You are on page 1of 12

Dating Mahabharata

- Two Eclipses in Thirteen Days
by Dr S. Balakrishna

Mahabharata war is considered by many to be a historical event. The epic
states that a singularly ominous pair of eclipses occurred in - Thirteen days
- some time before the war. Using modern astronomical software, our
article shows that a number of Thirteen day eclipse pairs were visible in
Kurukshethra. Article suggests some candidate dates for Mahabharata war.


Mahabharata is a
great epic, and is one of the pillars of present day Hinduism. The
Mahabharata story and its moral ethos have had profound influence on
millions over many generations. Mahabharata war is said to have occurred
before the transition of Dwapara Yuga to Kali Yuga. Dating the
Mahabharata war and start of Kaliyuga has been elusive and going on for
many centuries.

Aryabhata, is a famous early astronomer with contributions to science,
whose estimate of p, and the time of moon revolution around the earth are
so accurate, that his works are being extensively researched. Aryabhata

Page 1 of 12
(476-550 AD) stated that Kaliyuga started 3600 years before, when he was
23 years old, making the start as 3102 BC [Aryabhateeya ref-1]). It would
date Mahabharata war to around circa 3130-3140 BCJ.

Surya Siddhanta [Ref 2], a document evolved from roughly same period,
states that sun was 54 degrees away from vernal equinox when Kaliyuga
started on a new moon day, corresponding to February 17/18, 3102 BCJ, at
Ujjain (75deg47minE 23deg 15min N).

Varaha Mihira (circa 560
AD), another famous astronomer, stated that 2526 years before start of
Saka count (either Shalivahana saka starting in 79 AD or Vikrama Saka
starting in 57 BC) [Brihat Samhita Ref-3].

When Saptarishis (ursa major) was near Magha Yudhistira was king 2526
years before Saka time

Presently, traditional Sanatana Dharma followers consider that Kaliyuga
started at 3102 BCJ, when Sri Krishna passed away, and that Mahabharata
war occurred in 3138 BCJ. Millennium year 2000 AD is Kali 5102.

Like Homer's Iliad, another epic poetry from Greece, different scholars
have expressed opinions varying between the story of Mahabharata being
either total fiction or true record of historical facts. It took efforts by
Schliemann and others to show physical archeological evidence of
existence of Troy in present day Turkey, and Homer's poems having
historical relevance.

Bharata has been continuously and relatively densely lived in for thousands
of years and in Northern Bharata the archeological evidence is difficult to
come by because of many 100's of generations of people living in same
area. Hence, it is usual to look for Puranic and Vedic (written and oral
recitation) astronomical evidence to substantiate the time periods. As is true
of all such documents like bible stories, Scandinavian, Chinese, Japanese,
Egyptian and other documented local folklore, the historical truths are
likely to be anywhere between absolute truth to vivid imagination. An

Page 2 of 12
objective analysis can help in determining the likelihood of folklore being a
historical fact or not.

Mahabharata epic story was written by, Vedavyaasa (or Krishna
Dwaipaayana) after the Mahabharata war. Vyaasa is also credited with
codifying the existing branches of Vedas. It is perhaps the longest poem of
its kind of such antiquity. The presently known oldest version of
Mahabharata, based on its style, grammar and other features was probably
written down before the Gupta period. This Mahabharata text does not
refer to any Zodiac's or Raashis (a western concept probably
accommodated in to Jyotishya some time during 300BC to 200AD). The
linguistic style of the oldest version of Mahabharata clearly cannot be the
basis for determining if and when the events of Mahabharata occurred. It
probably may have been rewritten/re-rendered many times as the mode of
transference was by oral traditions as in the case of Vedic chandas prosody.
The known oldest version has nearly 90,000 to 100,000 poems dominantly
with 32 syllables Anushtup chandas, in 18 chapters called Parva's [ref-4
and 5].

