Sighted Moon vs Calculated Calendar: Who Decides?
There is a stretch around conjunction when the moon cannot be picked out in the glare. Not thin — gone from ordinary sight. The old crescent left the evening sky and thinned into the dawn; the returning crescent has not yet appeared.
Somebody has to decide which day that absence belongs to, and that decision is the whole sighted moon vs calculated calendar dispute. One camp says the month begins when a person sees the returning crescent and says so out loud. The other says it begins where the arithmetic puts it, worked out centuries ahead. Two more positions stand alongside: the invisible instant of alignment, and, for a minority, the full moon.
Behind all four sits a question the sky cannot answer. Not what the moon did. Who gets to say what it did.
Where in the lunation each camp plants its flag
The Hebrew word for month is chodesh, from the root chadash — to be new, to renew. It names what happens, not how you catch it happening. Scripture commands the day — "in the beginnings of your months, ye shall blow with the trumpets" — Numbers 10:10 — and never gives a procedure for finding one. Four families grew in that silence. Run a lunation forward and watch where each stops.
Start at the alignment. Conjunction is one instant, the same in every time zone, indifferent to cloud and calculable to the second, and that precision is its whole case. Its weakness is that nothing marks it: a shepherd in Iron Age Judah could not observe an event with no appearance. That is the Karaite objection in one line.
After conjunction, the first crescent eventually becomes catchable low in the west just after sundown. This is the answer that fits the word: if chodesh is renewal, the renewal is thin light returning where there was none. Philo, in first-century Alexandria, describes the festival when "the sun begins to illuminate the moon with a light which is visible to the outward senses" (Special Laws II). Karaite practice uses observation, as do some Messianic and Hebrew-roots fellowships. The cost is that eyesight is local: cloud over Jerusalem does not cloud Cairo.
Keep going about a fortnight and the moon is full, where a minority plant the head of the month on one verse read as tight parallelism: "Blow the horn on the new moon, on the full moon for our feast day." — Psalm 81:3. Many translations take chodesh and keseh there as two moments inside one festival month, and the rare keseh also appears in Proverbs 7:20 for the time of a husband's return. Sincerely held, textually thin.
The fourth position never looks up. Calculation opens the month where the arithmetic says, as mainstream Jewish practice has for roughly a millennium, and it buys certainty: a community cut off from every authority keeps the feast on the same night as one six thousand miles away, and can print the date for the year 2400.
Sighted moon vs calculated calendar: the real dispute is authority
The detail that reframes all this comes from an early rabbinic source. The Mishnah's tractate on Rosh Hashanah records how a month was declared: pairs of witnesses came before the court and were examined apart — how did you see the moon, in front of the sun or behind it, which way did it lean (Rosh Hashanah 2:6).
Then the case that gives the game away. Two men testified they had seen the waning moon in the east at dawn and a new crescent in the west that same evening — astronomically impossible. Rabbi Yohanan ben Nuri called them false witnesses. Rabban Gamliel accepted their testimony and sanctified the month anyway (Rosh Hashanah 2:8).
The neat opposition dissolves there. The month was made by a court's declaration, not by a photon reaching a retina — sighting furnished the evidence, a ruling made the date. When the tradition swapped testimony for arithmetic, it moved an authority that had never been the eyeball alone. And the communities that refused were not naive about weather — they were denying that anyone may stand between the sky and the congregation.
The tidy story about Hillel II, and the six centuries behind it
You will read everywhere that Hillel II published the calculated calendar in 358/9 CE, when Roman persecution made convening a court unsafe. Clean story, thinner evidence than the confidence with which it travels.
No contemporary rabbinic source records Hillel instituting the calendar now in use. The earliest attestation is a passage attributed to Hayye Gaon in a responsum dated 991/2 CE and preserved in Abraham bar Hiyya's Sefer ha-Ibbur (1123). Its context concerns intercalation: Sacha Stern reads it as crediting Hillel with fixing seven leap years in a nineteen-year cycle, not with creating every rule of the present calendar. The account therefore reaches us more than six centuries after 358/9 CE, through a textual tradition whose surviving versions are not identical.
Two bodies of evidence sit awkwardly against the tidy version. A dated exilarch's letter for 835/6 CE sets Passover two days earlier than the present calendar would, showing that the current rules were not yet in force. Then in 921/2 Palestinian leader Ben Meir and the Babylonian authorities announced festival dates two days apart, and communities across the Near East observed the same festivals on different days. The surviving evidence records a serious dispute between rabbinic authorities, not a single uncontested act that established every rule of the calendar.
The evidence places the emergence of the present fixed calculation in the ninth and tenth centuries, not in one fourth-century act. Stern's later work also cautions against treating 921/2 as the finishing point: a divergent rabbinic calendar is still attested in southern Italy in 946. Which cuts both ways. "Unchanged since the Sanhedrin" overstates it, and "a tenth-century invention" understates it just as badly.
