52, he excavated the site of Hastinapur, 60 miles northeast of Delhi. He discovered many layers of habitation. In the second-oldest layer, there was evidence that the habitation had been abruptly cut off by a massive flood of the Ganga; after which the city had been abandoned for more than two centuries. The archaeologists were able to date this flood to about 800 BCE.
Interestingly, Puranic texts including the Vayu Purana and the Matsya Purana mention that a flood destroyed Hastinapur during the reign of King Nichakshu, who was in the seventh generation after the Pandavas (he was the fifth ruler from the Pandavas’ grandson Parikshit. Parikshit ascended the Puru throne after Yudhishtir). They also mention that the severity of the flood forced Nichakshu and his people to abandon Hastinapur altogether and move to a new capital city, Kaushambi. Now, archaeologists had also discovered very distinctive “painted greyware” pottery in the second-oldest layer at Hastinapur; they had found exactly the same kind of pottery at other sites mentioned in the Mahabharata, that would have been contemporaneous with the time of the Pandavas (such as Ahicchatra, the capital of Drupada, Draupadi’s father, and also Kurukshetra, the site of the battle, Mathura, and Barnawa, or Varanavat, where the Pandavas had to flee arson). The painted greyware culture was carbon-dated to be roughly between 700 and 1200 BCE.
Even more intriguingly, there was archaeological evidence of “painted greyware” at Kaushambi dating from the period at which Hastinapur was abandoned – suggesting that Hastinapur’s displaced inhabitants may well have begun rebuilding their lives shortly after, in Kaushambi. All this led Dr BB Lal to identify the major flood that cut off habitation at the end of this “painted greyware” period, with the flood experienced by Nichakshu, which caused him to shift his people to Kaushambi. This, in turn, led him to hypothesize – based on the number of generations of rulers between Nichakshu and the Pandavas, and given that average reigns were not very long (perhaps 13 or 14 years on average) – that a date of 800 BCE for Nichakshu’s flood would suggest a date between 900 and 950 BCE for the Mahabharata war.
Decades before these archaeological finds, FE Pargiter had independently reached a conclusion about the time period of the Mahabharata battle. Pargiter, who was a civil servant and judge in British India, extensively researched Puranic and epic sources for the genealogical information they contained on various royal dynasties. He even rigorously established synchronicities between royals of different dynasties, so that he was able to assign them a generation number and say which kings of one dynasty would have been contemporary with those of others.
To arrive at a probable period for the Mahabharata war, Pargiter examined the information on royal genealogies from after the war, all the way up to Chandragupta Maurya’s time. As detailed in his books Ancient Indian Historical Tradition (1922) and The Purana Text of the Dynasties of the Kali Age (1913), several Puranas (such as the Vayu and the Matsya) have information on the number of kings in 10 contemporaneous dynasties between two fixed points of time. The earlier “fixed point” is the reign of Adhisimkrishna, who was the fourth successor of Parikshit (the Pandavas’ grandson). The later fixed point is towards the beginning of the reign of Mahapadma Nanda, who apparently exterminated these 10 dynasties. Essentially, the Puranas provide numbers of the kings in each of the 10 dynasties in between the times of Adhisimakrishna and Mahapadma Nanda.
Now, it is known that Chandragupta Maurya’s reign started in 322 BCE, a few years after Alexander’s invasion of 326 BCE. According to the Puranic sources available to Pargiter on the combined duration of the reigns of Mahapadma Nanda and his sons (who ruled just before Chandragupta Maurya), Pargiter finds that Mahapadma began reigning around 402 BCE. Allowing for the fact that it might have taken him about 20 years to exterminate the pre-existing dynasties, Pargiter assigns a time of 382 BCE to the later fixed point (when the ten contemporaneous dynasties came to an end).
