New Year in monsoon

While the Western world celebrates the New Year on the first of January, in Bangladesh, Burma, India and Nepal it falls in mid-April. Struck by the near coincidence of these New Years dates in the traditional calendars in the region, an observer inevitably begins to wonder why this has come about. In fact, if we look at Southasian calendars from centuries ago, we find that they have become increasingly out-of-sync with both the Gregorian calendar and the local seasons, by nearly a day every 60 years. In this way, the region's calendars seem to be going forward in time, like a clock ticking slightly more quickly than it should.

A clue to this mystery can be found if we look slightly beyond our borders. For millennia, Iran has celebrated the New Year, or navroz, on 21 March, because the Zoroastrians have a fixed calendar. The Parsi community in India also observes Navroz on the same day. This date should ring a bell in the mind of any schoolchild: it is one of the four most significant days of the year. As the Earth goes around the sun, every year we get two days with equal night and day – the equinoxes. Likewise, there are two 'extreme' days, one with the longest and another with the shortest days – the solstices. These four days are spaced roughly three months apart, and they mark the beginnings of the four seasons. The equinoxes appear in spring and autumn, and 21 March has long been marked as the beginning of spring. Traditionally, many cultures have also used the spring equinox to mark the New Year. Like the Persians, the Romans also used March as their first month. This is the reason why there is a series of English-language months with telltale, though incongruous, numerical names: for instance, the sept- in September means seventh, despite the fact that it is now the ninth month. Such a discrepancy is cleared up if the year begins with March, however.

If, since about 500 BC, it has been customary to celebrate the New Year in the spring, then could it be that Southasian calendars also used this equinox as the beginning of a year, at some point in the distant past? Careful readers would have already calculated that the difference between mid-April and 21 March is approximately 25 days. And, if our previous understanding holds true – that Southasian calendars are moving faster than the Gregorian calendar by a day every sixty years – then we can quickly estimate the era when our calendars began to go more quickly. We simply have to multiply 25 by 60, which gives us approximately 1500 years. That takes us back to sometime around 500 AD. What could have happened at that time, to push the region's calendars into such a slide?

Marking time
Before we dig into our history, it would do well first to understand why there should be a slide of this nature at all. Let us begin by looking at the problems that any calendar-maker faces. The first problem is the undeniable fact that a year, which we define by the length of time our Earth takes to orbit the sun, does not comprise of a whole number of days. A year is 365 days plus some extra hours, minutes and seconds. But a calendar, on the other hand, is supposed to mark our days, which we define by sunrise and sunsets.

The problem here is that there are two distinct definitions of a 'year'. First, we can call the time between two spring equinoxes our year, in which case the 'calendar' year will correspond to the rhythm of the seasons. Let us call this our 'seasonal' year, though astronomers call it a 'tropical' year. But second, as the Earth goes around the sun, from the point of view of the Earth the sun appears to move in the sky, and the constellations it moves past are the zodiac signs. For instance, if I move past a friend standing in front of me, then from my point of view, she would appear to move against the backdrop of distant landmarks, such as buildings and roads. The same thing happens to the Earth as it goes around the sun: the sun takes a year to come back to a certain point in the sky with respect to the stars. We can call this a 'stellar' year, also known as a 'sidereal' year in astronomical jargon.

Now let us add one bit of complexity to this picture. If, in addition to moving around my friend, I am constantly turning a bit myself, then my point of view at the end of my orbit would be different than it would be under the previous example. As a result, from my changed point of view, my friend would appear differently against the backdrop. In other words, she may not appear against the same reference point in the backdrop, even though I have come back to same point in my orbit.

This is exactly what happens in the case of the Earth. The planet's axis rotates very slowly, taking 26,000 years to complete a cycle. This motion is analogous to the slow turning of the axis of a toy top spinning on the ground; in the case of the Earth's axis, it is caused by the gravity of the sun and the moon. We know that this axis points roughly to the North Star (the star Polaris) in the sky. But this slow motion – called the 'precession' of the Earth's axis – has led the axis to point in different directions as the millennia have progressed, and it will not point towards Polaris in a few thousand years' time.

This is very slow motion, almost imperceptible, but it has a tangible effect on our calendars. For one, it makes the 'stellar' year different from the 'seasonal' year, because by the time the Earth comes back to the same 'seasonal' point in its orbit, the sun has not returned to the same point in its apparent path in the sky. This phenomenon was discovered by a Greek astronomer named Hipparchus around 150 BC. The fundamental problem he confronted was this: How does one track the advance of the sun against the backdrop of the stars, if one cannot see stars during the daytime? His solution was a stroke of genius. At the time of lunar eclipses, the moon is situated exactly opposite to the sun with respect to the Earth, so that the Earth's shadow falls on the moon. If there happens to be a lunar eclipse near the spring equinox, then by observing the position of the dark moon in the sky, we can work backward and find out the exact position of the sun in the sky on that day. Then we just wait for another lunar eclipse near a spring equinox day after a few years, and observe whether the sun's position has changed – and, if so, by how much.

Hipparchus observed the lunar eclipse on 21 April 146 BC and again eleven years later, on 21 March 135 BC. He then was able to figure out how much the sun had moved with reference to a particular star per year. From that, Hipparchus was able to deduce the difference between the seasonal and stellar year. Since then, astronomers have measured this difference extremely accurately. They have found that a seasonal year is 365.2422 days long, compared to 365.2564 days for the stellar year. This difference works out to approximately twenty minutes.

