Thanksgivukkah FAQ

(Crossposted to Jewschool.)

As many of you know by now, November 28, 2013, will be both (American) Thanksgiving and the 1st day of Chanukah! The possibilities are endless: deep-fried turkey; latkes with cranberry sauce and gravy; pumpkin sufganiot; I’m sure you have more in mind. This week an article by Jonathan Mizrahi on this calendar issue has been making the rounds. It has some excellent graphs illustrating both the rarity of Thanksgivukkah in our present era and the long-term drift of the calendar that will make Thanksgivukkah impossible in the future, but it somewhat overstates its primary claim that Chanukah and Thanksgiving are “a once in eternity overlap”. This FAQ answers some questions that this article has inspired in various other forums, and corrects a few nuances.

Many thanks to Stephen P. Morse for creating an excellent tool to answer calendar questions quickly (though if he’s reading this, I’d love to see the capability of going beyond 9999 CE, and of distinguishing between Adar and Adar I), and to Remy Landau for providing the raw data on the Rosh Hashanah drift (though if he’s reading this, what’s with the popup ads?).

If you have questions that aren’t answered here, we’ll try to answer them in the comments (and if there are a lot, we’ll put together a sequel).

1. What is causing the long-term drift in the calendar?

You’ll notice from Mizrahi’s graph that the Jewish holidays shift significantly from one year to the next (like seasonal variations in the weather), but also (on average) slowly drift later over long time periods (like climate change). The year-to-year shifts are because the Hebrew calendar is primarily a lunar calendar, and 12 lunar months are approximately 354 days – much shorter than the solar year of ~365.25 days. Without any correction, the Jewish holidays would continue to move ~11 days earlier every year. (This is what happens with the Islamic calendar, in which every year is 12 lunar months without exception, so over several decades the Muslim holidays traverse the entire solar year.) In order to keep the Jewish holidays roughly aligned with the solar year (so that Pesach is always in spring, etc.), an month is added every few years, so Jewish “leap years” have 13 lunar months instead. As the Greek astronomer Meton discovered, 235 lunar months (=19*12 + 7) are approximately equal to 19 solar years, so if we put the calendar on a 19-year cycle, and add an extra month to 7 out of every 19 years, it mostly works out.

BUT NOT EXACTLY. 235 lunar months add up to 6939 days 16 hours 595 parts. (In Jewish calendar math, “parts” are the basic subdivisions of an hour, instead of minutes and seconds. There are 1080 parts in an hour, so 595 parts is about 33 minutes.) In the Gregorian calendar, 19 solar years (on average) are 6939 days 14 hours 626 parts. That’s about a 2-hour difference. So the Jewish holidays (on average) shift about 2 hours later during each 19-year cycle, which adds up to a full day every 231 years.


2. Is this an issue of Julian vs. Gregorian calendars?

Not really. 19 Julian years (on average) are 6939 days 18 hours. So if the Gregorian calendar is closest to the actual solar year, the Jewish calendar is doing better than the Julian calendar at approximating it (but still not well enough). (Think of it this way: By definition, the Julian calendar deviates from the Gregorian calendar by 3 days every 400 years. The Jewish calendar deviates by slightly less than 2 days in the same time period.)


3. But there’s some mechanism in place to correct this drift before it gets out of hand, right?

Nope. If no action is taken, the Jewish calendar will continue to drift later and later, until Pesach is in summer, Rosh Hashanah is in winter, etc. And it’s not clear how any action could be taken, since there’s no Jewish pope or Sanhedrin or any sort of body empowered to act on behalf of the whole Jewish people. But on the bright side, (as Mizrahi mentions) if we wait tens of thousands of years, we’ll loop all the way around to where we started.

The Catholics do have a pope, and so even though Easter is on a similar 19-year cycle, they’ve instituted corrections to keep it from drifting. Easter and Pesach usually coincide, but in the years when they’re a month apart instead, let’s just say it’s not Easter’s fault.



4. If we did take action to fix the calendar drift, what would that look like?

Generally speaking, over the long term, we’d need a way to have (very) slightly fewer leap years. I’ll get into specific proposals in a future post.


5. If November 28 is the earliest possible Chanukah, does that mean all the other holidays in 2013 are the earliest they can be?

Yes! We’re also getting the earliest Purim (February 24), Pesach (March 26), Shavuot (May 15), Rosh Hashanah (September 5), and the other fall holidays.

However, because of the aforementioned calendar drift, this is true only locally, for the present couple of centuries. The earliest Rosh Hashanah used to be September 4 (which means Purim on February 23, and so on for the rest of the holidays), but that happened for the last time in 1766. The last September 5 Rosh Hashanah (until we loop all the way around, of course) will be in 2089; after that, the earliest will be September 6.


6. Chanukah is never early or late – it always starts on the 25th of Kislev! Ha ha!

Yuk yuk yuk. You’re very clever, and showing your allegiance to Jewish time rather than to the surrounding culture. But the solar year is an actual physical thing, not an arbitrary convention of the secular calendar (even if the months of January, February, etc., are arbitrary) – it corresponds to the earth’s orbit around the sun, and therefore to many readily observable features of the seasons. The architects of the Jewish calendar understood this, and that was why they instituted leap years, to make sure the season-dependent Jewish holidays ended up in their proper seasons. The concept of “the holidays are early this year” would have been very familiar to the rabbis of the Talmud (even if their response of “Do you think we should add an extra Adar?” would be unfamiliar to us).


