The Saros Cycle: Why the Same Eclipse Repeats Every 18 Years

The Moon does not orbit Earth in a perfect circle, and its orbital plane is tilted about 5 degrees against Earth's path around the Sun. That tilt is why a solar eclipse does not happen every new moon: most of the time the Moon's shadow misses Earth entirely, passing above or below. An eclipse only happens when the Moon crosses the plane of Earth's orbit, a point called a node, at almost exactly the same moment it is new. That coincidence is rare. The Saros cycle is the period after which it repeats.

Three lunar cycles have to synchronise for the geometry to recur. The synodic month, the time from one new moon to the next, is 29.53 days. The draconic month, the time for the Moon to return to the same node, is 27.21 days. The anomalistic month, the time for the Moon to return to the same point in its elliptical orbit, which controls how large the Moon appears, is 27.55 days. After 223 synodic months, 242 draconic months, and 239 anomalistic months, all three cycles land within hours of each other. That convergence takes 6,585.32 days: 18 years, 11 days, and roughly 8 hours.


The 0.32 of a day left over, about 8 hours, is not a rounding error to be ignored. Because Earth rotates once every 24 hours, that extra third of a day shifts the eclipse path approximately 120 degrees westward on the globe. A total solar eclipse that sweeps across India will, one Saros later, sweep across a region of the Pacific or South America. Three Saros cycles, 54 years and 34 days, a period called the Exeligmos, bring the path back to roughly the same longitudes, though still shifted in latitude.


This is why eclipse chasers who follow a single Saros series across decades see the same eclipse family slowly migrate across the planet. The eclipse of 11 August 1999, which crossed Europe and western Asia, belongs to Saros 145. Its predecessor in that series ran on 31 July 1981. Its successor ran on 21 August 2017, the Great American Eclipse, tracked live by NASA across fourteen US states. The next in the series falls on 2 September 2035, crossing Japan and parts of the Pacific.


The Babylonians had identified the Saros cycle by at least the seventh century BCE, working from clay tablet records of eclipses accumulated over generations. They called the underlying period by a name that translates roughly as the interval of repetition. They could not explain the orbital mechanics, that framework came much later, but they could count. Enough records of past eclipses, laid side by side, revealed that the gap between related events was consistently 6,585 days. That was enough to predict the next one.

Greek astronomers, including Hipparchus in the second century BCE, refined the underlying lunar periods to a precision that would not be improved significantly until the telescope era. The Antikythera mechanism, a geared bronze device recovered from a Roman-era shipwreck off the Greek island of Antikythera, encoded the Saros cycle mechanically, using a 223-tooth gear to track the synodic months. It is the oldest known analogue computer, and eclipse prediction was one of its primary functions.

NASA and the International Astronomical Union catalogue every eclipse by its Saros series number. Each series begins when the Moon's shadow first clips the polar regions of Earth at a shallow angle, then matures over centuries as successive eclipses move toward the equator, then fades as the shadow drifts off the other pole. A typical Saros series runs for 1,200 to 1,500 years and produces 70 to 80 eclipses. Saros 145, the series that produced the 2017 American eclipse, began in 1639 and will end around 3009.

ISRO's solar observatory Aditya-L1, launched from Sriharikota in September 2023 and positioned at the Sun-Earth Lagrange point L1, does not need an eclipse to observe the Sun's corona, it carries instruments designed to do that continuously. But the Saros cycle matters to solar physicists precisely because it allows them to compare coronal observations from the same eclipse geometry across decades, controlling for the geometry while measuring how the Sun itself changes. The cycle is not just a curiosity; it is a repeatable experimental condition.

The Saros cycle repeats the geometry of an eclipse with high fidelity, the size of the Moon's shadow on Earth, the duration of totality, the path width, but it does not freeze everything. The corona the Sun displays during totality changes with the solar cycle, which runs on an independent 11-year rhythm. An eclipse in the same Saros series can show a corona spiked with long streamers near solar maximum or a quieter, more symmetric halo near solar minimum. The clock repeats the stage; the Sun writes a different scene on it each time.

The Moon is also very slowly receding from Earth, at roughly 3.8 centimetres per year. Over the lifetime of a Saros series spanning a millennium, that recession is small enough to leave the eclipse geometry largely intact. Over geological time it is not: a billion years ago the Moon was close enough that total solar eclipses were longer and more frequent. A billion years from now it will be far enough that annular eclipses, where the Moon is too small to cover the Sun completely, will be the only kind possible. The Saros cycle will still exist then. It will just be counting a different kind of event.

The three lunar periods that produce the Saros do not care about human calendars, national borders, or the names astronomers give to things. They converge on 6,585.32 days because orbital mechanics has no other option. Every eclipse that has ever been recorded, and every one that will happen while the Moon still covers the Sun, belongs to a series that started before any civilisation wrote it down and will end long after any now living is gone.