During the medieval era, Indian astronomers were aware of this difference. During the first half of the first millennium, a new system of astronomy and calendar-making was being established. Particularly around 500-700 AD, a vigorous debate was raging regarding the phenomenon of precession. Some astronomers were already citing it, albeit cautiously, while others were categorically denying its existence. In addition, some other astronomers were arguing that the Earth's axis did move, but that it swung like a pendulum; this would mean that the difference between a seasonal and a stellar year would first increase, then decrease, come to zero, and then increase in the other direction. Therefore, this group suggested, there would be no damage to the calendars in the long run, if one waited long enough.

It is not clear exactly why, but the medieval Southasian astronomers adopted a value close to 365.2588 days as the length of a calendar year, a value closer to the stellar year (though still off by a few minutes) than to the seasonal year. According to this calendar, the date of the spring equinox is not fixed, but rather varies from year to year, shifting slowly over time. At some point, this length came to be taken as the standard, although latter-day astronomers kept noticing the discrepancy that was accumulating over centuries, and they continued to prescribe various 'corrections'. But the main framework of the calendar was tied to the stars, not to the seasons, and that too with a bit of error. So the Southasian calendar (which was eventually to spawn variations throughout the region) began to step out of rhythm with the seasons. This ultimately brings us back to our original figure: this calendar was faster than the seasonal year by roughly a day every 60 years.

Over the centuries, small errors in a calendar can add up. The New Year becomes more and more out of sync with the seasons, and so do the months. Festivals related to seasons are no longer attached to one another. For example, the Buddha Purnima, the birthday of the Buddha, was supposed to be on a full moon during the first month of the year, Vaisakh (Baisakh in Nepal and Boishakh in Bangladesh), according to Ashvaghosha's book Buddha-charita, written in the beginning of the first millennium. At that time, the New Year was celebrated on the day of the spring equinox, and Buddha Purnima would have been on the following full moon, meaning sometime before 20 April. At least, this would have been the case had our calendar systems adhered to the seasons and always begun on the spring equinox. These days, however, we celebrate this holiday in the middle of May – a shift of almost a month. As this calendar continues to slide, after a couple more millennia, Southasians could be celebrating Buddha Purnima at the peak of the summer monsoon!

Difficult reform
In India, a government committee was set up in 1955 to look into this matter, with the famous physicist Meghnad Saha as its chair. The committee eventually suggested that a 'civil' calendar be used, which would begin on 22 March every year. Of course, in instituting this new system Saha and the rest did not want to shock the people by changing traditional festival days. Thus, rather than scrapping days from the calendar, it suggested that one could live with the errors – but also ensured that there be no further increase in those errors. Since the calendar had already shifted by 23 days, the committee suggested that the first month should be called Chaitra rather than Vaisakh. This would mean a difference from the traditional calendar (or, as it stood in the 1950s) by about a week. Thereafter, it proposed ways similar to the Gregorian calendar to keep the Indian calendar pegged to the seasons. Although this 'Indian National Calendar' was initiated in 1957, it is hardly used; these days, it only appears in official documents, having failed to percolate into other spheres of life.

At the moment, nearly all of Southasia's traditional calendars need reform. They are either solar or 'lunisolar' (a mixture of lunar months and solar years), and are all pegged to the wrong notion of the length of a solar year. The solar calendars (such as the Nepali calendar) begin in mid-April, and are divided into twelve months of either 30 or 31 days. The lunisolar calendars (used in some states in India, such as Maharashtra and Andhra Pradesh, as well as by Buddhist and Tibetan calendars) are a bit complicated; the months here are as long as the lunar month, which is roughly 29.5 days long. From time to time, this requires some adjustment, by adding an extra month (a leap month) in some years so that the calendar follows the solar calendar. But they also follow the 'wrong' notion of the length of a solar year, and are thus in need of reform. The New Year in these calendars is fixed on the new moon day before the New Year according to the solar calendar, and therefore falls within a month of that in solar calendars.

Indeed, only the Islamic calendar works well in its current format, because it does not stick to seasons. Instead, it completely disregards the sun, and depends solely on lunar phases. Since a lunar month is approximately 29.5 days long, the Islamic calendar has 354 days with twelve lunar months, making it shorter than the Gregorian calendar year by about 11 days. As a result, Islamic festivals continue to be discordant with the Gregorian calendar, but since they are not hinged to any particular season, one does not encounter related difficulty. For example, the festival of Ramadan in 2005 began on 4 October, and this year will begin on 1 September.

In Bangladesh, a revised calendar, recommended by the Bangla Academy in 1966, suggested that the New Year be celebrated on 14 April every year. The idea again was that one could live with accumulated errors over the centuries, but could also take measures to prevent any further shifting of the calendar. Interestingly, in 2008 Bangladeshis celebrated their New Year on 14 April, while those in West Bengal celebrated it the following day, according to the traditional, 'incorrect' calendar.

Like in West Bengal, traditional calendars in Southasian countries that still tack festivals to seasons continue their slowdown. The worry here, however, is that a significant slide could eventually lead these festivals to lose both their meaning and relevance. While this might not be a calamity, since old habits die hard, one might as well begin the reform process sooner than later. Reform, of the kind that has been undertaken in India and Bangladesh, is sorely needed, but it must be ensured that this overhaul not be confined merely to official use.

~ Biman Nath is an astronomer at the Raman Research institute, Bangalore. His first novel, Nothing is Blue, will be published by Harper Collins (India) in  2008.