Quae vero post aequinoctium, vel in ipso aequinoctio, suum plenilunium habet, in hac absque ulla dubietate,
quia primi mensis est, et antiquos pascha celebrare solitos, et nos, ubi Dominica dies advenerit, celebrare debere noscendum est.(But that moon which is full after the equinox, or at the very time of the equinox, belongs to the first month, and on that day, without a doubt, we must understand that the ancients were wont to celebrate the Passover; and that we also ought to keep Easter when the Sunday comes.)
Abbot Ceolfrid of Jarrow Abbey, to Nechtan, King of the Picts (710 AD)
Saint Ceolfrid, above, is lecturing King Nechtan on the correct way of calculating the date of Easter, and emphasizing a deep connection to the Jewish festival of Passover, or Pesach, of which more later.*
As this post goes live, we’re about to see a slightly unusual Easter Sunday, on 20 April 2025. But before I explain what’s unusual about it, some preamble is required.
There are two broad groups of Christian churches: the Orthodox Churches, which trace their practices back to the Patriarchs of the Byzantine Empire; and the Western Churches (including Roman Catholicism and the various kinds of Protestantism), whose practices descend from the Popes of the Western Roman Empire. It’s probably fairly common knowledge that Orthodox holy days occur on different dates from their Western equivalents—Orthodox Christmas, for instance, which is celebrated on 7 January, rather than the date familiar in the West, 25 December. And many people are probably aware that the difference of 13 days is because of a difference in ecclesiastical calendars—the Western Churches use the Gregorian calendar, established in 1582 by Pope Gregory (which is also in widespread civil use); the Orthodox Churches, for ecclesiastical purposes, use the older, Julian calendar, established by Julius Caesar in 45 BC. And those two calendars have drifted apart, over the centuries, because the Julian calendar has observed leap years that have been omitted by the Gregorian calendar. For much more on that process of divergence, see my post “February 30th”.
So. We might naively expect that there would be a fixed 13-day difference between the dates of the Western Easter and its Orthodox equivalent, Pascha. But, as my fifty-year chart of Easter/Pascha Sundays at the head of this post shows, while Pascha is typically later than Easter, both festivals vary in date from year to year (which is why they’re called “moveable feasts”), and sometimes they stray far enough towards each other that the Easter dates of the two traditions coincide—which I’ve circled in red. This happens in 2025, and on fifteen other occasions during that specific fifty-year span.
In this post, I propose to explain why that it is, and also to account for the obvious, if intermittent, periodicity of those red-circled coincidences.
The reason the date of Easter leaps around in the way it does is because, as Ceolfrid summarizes at the head of this post, Easter is tied both to the season of the year and to the phase of the moon. And the reason for that is because, as he wrote, Christian Easter is linked to the Jewish Passover celebration. The Biblical accounts of Jesus’ resurrection after crucifixion make it clear that it occurred a few days after Passover. And by “Passover” the authors of the Gospels meant not the modern Jewish Passover holiday period, but specifically the date of 14 Nisan in the Jewish calendar.
The Jewish calendar is lunisolar, which is to say it takes account both of the seasons, and of the phases of the moon. Because the period from one full moon to the next is an awkward 29.5 days, lunisolar calendars accommodate this by alternating between months of 29 and 30 days. And because twelve lunar cycles, or lunations, add up to only 354 days, an extra month, called an embolismic month, is added from time to time so that the average length of a year works out to match the 365¼-day cycle of the seasons.
Nisan is the first month of spring. Like all Jewish calendar months, it begins and ends (approximately) with a new moon. So 14 Nisan corresponds (approximately) to the first full moon of spring—which is why Ceolfrid is so interested in the timing of the [spring] equinox and the full moon.
Come the spring, early Christians used to just check with their Jewish neighbours for the date of Passover, and celebrate accordingly. But by the time of the Council of Nicaea in 325 BC, there was a feeling (though by no means universal) that Christians should be coming up with their own way of setting a date for Easter. The debate over how exactly that should be done went on, at times venomously, for centuries, and eventually settled on the definition given by Ceolfrid at the head of this post—the first Sunday after the first full moon, on or after the vernal equinox. (Which is to say, the March equinox, but this debate was taking place exclusively in the northern hemisphere.)
So the ecclesiastical calendar that determines the date of Easter is a lunisolar calendar, just like the Jewish one. And the backward drift of the lunar months relative to the seasons, followed by the corrective lurch of an embolismic month, is why the date of Easter jumps around from year to year.
