Flat-Earthers do not have a consistent explanation for eclipses, but they all reject the scientific explanations for eclipses. If confronted with the fact it is possible to predict the occurrences of an eclipse, they would claim NASA are using the Saros cycle to predict eclipses. They would point to a web page on NASA that explains the Saros cycle and use it to 'prove' NASA uses the Saros cycle to predict eclipses: https://eclipse.gsfc.nasa.gov/SEsaros/SEsaros.html The Saros is 18 years, 11 days and 8 hours between two eclipses. There will be a similar eclipse a saros after an occurrence of an eclipse. The set of eclipses separated by the same saros is grouped into a saros series. A saros series will end, and a new one will replace it. Several saros series are active at the same time. While it is still possible to predict eclipses using the Saros cycle, it has several limitations: Cannot tell us the time and duration with high accuracy. Not possible to determine the path of a solar eclipse. Cannot predict the first eclipse in a Saros series. I created the following Python script some time ago to show that it is possible for everyone to predict eclipses without relying on NASA and that it is possible to do that using a geometric model, without using the Saros cycle. The script predicts lunar eclipses in the entire 21st century. It works by iterating every hour in the 21st century and calculate the separation between the Sun and the Moon. If the separation is more than 179.1°, then a lunar eclipse occurs. This also proves a lunar eclipse occurs because of Sun-Earth-Moon syzygy, and thus, it also invalidates the flat Earth model where such a thing cannot possibly occur.