
Sunset this evening—Saturday, 12 April, 2025—marks the beginning of the Jewish feast of Passover; sunset is also, for both Western Catholics and Eastern Christians, exactly one week from the beginning of the Easter Vigil or Paschal Vigil next Saturday evening.1
Why exactly are Passover and Easter, so to speak, in the same orbit, even if they don’t always coincide? Why does the Eastern Orthodox celebration of Christ’s resurrection—Pascha, as Easter is also called, not only by the Orthodox, but also by Catholics speaking nearly all languages other than English and German—usually diverge from the Western celebration by one to five weeks?
The short answer is … I mean, there is no short answer, it’s complicated! As I joked six years ago on Facebook:
In reality, this was an oversimplification, as I will explain. How does it all work? Let’s take it from the top!2
Our story starts in Egypt, some 13–14 centuries before Jesus was born. According to the book of Exodus, God freed the Hebrews from slavery in the spring, marking what would afterwards be reckoned the first month of the year, the lunar month of Aviv or Nisan. The Passover, commemorating the last of the ten plagues and the liberation from Egypt, was to be held at the time of the full moon, about halfway through the 29-day lunar cycle: i.e., the 14th of Nisan.3

Because the phases of the moon don’t perfectly line up with the tropical or seasonal year, 12 lunar months are about 354 days—about 11 days short of a full tropical year.4 Eleven days is too few for an annual 13th month, but it quickly adds up. Before long, without adjustments, the first month of Aviv or Nisan would drift back from spring into winter.5 Thus, like other ancient Near Eastern cultures, the Hebrews kept their calendar in line with agricultural practice by observing periodic leap years with an extra intercalated month, as determined by religious authorities—not quite every other year, but more often than every three years.
The Last Supper and Jesus’ crucifixion occurred in close proximity to the Passover.6 At the Last Supper, moreover, Jesus overtly linked his suffering and death to the Passover, and Christians have always understood his death and resurrection to have Passover significance for us and for humanity. In fact, the term “Paschal Mystery”—referring both to Jesus’ suffering, death, and resurrection and also to the Eucharistic liturgy which commemorates and represents the sacrifice of the Cross—is derived via Greek and Latin “Pascha” from the Hebrew term Pesach or Passover. Likewise, the name for Easter in virtually all languages (again, the notable exceptions being English and German) is cognate to Pascha/Passover.7 (The silly anti-Christian canard about Easter deriving from a pagan holiday is linguistic as well as historical nonsense!)
The Christian celebration of Jesus’ suffering, death, and resurrection is thus closely linked in its origins to the Jewish Passover, which, again, is tied to 14 Nisan, in principle always falling on the evening of the full moon in close proximity to the vernal equinox.8 How closely linked, though? Let’s see…
One very ancient Christian practice, called Quartodecimanism or “Fourteenthism,” observed the Christian Paschal Mystery on 14 Nisan, the original date of Passover, as calculated each year by Jewish authorities. The Paschal celebration could thus fall on any day of the week. However, another ancient custom associating Jesus’ resurrection with the first day of the week—Sunday—prevailed, leading to the practice of celebrating Pascha on the Sunday after Passover.
The strict tie to the observance of Passover soon became unsatisfactory, though. Some degree of antisemitic sentiment, along with theological resistance to Judaizing tendencies in early Christianity, was surely involved here (as well as in the Church’s ultimate rejection of Quartodecimanism). Even apart from this, though, the date of Passover at this point still depended on real-time calculations and decisions of Jewish authorities—leaving Christians waiting upon the rabbis to tell them when they would celebrate the holiest day of the Christian year.
Thus, the Church eventually settled on their own formula: The Christian Pascha would fall on the Sunday after the first spring full moon, i.e., the first full moon on or after the vernal equinox, whether or not this full moon was reckoned as Passover by Jewish authorities. The vernal equinox was reckoned as March 21st, which was accurate enough at the time—but there were a couple of complications, one of which would become increasingly unavoidable over the centuries…
The Julian calendar used at the time throughout the Roman empire was sophisticated enough to have a leap year with an extra day every four years, meaning that the average calendar year at the time was 365.25 days. This is very close to a true tropical or seasonal year—but Julian leap years were a very slight overcompensation, adding an extra day every 128 years, or three days every 400 years. As a result, by the 16th century, “March 21st” fell about 10 days after the true vernal equinox.

