Why the Hindu Calendar Is So Accurate

Take the length of a lunar month, measured from one new moon to the next.

The modern figure is 29.530589 days. The Surya Siddhanta, a Sanskrit astronomical text whose surviving form is usually placed somewhere between the fourth and sixth centuries, gives 29.530588. The gap between those two numbers is about seven hundredths of a second.

No telescope was involved. No clock capable of resolving a second existed anywhere on earth. The instruments available were a gnomon casting a shadow, a water vessel that drained at a known rate, and a sighting tube.

That number is the thing worth explaining, and the explanation is more interesting than the number.

Accuracy here comes from patience, not from instruments

The intuition most people bring to this is that a precise result requires a precise instrument. For a single measurement, that is true. For a repeating event, it is not.

If you record a new moon today and another one a month later, your uncertainty about the length of that month is roughly your uncertainty about each observation, which with a shadow and a water clock might be a quarter of an hour. But if you record a new moon today and one that happened four thousand months ago, and divide, the same quarter hour of error is now spread across four thousand months. The error per month has fallen by a factor of four thousand.

This is why the accuracy of Indian, Babylonian and Chinese lunar constants tracks the length of the observational record rather than the quality of the instruments. What those traditions had was a written record maintained continuously for centuries by people whose successors could still read it.

A quarter hour of error divided across a millennium of lunations lands you within a fraction of a second. The arithmetic is not subtle. The institution that keeps records for a thousand years is the hard part, and that is what actually got built.

What a luni-solar calendar is trying to do

The Hindu calendar is luni-solar, and that word describes a genuine problem rather than a design preference.

The month is tied to the moon, which matters because tithis, ekadashis, purnima and amavasya are lunar. The year is tied to the sun, which matters because the harvest is solar. These two do not divide into each other. Twelve lunar months come to about 354 days. The solar year is about 365. The shortfall is eleven days a year.

A purely lunar calendar lets that shortfall run, which is why the Islamic months rotate through the seasons and Ramadan moves. A purely solar calendar throws away the moon, which is what the Gregorian calendar does. The Hindu calendar refuses both and pays for it with machinery.

The machinery is the adhika masa, an extra month inserted when the drift has accumulated to about a month's worth. Seven of them fall in every nineteen years. That ratio is not arbitrary: 235 lunar months come to 6939.69 days and 19 solar years come to 6939.60 days, a difference of two hours. The two cycles very nearly close, and a calendar built on that near-closure stays aligned for a long time.

The Greeks found the same relationship and named it after Meton. The Babylonians were using it earlier. Three traditions arriving at the same nineteen-year cycle independently is not a coincidence, because the cycle is a fact about the sky rather than a convention.

The tithi is a genuinely different idea

Here is where the Hindu calendar stops resembling other calendars.

A tithi is not a day. It is the time the moon takes to gain twelve degrees on the sun. Thirty tithis complete the circle from one new moon to the next.

Because the moon's orbit is elliptical, its speed varies, so a tithi is not a fixed length. It runs from roughly nineteen hours to roughly twenty six. This means a tithi can begin and end within one solar day and be skipped in the calendar, or it can span two sunrises and be counted twice.

Most calendars would treat this as a defect and round it away. The Hindu calendar treats it as the actual unit and adjusts the days around it, which is why a panchanga has to state the exact clock time at which a tithi begins and ends, and why the same festival can fall on different dates in different cities.

That is not imprecision. It is the opposite. A calendar willing to say a date is different in Delhi and Chennai is one that has refused to average away something real.

Where the calendar does drift, and why

There is one place where the calendar and the seasons part company, and it is not a flaw. It is the place where you can see exactly how much the astronomers knew.

The Hindu calendar measures the year against the fixed stars rather than against the seasons. These are not the same year. The sidereal year, star to star, is 365.2564 days. The tropical year, equinox to equinox, is 365.2422 days. The difference is about twenty minutes, caused by the slow wobble of the earth's axis that western astronomy calls precession.

Twenty minutes a year is one day every seventy one years. Over a millennium it is fourteen days.