7. Is it true that Thanksgiving has never fallen during Chanukah before?

It’s true that, since Thanksgiving became a national holiday in 1863, no day of Chanukah has fallen on the 4th Thursday in November. But Thanksgiving wasn’t always on the 4th Thursday in November: originally, it was on the last Thursday in November (which could be either the 4th or 5th Thursday, depending on how many Thursdays were in November). The change happened under President Franklin D. Roosevelt, starting in 1939. (The motivation was to start the Christmas shopping season a week earlier, to help stimulate the economy. Can you imagine a time when the Christmas shopping season never started before Thanksgiving?) The original range for Thanksgiving was November 24-30; the current range is November 22-28.

And sure enough, there were two Thanksgivukkahs in the late 19th century, both on the 5th Thursday in November. Thursday, November 29, 1888, was the 1st day of Chanukah, and Thursday, November 30, 1899 was the 4th day of Chanukah.


8. If November 30 was the 4th day, that means Chanukah started on November 27! How is that possible?

Good question! After all, September 5 is the earliest possible Rosh Hashanah (and indeed, September 5, 1899, was Rosh Hashanah), and has been since 1766. And as we know from 2013, September 5 Rosh Hashanah corresponds to November 28! What’s the deal?

The answer is that, while all the holidays between Adar and Cheshvan have a fixed relationship (since all the months in between have a fixed number of days), Chanukah is different, since Cheshvan can have either 29 or 30 days. In “applesauce years”, Cheshvan has 30 days, and Chanukah begins on the same day of the week as Rosh Hashanah (as it will in 2013). In “sour cream years”, Cheshvan has 29 days, and Chanukah begins one day earlier, on the same day of the week as Shavuot. 1899 was a sour cream year, so Chanukah began one day earlier, on November 27.

However, Shavuot can never fall on Thursday, so Chanukah can never fall on a November 27 Thanksgiving.


9. Have there been any other November 27 Chanukahs since then?

No. 1899 was the last one, and will be the last one (until we loop around again).


10. Is it true that Thanksgiving will never again overlap with Chanukah?

Never say never. As Mizrahi points out, it will happen again in 70,000 years or so, when the calendar loops all the way around.


11. Ok, but seriously, what about during our (or our grandchildren’s) lifetimes?

After 2013, if we ignore what happens 70,000+ years in the future (and assume the rule for the date of Thanksgiving stays the same), Thanksgiving will never again fall completely during Chanukah.
However, Chanukah begins at sundown, and my family does Thanksgiving dinner in the evening (though I know not everyone does). And though the 1st day (or any other day) of Chanukah won’t fall on Thanksgiving again in the near-to-medium future, we’ll have a couple more instances when the first night of Chanukah is on Thanksgiving night: November 27, 2070, and November 28, 2165. (The first of those will be just before my 91st birthday, and I hope to be celebrating with my grandchildren (and maybe great-grandchildren), just as my son and I got to celebrate Thanksgiving this year with my nonagenarian grandfather.) After 2165, that’s it.

The partial Thanksgiving-Chanukah overlap happened once before, on November 28, 1918. Perhaps some of our older readers remember this?


12. Wait, I wasn’t alive in 1918, and I think I remember Chanukah starting on Thanksgiving night.

No, you don’t. You’re thinking of 2002, when Chanukah started on Friday night, November 29 (the night after Thanksgiving).


13. Have we had other November 28 Chanukahs in recent memory?

Yes, most recently in 1994. But that year, Chanukah began on Sunday night (Thanksgiving was November 24). After 2013, we’ll have it again in 2032, when Chanukah begins on Saturday night (following Thanksgiving on November 25).


14. Is it fair to say that Thanksgiving repeats on a 7-year cycle?

It’s not strictly true, since the 4-year Gregorian leap year cycle interferes with this. But Mizrahi’s point is about long-term averages, so his statement that “[y]ou would therefore expect them to coincide roughly every 19×7 = 133 years” is valid to within reasonable precision. The point is that the probability of Thanksgiving falling on any of the 7 allowed days is roughly equal, and is uncorrelated with the 19-year cycle (since 7 and 19 are both prime numbers).


15. While we’re on the subject of early Jewish holidays coinciding with American holidays, can Rosh Hashanah fall on Labor Day?

Yes! Labor Day is the first Monday in September, so it can fall anywhere from September 1 to 7, and part of that range is in the allowed range for Rosh Hashanah. It won’t happen this year, when Labor Day is September 2 and Rosh Hashanah is September 5. The last time was September 7, 1964, and the next time will be September 6, 2032.

As for Rosh Hashanah starting on Labor Day evening (so that 2-day Rosh Hashanah observers get a 5-day weekend), the last time was September 5, 1994, and the next time will be September 6, 2021.