This led, in the West, to the Gregorian reform of the calendar. In 1582, on the authority of Pope Gregory XIII, Catholic countries dropped ten days from the month of October (a month chosen to avoid conflict with important Church feasts). Thus, 4 October 1582 was directly followed by 15 October—and the following year 21 March again basically coincided with the true vernal equinox.9 Additionally, based on more advanced astronomical measurements at the time, the cycle of leap years was adjusted to skip leap years in three century years out of four: that is, in every century year except those divisible by 40010—an incredibly accurate schema.11
Yay! The pope fixed everything, right? Well, not exactly…
In the Protestant world, where papal leadership was not exactly highly thought of, the superiority of the Gregorian calendar took awhile to gain acceptance. Most Protestant countries were on board by 1700, but England, Wales, and the American colonies held onto the Julian calendar until 1752. (Among other brain-melting implications, this means that the birth of George Washington in 173212 was originally recorded in Julian reckoning as the 11th of February,13 a date already reckoned in Catholic countries, and today reckoned pretty much everywhere, as the 22nd. Also, had England and the colonies delayed a mere quarter century longer before ditching the Julian calendar, the 4th of July would originally have been recorded as the 23rd of June!)
In Orthodox countries (where the pope’s status was even lower than among Protestants!) the Gregorian calendar took far longer to take hold, even for civil purposes. How much longer? Russia accepted the updated calendar only in 1918, and Greece held onto the Julian calendar until little more than a century ago, 1923.
That’s for civil purposes! For the reckoning of Pascha, the Orthodox world still reckons the vernal equinox on Julian “21 March,” or what for all other purposes we now call 3 April—a discrepancy of 13 days. That disconnect was only 10 days back in 1582, but remember how the Gregorian calendar skips three leap years every 400 years to stay accurate? There you go. The Julian calendar is slowly, inexorably sliding further from tropical or seasonal reality, and the Orthodox Pascha is sliding with it.
Specifically, under the Gregorian calendar, the latest possible date of Easter14 is currently 25 April.15 But the Orthodox Pascha can fall 13 days later, on 8 May—almost 50 days after the actual vernal equinox—and, again, the discrepancy is only growing. Right now, Easter and Pascha coincide roughly 30 percent of the time. By the year 3000, though, the Julian equinox will be reckoned on what we call 10 April—a 20-day discrepancy!—and Orthodox Pascha will potentially fall as late as 15 May!
Whew! That’s it, right? Not quite—believe it or not, the complications don’t end there!
In 2022, I noticed that the first spring full moon fell on Friday, 15 April—well after both 21 March and 3 April. In theory, this should have made 2022, like 2025, a year of convergent Easter/Pascha dates—but while Easter in the West did follow on Sunday, 17 April, Pascha in the Orthodox world was a week later, on April 26. How is that possible? It turns out that, just as neither Catholics nor Orthodox Christians follow the exact astronomical date of the equinox, neither follows the exact astronomical date of the full moon either! Nor, for that matter, do the Jews for Passover purposes.
Instead, the phases of the moon are, for the liturgical purposes of Jews, Catholics, and Orthodox Christians, determined by a 19-year table called the Metonic cycle, devised by in the 5th century BC by a Greek mathematician and astronomer named Meton.16 Meton calculated that while 12 lunar months is less than a topical year, and 25 months is more than two years … and 37 months is a bit less than three years … 19 tropical years works out to almost exactly 235 lunar months, bringing the phases of the moon and the seasonal state of the earth into nearly perfect sync again. He thus devised a 19-year table with seven 13-month leap years. This basic calculus remains the foundation for the calculation of the 14th of Nisan in the Jewish calendar and the Paschal full moon in both the Julian and Gregorian calendars.
The Metonic cycle is actually more precise than the Julian calendar, though not as precise as the Gregorian calendar. Like the Julian calendar, it’s a tiny bit too long, adding almost two hours and five minutes every 19 years and an extra day about every 200 years. As a result, none of the three feast days tracks exactly with the true astronomical full moon.17 In particular, while the Gregorian reform included resetting the liturgical full moon astronomically at the time, and the math was, I believe, adjusted a bit, it’s still a tabular approximation and it continues to drift a bit. What happened in 2022, then, was this: The astronomical full moon was on Friday, 15 April, while the Catholic ecclesiastical full moon was, I think, the next day, Saturday the 16th—close enough to make no difference in this case. The Orthodox ecclesiastical full moon, though, was a little later, so Pascha was delayed till the following Sunday.18
The upshot is that the Catholic date of Easter is good to go for many millennia, while the Orthodox Pascha (and, to a lesser extent, the Jewish Passover) are slowly sliding toward summer. Why does this matter and what might be done about it? More to come!