But it’s not an astronomical calendar—it doesn’t reflect the exact timings of the equinox, or of the full moon. Instead, what’s used is a fixed date for the “ecclesiastical equinox”, 21 March; and a set of calculations, referred to as the computus, which determines the date of the “ecclesiastical full moon”. The ecclesiastical full moon of interest is the one that falls within the 29-day period from 21 March to 18 April, inclusive—this calculated full moon is called the Paschal Full Moon†; and, for brevity, I’m going to refer to the prescribed period in which it occurs as the Paschal lunation.
The computus used at the time of Ceolfrid was a fairly simple one based on the fact that 235 lunations are very similar in duration to 19 years. This is the Metonic Cycle, which I described in more detail in my post about Blue Moons. When Pope Gregory introduced the Gregorian calendar in 1582, his astronomers also tweaked the computus, which was beginning to get out of phase with the real lunations—with the result that the Western Paschal Full Moon aligns pretty well (within in a day or so) with the astronomical full moon, whereas the Orthodox Paschal Full Moon, using the earlier computus, has continued to drift, and is now running about four days late.
Time for a chart. Here are the Paschal Full Moons, according to the Western computus, for the years 2001 to 2050, compared to the actual astronomical full moon dates.
I’ve also plotted the boundaries of the Paschal lunation. You can see that the Western computus generates Paschal Full Moons that match the real full moon pretty well, while keeping them constrained within the 29-day period of the Paschal lunation. But notice the years 2019 and 2038‡, in which the astronomical full moon occurs on 21 March, Greenwich time, but the computus places its calculated full moon on 20 March, too early for the Paschal lunation, so that the Paschal Full Moon is delayed until 17 April.
Notice also, the prominent diagonal trend in full moons at three-year intervals. This is a product of the 11-day mismatch, already alluded to, between the duration of 12 lunations and one calendar year, which means that full moons come about 11 days earlier each year.
It is impossible for three 11-day backward jumps to be accommodated within the 29-day Paschal lunation. On the third backward step, the full moon will necessarily fall before the equinox, and the Paschal Full Moon will occur one lunation later. Two backward jumps of 11 days, followed by a forward jump of one lunation minus 11 days, means that, after three years, the Paschal Full Moon will have moved backwards in the calendar by three or four days. (Though, on occasion, the computus produces only a two-day leap, as between 18 April 2019 and 16 April 2022.) This is the diagonal trend that’s so striking in the chart above, and it will be important in explaining the repeating coincidences between Western Easter and Orthodox Pascha.
To the chart above, I can now add the dates of Easter Sunday associated with each Paschal Full Moon:
Easter Sundays can occur as early as the day after the Paschal Full Moon, if that falls on a Saturday; but can be a maximum of seven days after the Paschal Full Moon, if that falls on a Sunday. This means the earliest possible date for Easter is 22 March, if the Paschal Full Moon falls on a 21 March that is also a Saturday. There’s no such event on my chart—the last time it happened was in 1818. The latest possible date for Easter is 25 April, seven days after the end of the Paschal lunation. You can see one of those in 2038, when the Paschal Full Moon falls on Sunday 18 April, with Easter on the following Sunday.
You’ll see I’ve also added diagonal lines marking the “trajectory” of Sunday from year to year. Because 52 weeks add up to one day less than the length of a 365-day calendar year, the days of the week move one day earlier each year, and two days earlier if there has been a leap day. This extra regression every four years accounts for the kinks in the “Sunday trajectory” lines.
The movement of Sunday to earlier dates, year on year, almost matches the three-year regression of the Paschal Full Moon I’ve been discussing, but is slightly faster. So the Paschal Full Moon and Easter Sunday slowly converge on each other. In 2030, for example, the Paschal Full Moon falls on Wednesday 17 April, with Easter on Sunday 21 April. After three years, in 2033, the Paschal Full Moon fall three days earlier, on 14 April, but an intervening leap year means that Sunday comes four days earlier, so that the Paschal Full Moon is now on a Thursday. Another leap year in 2036, and the Paschal Full Moon is on a Friday. No leap year before 2039 … still on a Friday. But then it’s on a Saturday in 2042; and in 2045 the convergence is complete—the Paschal Full Moon falls in a Sunday, and Easter leaps away to the following Sunday. This slow convergence will also be relevant to the runs of Easter/Pascha coincidences.
I can plot the Paschal Moons and Pascha dates from the Orthodox computus on my same chart, if I mark Julian dates on the right side:
The earliest and latest dates for Paschal Full Moons, and the latest date for Pascha, are exactly the same in the Julian calendar, just time-shifted by 13 days relative to the Gregorian dates. And notice how poorly the Orthodox computus tracks the real astronomical full moon, occurring four or five days later. But we can see exactly the same three-year recurring patterns in Paschal Full Moon and Pascha dates as were evident in the Western computus, and exactly the same convergence between Paschal Full Moon and Pascha Sunday.