The Bhishma Parva and
Udyoga Parva (specific chapters of Mahabharata) provide considerable
astronomical/astrological descriptions and omens as the Mahabharata war
was approaching. It describes a period of draught, with many planetary
positions. Then there is this clear reference to pair of eclipses occurring on
13th day as shown below.

Fourteenth day, Fifteenth day and in past sixteenth day, but I have never
known the Amavasya (New Moon day) to occur on the thirteenth day.
Lunar eclipse followed by solar eclipse on thirteenth day is in a single lunar
month etc...

Page 3 of 12
This reference to Thirteen day eclipse pair appears to be a unique
astronomical observation.

Mahabharata text also refers to retrograde motions of planets prior to war
and provides their location with reference to 27/28 Vedic star locations.
Mahabharata Drona Parva also refers to Jayadhratha's killing during a
dark episode on 13th day of the war, which some consider as another short
solar eclipse.

This document is basically concerned with analysis of all eclipses visible at
Kurukshethra (Location where Mahabharata war took place, north of New
Delhi, Longitude 76 deg 49 min East, Latitude 29 deg 59 Min North) from
3300 BC to about Buddha-Mahavira-Parshvanaatha time of about 700BC.
Analysis of the time between successive eclipses, specifically time between
end of one and beginning of other has been made, with a view to look at
astronomical feasibility of back-to-back eclipses in 13 days, using modern
astronomical computer software.

Another major issue of how did observers of the period define and
determine period between eclipses when no clocks existed, has been

Lunar eclipse occurs when Earth's shadow falls on the Moon. There are
about 150 lunar eclipses per century. Lunar eclipses can occur only at full
moon, and can be either total or partial. Further they can be umbral and or
penumbral. Total lunar eclipses can last up to 2 hours, while partial lunar
eclipses can last up to 4 hours. Any observer on dark face of earth can see
when lunar eclipse when it occurs. During period 3500BC to 700 BC,
nearly 4350 lunar eclipses have probably occurred. A good fraction of these
would have been visible in Kurukshethra [ref-6].

Solar Eclipse occurs when Moon's shadow falls on earth observer. About
240 solar eclipses occur every century. During period 3500BC to 700 BC,
nearly 6960 Solar Eclipses have occurred. Solar can occur only at new
moon. Solar eclipses may be total or annular. Total solar eclipses can last
up to about 8 minutes, and partial solar eclipses can last up to 115minutes.
The shadow of moon has a limited size of few thousand miles falling on
nearly 8000-mile diameter earth. Hence, solar eclipses can be seen only in a

Page 4 of 12
limited range of longitude-latitude where the shadow falls. Elsewhere, even
though sun is visible, eclipse will not be seen.

Eclipse evaluating computational software and its validation in present
Astronomical calculations have been greatly improved since past 30 years,
particularly with considerable amount of trajectory work conducted in
Moon and other scientific projects. High accuracy computer models and
software have been developed. These are validated against databases from
US Naval Observatory's Interactive computer Ephemeris, and Jet
Propulsion Laboratory in California. One such code is LodeStar Pro copy
righted by Wayne C Annala in 1994 [Ref- 7]. The Lodestar Pro was
checked for historical eclipses of 1000-2500 BC from clay tablet records of
Mesopotamia area presently available with British Museum. Wayne
Mitchell has analyzed this data [Ref-8]. Lodestar Pro provided excellent
match with ref-8.

Eclipses at Kurukshethra
During the period of our interest, 3500BC to 700 BC, nearly 4350 Lunar
Eclipses and 6960 solar eclipses have occurred on earth. Of these nearly
673 solar and lunar eclipses occurred in pairs of time gap of about nominal
15 days corresponding to roughly half lunar month. We need to search
amongst these 673 for eclipse pairs visible in Kurukshethra, which
occurred in 'Thirteen' days.

A very detailed scan of all the visible lunar and solar eclipses for every year
from 3300BC to 700 BC was made on the Lodestar software for
Kurukshethra location. These are tabulated and plotted. Maximum eclipse
time gap (end of one eclipse and beginning of next eclipse for naked eye
observers) was found to be about 379 hours while the minimum was about
332 hours. A plot of time gap between back-to-back eclipses versus eclipse
pair number is shown below. (This time corresponds to maximum to
maximum - not end of one to beginning of next as in the future table).