What critics mean when they call the postponements a fudge
Once the calendar ran on paper, four rules — the dechiyot, "postponements" — could shove the new year forward a day or two. The mechanics are worked step by step in how Passover dates are calculated. The objection is what matters here.
Three of the four have no immediate astronomical trigger: one helps keep the Day of Atonement from abutting the Sabbath, while two prevent year lengths the system does not allow. Critics argue that moving the first of a month for scheduling makes the calendar a human ruling rather than a direct report of the sky. Karaite critics have long pressed that point.
The defence is not weak. The remaining rule, molad zaken, fires when the mean conjunction falls after noon, and its rationale is whether a crescent that young could realistically be caught over the Land of Israel that evening. Sighting logic, sitting inside the calculated system and doing sighting work. In intention the two camps stand closer than their published dates ever let on.
How far apart the two methods can put a feast
Usually less than you would fear.
Very young crescents are difficult to see. Angular separation, altitude, haze, latitude and observer skill all matter. A community farther west can sometimes report a crescent later on the same civil day that Jerusalem reports nothing.
Lunar months are normally twenty-nine or thirty days, so an ordinary sighting dispute does not accumulate without bound. Later observations may realign the dates or leave them a day apart.
The exception is the year the intercalation decisions part company — the fixed calendar inserting its thirteenth month on schedule while an observational community waits on barley in the fields of Israel. Then the same feast stands a whole lunation apart, and a family keeps Passover a month from cousins in the next street. The four rules set against each other with dates get an article of their own.
Common questions
Should the month begin at conjunction or first visible crescent?
Scripture names neither. Chodesh means renewal without saying how the renewal is detected, and no verse gives a procedure. Conjunction offers one worldwide instant that nobody can witness. First crescent offers something witnessable, at the cost of weather, latitude and delay.
Was the calendar really fixed by Hillel II in 358 AD?
Not in the complete form used today. No contemporary rabbinic source records such an institution. The earliest attestation is a passage attributed to Hayye Gaon in 991/2 CE and preserved by Abraham bar Hiyya; in context it concerns the nineteen-year intercalation cycle, not every rule of the later calendar. A dated letter from 835/6 gives a Passover date two days earlier than today's calculation, so the present system was not yet in force.
What are the postponement rules and why do critics object?
Four rules can push the new year forward. One keeps the Day of Atonement from abutting the Sabbath, two keep the year within a permitted length, and molad zaken turns on whether the crescent could have been seen. Critics say the first three are scheduling decisions dressed as celestial facts.
How far apart can the two methods put a feast?
In an ordinary year, usually a day or two, depending on weather and the rule each community uses. A disagreement over adding a thirteenth month can leave two communities keeping the same feast about one lunation apart.
When the congregation across town keeps a different date
Sooner or later this stops being theoretical. Two households in one city keep the same feast a day apart, both sincere, both able to show their working, and neither able to settle it from a verse — because the verse everyone quotes names a day and not a method.
Day 4 is no use in that argument, and it is worth saying why rather than posing as neutral: it does not start months from the moon at all. Its year is 364 counted days — 52 whole weeks, four seasons of 91 — anchored to the March equinox, read from a canonical per-year table rather than computed or averaged. After day 364 the counting stops for one or two uncounted days until the next equinox re-pins Day 1, which always falls on the week's day 4. No witnesses, no molad, no postponements. Enoch, Jubilees and the Qumran calendrical texts all run a year on that shape, though whether the Temple ever did is unsettled among scholars and the app does not pretend otherwise. A different answer, not a better one, and one that will never print a sighted-crescent date. Why Day 4 takes that road is worked out here.
Which leaves the older question, the one the Mishnah had already met. Rabbi Yehoshua believed his teacher had sanctified the wrong month, and said so. Rabban Gamliel ordered him to appear, staff and purse in hand, on the day Yehoshua's own reckoning made the Day of Atonement — an ordinary errand if Gamliel was right, the holiest fast of the year if Yehoshua was. He went. Gamliel rose and kissed him on the head: come in peace, my teacher and my student (Rosh Hashanah 2:9). Nobody there conceded the astronomy. What they settled was how to keep a feast beside somebody who dates it differently, and the sky was never going to decide that for anyone.
Sources: Mishnah Rosh Hashanah 2 · Philo, The Special Laws II · Sacha Stern, Calendar and Community (Oxford University Press, 2001) · Sacha Stern, "Jacob Mann's Contribution to the History of the Jewish Calendar" · Princeton Geniza Project: Exilarch's calendar letter, T-S 8G7.1 · Sacha Stern et al., The Jewish Calendar Controversy of 921/2 CE (Brill, 2019)