So, finally, I can plot both Easter and Pascha on the same chart:
I’ve removed the astronomical full moons, but retained the other details. The Easter/Pascha coincidences now appear as red/blue rosettes, where the two symbols are superimposed.
The earliest possible Easter/Pascha coincidence is if Easter occurs on 22 March (Julian), 4 April (Gregorian), which is the earliest possible Pascha date. One of those occurred in 2010. The latest possible coincidence occurs if Pascha falls on 12 April (Julian), 25 April (Gregorian), which is the last possible Easter date—and we have one of those on the chart, too, in 2038.
Within this three-week period when Easter/Pascha coincidences are possible (4 April to 25 April), there’s another rule in operation. Because the Orthodox Paschal Full Moon always falls four or five days after the astronomical full moon, and the Western Paschal Full Moon tracks the astronomical full moon pretty well, we’ll only ever get a coincidence if Western Paschal Full Moon falls early in the week (Sunday, Monday, Tuesday), leaving enough “calendar room” for the Orthodox Paschal Full Moon to occur later the same week.
But in years like 2042, for example, when the Western Paschal Full Moon falls on Saturday, 5 April, with Easter the following day, the Orthodox Paschal Full Moon is pushed into the following week, and Pascha occurs the week after Easter.
So there are three possibilities:
- Easter occurs before 4 April, and Pascha is four or five weeks later, after the next astronomical full moon
- Easter falls between 4 and 25 April, but late in the week, and Pascha follows a week later
- Easter falls between 4 and 25 April, early in the week, and there’s an Easter/Pascha coincidence
Finally, there’s that three-year cycle after which Easter and Pascha return to almost the same position in the calendar, but three days earlier, and Sunday migrates three or four days earlier in the same time period. So once an Easter/Pascha coincidence occurs, it tends to return three years later on a slightly earlier date. You can see that the coincidence we’re about to have on 20 April 2025 will return on 16 April 2028, 13 April 2031, 9 April 2034, and 5 April 2037—after which Easter occurs too early, on 1 April, and Pascha leaps forward to 6 May.
So that’s one way these runs of Easter/Pascha coincidences can end—when the repeating sequences creep so early that the Easter falls before 4 April.
The other way these runs end is if the gap between Western Paschal Full Moon and Easter Sunday dwindles below four days, so that there’s no room to fit an Orthodox Paschal Full Moon into the gap. That happens with the run of Easter/Pascha coincidences in 2011, 2014 and 2017. It starts with the Western Paschal Full Moon on a Sunday, affording plenty of room for the Orthodox Paschal Full Moon to fit in five days later. By 2017, Western Paschal Full Moon is on a Tuesday and the Orthodox Paschal Full Moon, four days later, is on Saturday. And in 2020 the sequence breaks, because Western Paschal Full Moon is on a Wednesday, and Orthodox Paschal Full Moon is on Easter Sunday, pushing Pascha a week later.
So that’s the story for this year’s Easter/Pascha coincidence. Mark your diaries for the next one, on 16 April 2028.
* Ceolfrid’s letter comes down to us because it’s quoted by Saint (“The Venerable”) Bede in his book Historia Ecclesiastica Gentis Anglorum (“Ecclesiastical History of the English People“), Book V, Chapter XXI (731 AD). Bede was a pupil of Ceolfrid’s. Ceolfrid uses the Latin word Pascha indiscriminately to refer to the Jewish Passover and Christian Easter. The translation I’ve used, by A.M. Sellar (1907), carefully separates the two meanings.
† Paschal means “pertaining to Easter”. It derives from Hebrew Pesach, “Passover”, which became Pascha in Latin and Greek. The Greek version gave its name to the Orthodox Easter festival; the Latin to the names for Easter in various Romance languages: Pasqua in Spanish, Pâques in French, for instance.
Bede (see note above) wrote that the English name, Easter, came from a pagan goddess of the Spring, Eostre, who had given her name to the lunar month in which Christians now celebrated Easter. But in Stations of the Sun, Ronald Hutton suggests that it’s just as likely that Anglo-Saxon Estor-monath meant something like “month of opening” or “month of beginning” (Spring being considered the start of the year), and that Bede’s otherwise unattested Eostre either never existed, or was the name of a dawn-goddess, like Greek Eos, who was unrelated to springtime in general and the name of the month in particular.
‡ Notice that 2019 and 2038 are 19 years apart—this is the Metonic Cycle, on which the computus is based. If you scan across my chart, you’ll see how full moons fall on, or near, the same date at 19-year intervals.