Page 5 of 12
The plot shows that during the period 3300BC to 700 BC, (Julian year
corresponds to zero at 4712 BC- an imaginary date- Our range corresponds
to 1412 Julian year to 4012 Julian Year) nearly 672 pairs of eclipses
occurred on earth, which in principle may have been visible at
Kurukshethra. Amongst these, nearly 32 pairs would be occurring for
period less than 14 days. Many of these were found to be weak penumbral
eclipses of moon, and solar eclipses had such low obscurity as to raise the
issue whether any body could see them. Six pairs of 'thirteen day' eclipses
could be seen unambiguously.

Definition of Day and issue of timing determination
It is easy for us, in present time, to precisely analyze the eclipse times
based on a 24 hour per day time clock. However many thousand years ago,
such a time evaluation would clearly be irrelevant. Hence the count of the
day and time had to be based on clear, natural and unambiguous events
such as sunset to sunset or sunrise to sun rise. Hence in all the analyses,
presented below, the time of relevant sun rise or sun set is indicated such
that the eclipse beginning and end can be evaluated with reference to the
sun rise or sun set. In modern day definition, the period from sunrise to
next sunrise is never 24 hours except on equinox day. On all other days, the
time will be either less than 24 hours (when day light time is shrinking) and
more than 24 hours (when day light time is increasing). For people of
ancient times, sunset-to-sunset or sunrise-to-sunrise would be the logical
definition of a day. Using this definition, it is possible to determine whether
an eclipse pair occurred in 'Thirteen days'.

Page 6 of 12
Kurukshethra eclipses and some planetary retrograde
The table below shows six pairs of eclipses, which can be analyzed further
to determine whether Mahabharata war and events could occur then.
Six eclipse pairs visible at Kurukshethra occurring in less than or near 14 days
Events in red not visible due to sun rise (Lunar) or sun set (Solar)
Year BC Eclipse Julian day Initial con Max End Sunrise Sunset end/start date

Year Julian Initial
Eclipse Max End Sunrise Sunset End/Strt Dt
BC Day Con
3129 Solar 18:53:48 19:48:04 20:38:54 19:22 13d20h20m
3129 Lunar 16:58:50 18:21:36 19:44:21 19:17
2529 Solar Jul 11 03:50:53 04:36:27 05:24:36 05:12 13d20h8m
2529 Lunar Jun 27 03:29:54 05:13:45 06:57:36 05:07
2056 Solar 16:50:19 17:52:24 18:48:02 17:38 13d21h40m
2056 Lunar 16:27:47 18:12:55 19:58:05 17:32
1853 Solar 15:47:28 17:00:02 18:03:38 17:29 13d22h14m
1853 Lunar Jan 13 16:17:56 17:24:16 18:30:37 17:36
1708 Solar Mar 27 04:55:14 05:47:28 06:44:15 06:37 13d20h18m
1708 Lunar Apr 10 03:02:36 04:46:36 06:30:55 06:19
1397 Solar Jul 04 19:00:34 19:36:54 20:11:34 19:21 13d21h30m
1397 Lunar Jul 18 17:41:38 19:34:00 21:26:30 19:23

Location of Kurukshethra 76 deg 49 min East, 29 deg 59 min North

After serious analysis of all the eclipses, six eclipse pairs from 3129 BCJ,
2599 BCJ, 2056 BCJ, 1853 BCJ, 1708 BCJ and 1397 BCJ clearly are the
best candidates for Mahabharata war year from 'thirteen day' eclipse pairs
view point. There are others that have low obscurity for solar eclipse, or
have dominant penumbral lunar eclipse content and hence do not constitute
strong candidates for the Mahabharata war.