From my last homily:
This year, the Jubilee Year 2025, is one of those happy years in which virtually all Christians—Catholics, Protestants, and Eastern Orthodox and other Eastern churches, including those Eastern Churches who don’t follow the same calendar we do—will celebrate Jesus’ resurrection on the same day, April 20.
Easter won’t align this year with another related celebration: the Jewish Passover, the holy day for which Jesus went to Jerusalem to celebrate the Last Supper with his disciples before being arrested and crucified. Even here, there’s a point of contact: The eight-day Passover season begins this year on the evening of April 12th, but it ends on the evening of April 20. This reminds us that Jesus’ death and resurrection are our Passover, and as the sacrifice of the Passover lamb in Egypt was linked to the liberation of the Israelites from slavery to the Egyptians, so Jesus’ sacrifice on the cross and his resurrection at Passover time sets us free from slavery to sin and death.
This is where I usually say I’m not an expert on whatever topic I’m trying to explain. And it’s true that I’m not a liturgical expert, particularly as regards Jewish ritual and tradition—though on this topic I believe I write with somewhat deeper knowledge than many topics I try to speak knowledgeably about. Still, I make mistakes, and this is a massive and complicated subject, and some of what I’ve written here reflects research I did years ago. As always, I welcome additional insights and corrections about any topics on which I may have misspoken.
Lunar months begin, of course, with the new moon. One week, 7 days, is roughly a quarter of a lunar cycle: one week to half moon and two weeks (a fortnight or 14 days) to full moon. (It’s actually a little longer than that, so lunar calendars typically alternate between 29 and 30 days.)
In the ancient world it was common for calendars to start in the spring, when things began happening in agricultural societies. For example, the old Roman calendar, also a lunar calendar, was originally just 10 months, from March to December—304 days, with a “monthless” winter gap of about 61 days.
The Gregorian calendar we now use, like its Julian predecessor, accounts for those 11 days by making the 12 months longer than a true lunar month: 30 to 31 days, as opposed to 29 or 30. The fact that February is the short month (28 days except during leap years) is a relic of the days when March was the first month of the year and winter was just a hole in the calendar; see note 3 above.)
For example, the Islamic calendar is exactly 12 lunar months with no adjustments to keep in line with the seasons. This means that the Islamic calendar is about 11 days shorter than the tropical year, so every lunar month, including the holy month of Ramadan, can and eventually does fall in any season of the year. For example, it takes about eight years for Ramadan to move from spring to winter, another eight years to move to autumn, and over about 33 years it cycles through the entire year.
There appears to be a difference in interpretation between the Synoptic Gospels, which present the Last Supper as a Passover meal, and the Gospel of John, which seems to present Jesus dying on the day of Passover. There are different theories and ways of trying to reconcile these accounts, but at any rate Jesus died in close proximity to Passover.
In Romance languages like Italian and Spanish, if it is necessary to differentiate the Christian “Passover” from the Jewish one, more specific language may be used: for example, in Italian, Pasqua ebraica or “Hebrew Passover” vs. Pasqua di resurrezione for Christian Passover/Easter.
Perhaps gratuitously complicating the topic, “Passover” was originally 14 Nisan, the day when the Passover lambs were sacrificed. When the sun went down and the Passover meal was eaten, that was reckoned as 15 Nisan, the first day of the weeklong festival of Unleavened Bread. Over time, though, terminology and praxis drifted in a number of ways:
Passover day and the festival of Unleavened Bread, originally distinct, were conflated into a unified “Passover season.” (We see this in the New Testament where Luke writes that “the feast of Unleavened Bread drew near, which is called the Passover.”
The emphasis on “Passover” shifted from the slaughter of the lambs to the eating of the Passover meal; thus 15 Nisan, not 14 Nisan, became “Passover day.”
In the New Testament, the day when the lambs were sacrificed—14 Nisan—is sometimes called “the first day of Unleavened Bread” (cf. Mark 14:12)!
Ish. Tropically or seasonally speaking, the real vernal equinox is best approximated as the 20th of March, not the 21st—though because of leap years and the rare adjustments, as well as time zone variations, it can be as early as the 19th or as late as the 22nd. For liturgical purposes, though, the equinox is reckoned as the 21st. (Wait till I tell you about the full moon!)