His next task was to assign a time to the earlier fixed point. Now, each of the 10 dynasties mentioned by the Puranas had different numbers of kings ruling between these two points – these numbers ranged from 20 to 32. Kings from dynasties with a relatively high number of rulers in a fixed time interval would have had shorter reigns on average. Pargiter averages the number of kings over these 10 dynasties, to 26 kings. On the basis of his study of many royal dynasties all over the world, he then assigns an average reign of 18 years to each king, and assuming 26 such reigns, he obtains a length of 468 years for the interval between the two fixed points. This gives him a date of 850 BCE for the beginning of the time interval, which started during Adhisimkrishna’s reign. Using the fact that Adhisimkrishna is six generations junior to the Pandavas, Pargiter then suggests an approximate date around 950 BCE for the Kurukshetra war. This date could be somewhat earlier, or later, depending on the generation lengths assumed.
Earlier, I and a co-author independently found an approximate date for King Rama’s time (around 1350 BCE). Our conclusions were based on Professor Iyengar’s findings on the dates of some ancient astronomers, Pargiter’s information on some dynasties contemporaneous with Rama, and on some data in a specific Rigvedic hymn. Details on how we arrived at our conclusion are available in our article in HT Books.
Now, according to the genealogical tables compiled in Pargiter, the Pandavas are 30 generations after Rama’s time (while Rama corresponds to generation number 65 in Pargiter’s list, the Pandavas are generation 95). This, along with our finding of approximately 1350 BCE for Rama’s time, in addition to Professor BB Lal’s and Pargiter’s independent conclusions of the approximate times for the Mahabharata war, made me zone in on 800 to 1000 BCE as a likely interval for the war. However, I wanted to try and pinpoint an exact date.
To do this, I used data maintained by Fred Espenak of NASA. I checked for solar eclipses visible from Delhi (the nearest geographical point in the database to Kurukshetra) within the interval 800-1000 BCE. I looked for a solar eclipse roughly in or around the month of Kartika, that started after midday, and ended before sunset (fortunately, all these details were available in the data).
I found only one solar eclipse satisfying all these criteria. This happened on 5th November, 900 BCE, so it was in the month of Kartika. The eclipse started at 14.25 (after midday) and ended at 15.54, and this was before sunset on that day. (The database specifically indicates if an eclipse happens to be ongoing at sunset or if it is already over.) I checked the historical table of moon phases (also maintained by Espenak) and found that there was a full moon on 21st October, 900 BCE. This would have been a “Kartika purnima” and would have been very close to the start of the war.
Many attempts at tracking the date of the war using astronomical data involved a hunt for a pair of eclipses – with a lunar eclipse preceding a solar eclipse. However, from studying the text of the conversation Vyasa has with Dhritarashtra when he visits him during the Kartik purnima before the war, it is clear that he does not refer to a lunar eclipse. He says
Alakshyah prabhaya heenah pournamasim ca kartikimChandro abhut agnivarnashcha samavarne nabhastale. (6.2.23)
This can be translated as follows. “On this full moon night of the Kartik month, the moon’s light is dim and it is hard to see because both the moon and the sky have exactly the same colour, the colour of fire” (my translation). Now, this rules out a lunar eclipse because one of two things can happen during a lunar eclipse. Either the moon is red (a blood moon) while the surrounding sky is black (NASA explains that this happens due to refraction) in which case, though the moon may be the colour of fire, it is certainly not hard to see (alakshyah) because the sky is not the same colour as the red moon. Alternatively, both the moon and the sky can be the same colour – black – in which case the moon is certainly hard to see, but then, as fires cannot be black, neither the moon nor the sky would be described as having the colour of fire. Lunar eclipse data from NASA show that there was no lunar eclipse on the night of 21st October 900 BCE, which confirms my reading of the text.
Why might even a full moon have appeared dim and small (as described by Vyasa)? This could be the case when the moon is very near its apogee (the farthest distance from the earth along its orbit) at times very close to the full moon. This is technically called a micro-moon. To check if this could have applied to October 21st, 900 BCE (my candidate Kartik purnima night shortly before the Mahabharata war), I used Fourmilab’s lunar perigee and apogee calculator, finding that the apogee actually occurred on 20th October, 900 BCE. So, even on the full moon night, the moon would have been pretty far away from the earth, and its light would have appeared dim.
