The Days the Calendar Refuses

The Hook

Six chapters of astronomy have brought us here. The Rambam has taught the length of the lunar month down to a fraction most of us have never encountered outside this book, twenty nine days, twelve hours, and 793 parts of the 1080 parts into which the hour is divided. He has shown how to carry that number forward from a fixed starting point to any month in any year, and he has done it with the confidence of a man who knows exactly what he is doing.

And then chapter seven opens, and the very first thing he says is that the answer the calculation gives you is sometimes the wrong answer. Rosh Chodesh Tishrei is never established on a Sunday, a Wednesday, or a Friday. Never. If the molad falls on one of those three days, you move Rosh Chodesh to the day after, and Rosh Hashanah with it, and every festival of the year shifts along behind it.

There is something almost impudent about this. A man spends six chapters teaching you to compute, and then tells you that on three days out of seven the computation does not get the last word. He is not correcting an error in the math. The math is fine. He is telling you that the math was never the point.

Not on Sunday, Wednesday, or Friday

The rule has a name made of its own letters. Aleph, dalet, vav, spelled out as ADU, standing for the first day, the fourth day, and the sixth. Molad on Sunday, Rosh Chodesh Monday. Molad on Wednesday, Rosh Chodesh Thursday. Molad on Friday, Rosh Chodesh Shabbat. Three doors are closed and the day simply walks to the next one.

Then a second rule, and this one is about the hour rather than the weekday. If the molad occurs at midday or after midday, Rosh Chodesh is established on the following day. The Rambam gives the example: the conjunction on Monday, six hours after daybreak or later, and Rosh Chodesh is fixed on Tuesday. The moon has technically been reborn, and the day is handed to it anyway.

Now put the two rules in the same room and watch what happens. The molad falls at midday, so the day moves forward one. And the day it moves onto is a Sunday, or a Wednesday, or a Friday. The Rambam does not blink. If that following day is one of the three, Rosh Chodesh is postponed again, and established on the third day from the conjunction. His illustration is the one worth memorizing, because he gives it in numbers: a molad on Shabbat at midday, siman seven and eighteen, which is to say the seventh day at the eighteenth hour counted from nightfall, exactly noon. That year Rosh Chodesh Tishrei is Monday. The moon was born on Shabbat and the year begins on Monday, two full days later, and nothing has gone wrong.

Midday, and a Single Unit Before It

Two more postponements follow, and they are the strangest sentences in the chapter because they are pure number. In an ordinary year, if the molad of Tishrei falls on the night of the third day at nine hours and 204 parts of the tenth hour or later, siman gimel, tet, resh dalet, Rosh Chodesh is not fixed on Tuesday but on Thursday. And in a year that follows a leap year, if the molad falls on Monday at three hours and 589 parts of the fourth hour after daybreak or later, siman bet, tet vav, tav kuf peh tet, Rosh Chodesh is not fixed on Monday but on Tuesday.

Read those slowly and something starts to feel almost unbearable. A part, a chelek, is a 1080th of an hour, about three and a third seconds. The whole shape of the coming year, which festival lands on which weekday, when families travel, how long the year runs, turns on whether a computed lunar conjunction arrived at 204 parts past the hour or at 203. And the Rambam says so explicitly, in the very next breath. If the molad of that ordinary year came out one part earlier, at 203, Rosh Chodesh is established on Tuesday after all. If the year after a leap year came out at 588 instead of 589, Rosh Chodesh stands on Monday.

He has already done this once, in the opposite direction. If the molad occurs before midday, he writes, even by a single part, Rosh Chodesh is established on the day of the conjunction itself, provided that day is not one of the three. One chelek. Three and a third seconds on a clock nobody was watching, in a conjunction nobody can see, computed for a moon that at that moment is invisible in the sky. The entire year hangs on it, and the Rambam states the boundary without a trace of anxiety, the way you would state the width of a doorway.

The Reason He Actually Gives

Here is where the chapter turns, and where most of us have been carrying around the wrong explanation for years. If you ask why Rosh Hashanah cannot fall on a Sunday, Wednesday, or Friday, the familiar answers come quickly. So that Yom Kippur will not fall adjacent to Shabbat and leave two days without burial or fresh food. So that Hoshana Rabbah will not fall on Shabbat, when the willow may not be taken. These are real considerations and they are widely taught.