One typical eclipse pair of the six is illustrated using Lodestar Pro views of
the relevant sunset/sunrise periods. The light/day transition is clearly shown

Page 7 of 12
in all the eclipse, which would form the only method of determining that
the eclipses occurred in less than fourteen days, which has to be called
thirteen-day eclipses. Planets Sani (Saturn) and Brihaspati (Jupiter), Shukra
(Venus) in retrograde motion are illustrated for period around the eclipse
Solar-Lunar eclipse pair from Julian year 3129 BC

Fourteen days later at same time

Page 8 of 12
Let us now look at how any observer can study these eclipses and conclude
that the pair occurred in 'Thirteen days'. The figures above show the
pictures of day/night sky for a pair of Solar-Lunar eclipses, end of lunar
eclipse being only 13 days and 20 hours before start of a solar eclipse. On
Julian August 11 afternoon, a solar eclipse begins 20 minuets before sunset
and it is still on going at sunset. Fourteen days later (On Julian August 25)
in the evening at sunset a lunar eclipse is already occurring. It clearly
suggests that eclipse started on the 13th day after the previous eclipse!
Obviously the end of lunar and start of solar eclipses were less than 14 days
period, or occurred in 13 days. This could be concluded without the benefit
of modern clocks.

The dates of this eclipse pair are Julian 3129 and Julian month of August.
In ancient Bharata, since the full moon occurred on Proshtapada, the
month would be considered as Bhadrapada. Normally, this is the monsoon
rainy season in North India. However, there are many occasions when
monsoon fails. The epic states that draught like conditions existed. Even
during normal monsoon the sky is occasionally clear for the eclipses to
have been witnessed.

The two planets Jupiter, and Saturn are in motion (vakri) and these do
occur during 3129 JBC as illustrated below. Motion of Angaraka or Mars is

Items in red show retrograde or Vakri motion

Graha (Planet) 3129BCJ Mahabharata text
Brihaspati (Jupiter) U.Ashada/Shravana Shravana-Vishakha
Sani (Saturn) Revati Shravana-Vishakha
Angaraka (Mars) U.Ashada/Shravana Magha
Shukra (Venus) U Phalguni Poorva Phalguni
Ravi (Sun Solar) U Phalguni Rohini

Page 9 of 12
The location of the planets at the time of eclipse pair is shown in table
above. Clearly, only Brihaspati, and Shukra are the only planets near
locations indicated in the Mahabharata text. This date of 3129 BCJ is a
serious candidate date for consideration of Mahabharata war.

Analysis of the Eclipse tables
The first and oldest eclipse pair from 3129 BC is unique. Aryabhata
estimated that Kaliyuga started in 3102 BC. So does Surya Siddhanta.
These fit the Puranic description that Sri Krishna passed away in 3102
BCJ, which is 27 years after the war. Our study confirms that Kaliyuga
could have started in 3102 BCJ.

The second date 2559 BCJ is also unique in that Varaha Mihira stated that
2526 before start of Saka, Yudhishtira was the ruling king. If it Saka was
Vikrama it would make Yudhistira as king in 2583 BCJ, which is before
Mahabharata War. Yudhistira was also king for a short time before war,
before he lost it in a game of dice to Sakuni/Duryodhana. This date is also
an excellent candidate for Mahabharata war. There is another event that
occurs in 2559 BC. While the eclipse pair occurred in lunar month
Shravana, there is another short solar eclipse in Pushya. On 13th day of
Mahabharata war, it is said that Jayadhratha was killed when Sri Krishna
covered the sun for a short time just before the sunset. This event could be
looked upon as a solar eclipse. A study of year 2559 shows that another
solar eclipse did occur in Pushya lunar month (Julian Dec 06, 2559) some
40 days before the winter solstice (Uttara ayana).

The third candidate is eclipse pair from 2056 BCJ. It occurs in
Margashira/pushya months, the lunar eclipse occurring when moon is
between Punarvasu/pushya nakshathra, and would be right in the middle of
war. Hence is not a very serious candidate for Mahabharata war.