From the way events unfold during the fourteenth day, it seems that Krishna may have realized that a solar eclipse was about to occur, which is why he told Arjuna that the “sunset” was fake but that no one else would realize that. Could eclipses be predicted in ancient India? Interestingly, Professor RN Iyengar, in his research on the archaic astronomers Parashar and Vriddha-Garga, not only dates them to about 1400 BCE (in other words, about 500 years before the solar eclipse I find), but he also details how Parashar himself already had 1300 years’ previous night sky observations of comets, eclipses, stars and planets to work with. These were all passed on from generations of teachers to their students. Even though Parashar and Vriddha-Garga did not use mathematical astronomy, many centuries’ worth of observations had led them to start identifying patterns in eclipse sequences (Iyengar 2008).
The next development in Indian astronomy that we’re aware of represents the Pancha Siddhantas. These were early works containing mathematical astronomy. Varahamihira (a much later astronomer, writing in the early 6th century AD) had preserved a large portion of their texts and reproduced them. Now, of these, the Vashishta siddhanta can be dated to roughly 1100 BCE, because as the translator (Sastry, 1995) of Varahamihira’s collection of Pancha Siddhantika notes, some of the discrepancies in the position of (celestial) objects between Varahamihira’s time (about 500 AD) and the position calculated in the Vashishta Siddhanta suggests a time gap of 1600 years between the two. A date of 1100 BCE for the Vashishta Siddhanta would not be too startling given that Parashar and Vriddha Garga had worked three centuries before this. The Vashishta Siddhanta lays down a relatively simple math calculation for predicting if a lunar eclipse would occur. One needed to know the longitude of the moon, the sun, and the angle between the full moon and the ecliptic node (which is called “Rahu” in these texts). It also provides some additional calculations involving the latitude of the moon which predict the duration of the eclipse. The method used to compute whether a solar eclipse would occur (given in the other Siddhantas) is also very similar to the method used to predict a lunar eclipse.
Now, Varahamihira mentioned that being able to predict a solar eclipse was regarded as a great secret. This was because those in positions of power, who had this special knowledge beforehand, could use it to their advantage. So, astronomers in the know tended to stay silent about their ability to predict solar eclipses. However, since the Mahabharata war would have occurred 200 years after the Vashishta Siddhanta, it’s likely that astronomers could predict solar eclipses by then. In fact, Krishna would have had access to two well-known astronomers in the Pandava camp – Asita and Devala, who were brothers of the Pandavas’ chief priest, Dhaumya. So, it is completely plausible that he might have been able to consult them, and learn about the solar eclipse of the 14th day of war.
Finally, since the Mahabharata, and other sources, say that a massive flood occurred in Dwaraka roughly 36 years after the Kurukshetra war, I tried to find corroborative evidence of the flood. Data on actual historical floods or tsunamis from our part of the world is extremely scarce, and when it exists it does not go back too far in time. Intriguingly, though, geologists have recently found some evidence of prehistoric tsunamis. Triggered by the 2004 Indian Ocean tsunami, geologists examined a coastal cave in Aceh, Indonesia, and found that it had preserved evidence of prehistoric tsunamis that mostly originated in the Andaman-Nicobar region and most likely affected India. In a paper published in Nature, (Rubin et al 2017) the geologists used radiocarbon dating to date each such tsunami. They find evidence of a tsunami in the interval 822 BCE – 1000 BCE (which they interpret as the latest and earliest possible dates for that tsunami). This interval is consistent with the date I find for the Mahabharata war.
Considering all this, I would say that the Mahabharata war was fought from 23 October to 9 November, 900 BCE. Of course, there have been previous attempts to date the Mahabharata war. I relied on the expertise of archaeologist BB Lal and on Pargiter’s research to identify a time interval within which to conduct my search and was able to get a very specific, unique date by a careful reading of the text, thanks to Espenak’s eclipse data.
Brishti Guha has a PhD in economics from Princeton and is currently an associate professor at the School of International Studies, JNU
The views expressed are personal