So, for example, following the Gregorian reform, 1600 was a leap year, but 1700, 1800, and 1900 were not; 2000 was a leap year, but 2100, 2200, and 2300 will not be. The “average” calendar year is now slightly less than 365.25 days; it is 365.2425 days, which is very, very, very close to the actual solar year of 365.2422 days.
How accurate? Depending on the method of calculation, the Gregorian calendar drifts by a day roughly every 3,000 to 7,000 years! By the year 4,000 or so, then, it may be necessary to modify the Gregorian system to allow leap years in years divisible by, um, 4,000, or something. There’s also the gradual slowing of the rotation of the Earth to consider. These are not problems we need to solve today on Substack.
Washington’s birth year was originally recorded as 1731 rather than 1732 because at the time the New Year was reckoned, not from 1 January, but from 25 March, “Lady Day” or the Feast of the Annunciation. Washington was 20 years old in 1752 when England, Wales, and the American colonies adopted the Gregorian calendar, dropping 11 days in September and adopting 1 January as New Year’s Day. (So Washington was 20 for only nine months!)
For simplicity’s sake I will use “Easter” to refer to the Western celebration, although as previously noted Latin Catholics speaking languages other than English and German largely use cognates of “Pascha.”
This happens when a) 21 March falls on a Sunday and b) the Church reckons the last winter full moon one day earlier, on Saturday the 20th, with the next full moon falling on Sunday, 18 April. This last happened in 1943, and will next happen in 2038! Note that after 2400 this range will change for reasons related to Metonic drift (see note 13 below and associated discussion above).
Thus, the calculation for the date of Easter, traditionally given as “the first Sunday after the first full moon on or after the spring equinox,” is actually as follows:
The equinox is reckoned as 21 March (though astronomically it can fall anytime between 19 March and 22 March, depending on time zone).
The ecclesiastical full moon on or after 21 March is determined by a tabular lunation.
Easter falls on the Sunday after that ecclesiastical full moon.
Among other implications, this means that, like Orthodox Pascha, though not nearly as quickly, the Jewish Passover is slowly sliding from the start of spring in the direction of summer.
In terms of the Catholic calculation of Easter, the oddest symptom I have found so far of this lunar imprecision is in 1981, when Catholic Easter fell on 19 April—the exact date of the astronomical full moon! Easter is supposed to be the Sunday after the full moon, or 26 April—a day later than the formula currently allows Easter to be. However, if we were also using the real vernal equinox, then the previous astronomical full moon, which fell on 20 March, would qualify, and Easter 1981 would have fallen on 21 March—a day earlier than the formula currently allows!
What fun. I've liked a good calendar deep dive, ever since finally finding out that leap year isn't just "every 4 years", back when working on the Year 2000 problem. This also feels like exactly the sort of entertaining rabbit hole that comes up whenever I look into Passover things, like the first time I participated in the Seder and was deeply confused why Maxwell House Coffee, of all people, published the Haggadah. That's right up there with Michelin Stars and Guinness Book of World Records for advertising pitches that became so culturally fundamental that they transcend sales and actually become somehow "good", which is a pretty wild transformation.
Huh, and I guess actually fits the Easter theme in a weird way.
An interesting wrinkle re. the problems in your footnote 11: now that we have atomic clocks, the official clock time is actually more "accurate" (i.e. more mathematically regular) than the solar phenomena themselves, since the earth's rotation is affected stochastically by geological events. Hence "leap seconds" are not added on a fixed schedule, but scheduled as needed by some person in authority (very likely the Astronomer Royal).
(Or were. Apparently leap seconds are difficult for programmers to deal with, and the current plan is to let the error accumulate up to a minute or more and let posterity deal with it. I find this vaguely upsetting.)
However, IIUC, leap seconds are only designed to correct the length of the calendar *day*, not the year. If the Gregorian calendar continues to be used for centuries into the future, some yet-to-be-determined correction will still have to be applied. Per Wikipedia, the 4000-year idea has been proposed before,[1] though I suspect it would make more sense to apply one-off corrections at somewhat shorter intervals, since the 4000-year cycle *still* wouldn't reduce the error to zero, and the length of the cycle is getting a bit ridiculous at that point.
[1] https://en.wikipedia.org/wiki/Gregorian_calendar#Accuracy