You can see the result on any calendar. Makar Sankranti marks the sun's entry into Capricorn, and it was originally the winter solstice. The solstice falls on 21 December. Makar Sankranti now falls on 14 January, twenty four days later.

Those twenty four days are precisely the accumulated drift. The Lahiri ayanamsa, the official Indian correction, places the coincidence of the sidereal and tropical zodiacs at 285 CE. From 285 to now is 1741 years, and at one day per 71 years that is 24.7 days of drift. Add that to 21 December and you get the middle of January.

The festival did not move. The definition held, and the season slid out from under it.

Whether that is an error depends on what the calendar is for

There are two ways to read this and only one of them is defensible.

The weak reading is that the ancient astronomers made a mistake and did not know about precession. This is false. Precession is described in Indian astronomical texts, and the ayanamsa is a measured, updated quantity precisely because the drift was known. Nobody was surprised by it.

The stronger reading is that the sidereal year was chosen. For the purposes the calendar was built to serve, which are the positions of the sun and moon against the nakshatras, the sidereal frame is the correct one. A muhurta calculation wants to know where the moon actually is among the stars. Anchoring to the equinox instead would have broken that in order to keep the festivals in season.

The tradition took the astronomy and let the seasons drift. A reformer wanting seasonal stability would have chosen the other way and broken every nakshatra calculation in the process. Both are defensible. What is not defensible is describing the choice as an oversight.

The disagreement that is still open

There is an unresolved question inside all of this, and it is worth stating plainly rather than hiding.

The ayanamsa is the angle between the tropical and sidereal zodiacs, and its value depends on when you say the two coincided. Different schools answer differently. The Lahiri ayanamsa, adopted by the Indian government's calendar reform committee in 1955, is the most widely used. Raman's differs. The Krishnamurti system differs again. The spread between them is a degree or two.

A degree or two sounds small and is not. It can move a sankranti by a day, which moves a festival, which means two panchangas built on different ayanamsas print different dates for the same event in the same city.

This is a real and current disagreement, not a historical one. Anyone who tells you the Hindu calendar produces a single unambiguous answer is describing one school and leaving out the others.

What the record shows

What the record supports: that the Surya Siddhanta's synodic month is accurate to a fraction of a second; that its sidereal year is within three and a half minutes of the modern value; that Aryabhata's length of the sidereal day, 23 hours 56 minutes and 4.1 seconds, differs from the modern figure by about a hundredth of a second; that precession was known and quantified; and that the nineteen year intercalation cycle was in use.

That list deserves to be stated without hedging. For centuries on end, the astronomical constants preserved in Sanskrit stood with the best on earth, and were unmatched in Europe until early modern times. They were achieved by the only method that can achieve them, an unbroken observational program sustained across many centuries, which is the hardest scientific instrument any civilization has ever built. Plenty of cultures produced a brilliant observer. Very few built an institution that could hand his observations to a successor five hundred years later in a form the successor could still use, and India did it while also folding the results, day by day, into the religious life of every household through the panchanga.

The comparison that puts it in proportion is the calendar Europe was using over the same centuries. The Julian year ran eleven minutes long, nobody corrected it, and the error compounded until the equinox was falling ten days from where the calendar said it should. In 1582 the Pope had to delete ten days by decree, and Europe went to bed on the fourth of October and woke up on the fifteenth. The Hindu calendar has never needed a rescue of that kind. Its correction is not a patch applied after the fact but part of the design, the adhika masa absorbing the drift every nineteen years, which is why machinery defined thousands of years ago still lands a festival on the right tithi tonight.

And the unit at its heart is its own. The true tithi, measured from where the moon actually stands rather than from an average, is the reason a panchanga states beginnings and endings to the minute, and the reason it can say honestly that the same festival falls on different days in Delhi and in Chennai. That is not imprecision wearing a religious costume. It is a religious tradition insisting on getting the sky exactly right, for thousands of years, and succeeding.

The same tradition preserved a figure for the speed of light that Europe could not better until 1862, and that story is its own article: the speed of light in the Rigveda.