They are not what the Rambam says. He asks his own question, why do we not fix Rosh Chodesh on the days of ADU, and he answers with astronomy. Because this reckoning tracks the gathering of the sun and the moon according to their mean motion, not according to their true position in the sphere, as he explained earlier. Therefore they made a day of fixing and a day of deferring, in order to meet the day of the true conjunction. Tuesday we fix, Wednesday we defer, Thursday we fix, Friday we defer, Shabbat we fix, Sunday we defer, Monday we fix. The permitted and forbidden days alternate, and the alternation is a net thrown to catch a moment the arithmetic cannot pin down.

And he says the same principle stands behind the other four postponements as well, and then he offers proof. There are times, he writes, when the molad falls on the night of the third day and Rosh Chodesh is deferred to Thursday, and often enough the moon is not seen on Thursday night, and not even on Friday night. Which tells you the sun and the moon did not truly gather until Thursday. In other words: look up. The postponement was right and the calculation, taken alone, would have been early.

That is a stunning admission to place at the end of a chapter of this precision. The calculation is an average. Averages are correct about the long run and wrong about tonight. So the law builds its own margin of error directly into the calendar, not as an apology but as a mechanism.

The Unifying Principle

Something in this arrangement matches the structure of the inner life more closely than it first appears. The Alter Rebbe taught in the Tanya that the soul's service runs in two motions, ratzo and shov, a running out and a returning, an expansion and a contraction, and that neither one alone is service. A run without a return burns out and a return without a run never left. The Rambam has just built a calendar out of exactly that rhythm. A day of fixing, a day of deferring. We establish, we postpone, we establish. The forward motion and the holding back are not in tension. They are one instrument with two strokes.

The Baal Shem Tov taught that a descent is for the sake of an ascent, that what looks like a step backward is often the mechanism of the step forward. A postponed Rosh Hashanah is not a delayed Rosh Hashanah. It is a Rosh Hashanah that has been aimed better. The day was not taken from the moon. It was given to the moon that is actually there rather than the moon the equation predicted.

And underneath everything is what the Sages heard in the verse about the appointed times, that God says to Israel, they are yours, even when you err. The Lubavitcher Rebbe returned to that line constantly, because it says something almost frightening about partnership. The court is not reading the calendar off the sky and reporting it. The court is making it. Which is why the court is permitted to install four deliberate postponements and a rule of alternation, and why the year those postponements produce is not an approximation of the real year. It is the real year.

Modern Application

Most of us have the opposite relationship with our own instruments. We build a model, a forecast, a set of metrics, and then we defend the output as though the output were the world. The Rambam builds the most sophisticated computational tool in medieval Jewish law and then legislates, in advance and in public, the four situations in which its answer must be overridden and the reason why. The confidence is not in the number. It is in knowing the number's tolerance.

There is a harder application in the chelek. Nothing in your week is decided by three and a third seconds, but plenty is decided by margins just as invisible: the sentence you almost sent, the phone call placed just before the mood turned, the day you started something rather than the day after. We tend to believe that the important things are decided by the important things. This chapter says that a year of festivals can hinge on one part of an hour, and that a serious person does not find this absurd. He computes it, writes it down, and lives by it.

The Closing

Rosh Hashanah is the only festival the Torah sets on the first of a month, and the Psalm calls it the day covered over, bakeseh, the day when the moon is hidden. The Sfat Emet dwelt on that concealment as the point rather than the problem, the year beginning precisely when there is nothing to see. This chapter is the legal shape of that idea. The court fixes the head of the year on a night when the moon cannot be sighted, using a reckoning it openly admits is an average, protected by postponements whose stated purpose is to catch a conjunction nobody can observe.

Three days are closed, four instances move the date, and one part of one hour decides between Tuesday and Thursday. None of it is guesswork and none of it is certainty either. It is what precision looks like when it is honest about its own reach: exact to a 1080th of an hour, and humble enough to build in a day of deferring so that the year begins on the day the moon is truly reborn rather than the day the equation said it would be. Fix what you can compute. Then leave a day open for the thing you cannot see.