The fourth candidate is eclipse pair from 1853 BCJ. It occurs in month of
Magha very near the winter solstice or Uttara Ayana. It is not a very good
candidate for Mahabharata War

The fifth candidate of eclipse pairs occurred in 1708 BCJ. This eclipse pair
occurs in month of Phalguna, just after Uttara Ayana and is a bad

Page 10 of 12
The last candidate of eclipse pair occurs in 1397 in the month of
Bhadrapada. It is a reasonably good candidate for Mahabharata war.
Again, there was no solar eclipse during the period prior to Uttara Ayana.

The aim of this work was to analyze the unique statement that
Mahabharata war took place when an ominous pair of eclipses occurred in
'Thirteen days'. Initially, Mahabharata texts, contemporarily accepted as
most authentic were reviewed and relevant data about Mahabharata and
astronomical planetary observations have been presented.

Firstly, a search of all eclipses during the period 3300 BCJ to 700 BCJ
visible at Kurukshethra, where Mahabharata war took place was made.
Amongst nearly 672 possible eclipse pairs, the time from end of one to
beginning of next eclipse was found to vary between 13.8 days to 15.8
days. Eighteen naked eye visible eclipse pairs with less than 336 hours
(14days) of time gap were found.

The second issue was, what was the definition of a day, and how was the
determination that eclipses occurred in 'thirteen days' made, has been
addressed. Day was taken to be the time between either successive sunrise
or successive sunset. This is particularly important when clocks did not
exist. Using this method, it was easy to demonstrate that observers from
3000 to 5000 years ago could identify accurately a 'Thirteen-day' eclipse
pair when they occurred.

Six pairs amongst these, found to be good candidates for Mahabharata
war, have been illustrated, showing how any observer could conclude that
the eclipse pairs occurred in less than 14 days or in 'thirteen days'. The
locations of Jupiter, Saturn, Mars, Venus, Sun and Moon, during the
eclipses were identified with reference to 27 star locations. The positions of
all these planets during the eclipse pair do not totally agree with
Mahabharata text, but some do agree.

Finally, it is found that two dates suggested by Indian astronomers
Aryabhata, Varaha Mihira are credible dates for Mahabharata war. It
would appear that 3129 BCJ is a first candidate for Mahabharata war
followed by 2559 BCJ. Four other dates viz., 2056 BCJ, 1853 BCJ, 1708
BCJ and 1397 BCJ are other candidates which qualify as 'Thirteen day'

Page 11 of 12
eclipse pairs.

In conclusion, this article has tried to address the basic issue, whether
'Thirteen day' eclipse pairs are astronomically possible. The conclusion is
that such eclipse pairs have occurred and observers could easily identify the
duration using sunset/sunrise transitions. 3129 BCJ and 2559 BCJ dates
appear to be very viable dates for Mahabharata war as are a few others.
This study provides modern scientific support one critical astronomical
statement made in Mahabharata Bhishma Parva that 'Thirteen day' eclipse
pair occurred in Kurukshethra before the Mahabharata war.

February 23, 2002

1. Aryabhateeya by Brahmagupta, S.Shukla,New Delhi, INSA 1976
2. Surya Siddhanta: Translation of an Ancient Indian Astronomical Text.
Translation by Bapudeva, Varanasi, 1860.
3. Varahamihira's Brihat Samhita- M Ramakrishna Bhat, Motilal Banarasidas
Publications, 1981
4. Ramashesha Shastry Bhagavata Mahapurana,
10th skanda, Upodghata (in Kannada script), 1930
5. John Smith web page - Mahabharata Text checked by Bhandarakar
Oriental Research Institute
6. Eric Weisstien, World Of Astronomy web page
7. Wayne Annala, Lodestar Pro Manual, 1994
8. Wayne Mitchell
Ancient Astronomical Observations and Near Eastern Chronology
Journal of Ancient Chronology Forum, Volume3

Related Article
The Date of the Mahabharata War by Pradip Bhattacharya

Image by Ashok Dongre
Charts by the author.


Page 12 of 12