Thus far the day has spoken. The sign asks for three nights, the week carries two sabbaths for it, the fourteenth of Nisan is the stronger day of death, and the tradition that says "Friday" turned out to have no independent memory. Everything points to the middle of the week. But a day is not a date: Wednesday — which Wednesday? There are fifty-two of them in any year. To turn a weekday into a date the year is needed, and that stands not in the text but in the sky. So the inquiry follows the moon that the previous chapter pointed to, and asks: which years permit a Passover on the right weekday at all?
And let me say at once where this chapter does not end up, for that is the core of its honesty. The stars do not deliver the Wednesday. Whoever hopes that astronomy will drop "25 April AD 31" out of the sky will be disappointed. What astronomy delivers is something more modest and something stronger at once: a two-horse race, and a funnel.
Begin with the month, for that is where everything begins. The Jewish month was not calculated but seen. At the end of the old month the calendar commission convened and waited for witnesses who could report the first lunar crescent under oath; if they saw it, the new month began the next day; if they did not, the old month received a thirtieth day and the new one began after that. A month could count no fewer than twenty-nine and no more than thirty days. That is not a theory; that is how it worked. And it means that the date of 1 Nisan in any given year is not a matter of preference but of celestial mechanics: when, after the last new moon before spring, was the first crescent visible above Jerusalem?
For the year 31 that answer is sharp. The moon stood new — in exact conjunction with the sun, invisible — on 10 April, around midday. That evening, 10 April, the crescent was still too young and too close to the sun to be seen. The following evening, 11 April, it stood well above the horizon: nearly thirty hours old, thirteen degrees from the sun, setting a good hour after it. That is no borderline case. That is a crescent every observer sees — one that clears the thresholds of all the standard visibility criteria with room to spare (see the Workbench). Visible on the evening of 11 April means: 1 Nisan = 12 April. And 1 Nisan on 12 April means 14 Nisan — the day of the Passover lamb, the day of death — on 25 April. And 25 April AD 31 is a Wednesday.
Expertthe month is seen, not counted
The hinge beneath everything is the distinction between the conjunction (the astronomical new moon, when moon and sun share the same ecliptic longitude — invisible) and the first crescent (the "new light", the narrow arc of light that first becomes visible one or two evenings later). The Jewish month began not at the conjunction but at the crescent — at observation, not at calculation. That difference of one or two days is not trivial: it determines on which day 1 Nisan falls, and therefore on which weekday 14 Nisan lands. Whether the month was also bound to a second requirement — that Nisan might begin only after the spring equinox — is the narrowest link in the whole case, and it receives a chapter of its own (ch. 8). Here only one thing counts: the crescent of 11 April AD 31 was there, beyond dispute.
Now the second step, and it is the heart of this chapter. For recomputing one year is not enough; you must see the whole field. Lay the crescent computation across all the years in which the crucifixion can reasonably fall — roughly AD 26 to 36 — and see on which weekday 14 Nisan falls each time. Do that, and no "Wednesday of its own accord" appears anywhere. What appears is a race between two horses.
That window does not belong to the Wednesday school. The field itself keeps it just as wide: Helen Bond closes her survey of the dating with the finding that "the precise date can no longer be recovered. All that can be claimed [...] is that Jesus died some time around Passover [...] between 29 and 34 CE" (Bond, 2013). Within that window AD 31 is an ordinary candidate, no outlier; that the year appears on no shortlist is due not to the window but to the Friday assumed before the weighing (ch. 1). And the measure holds in return: whoever cites Bond's window also takes along her conclusion that the date can no longer be recovered from the available data. Against her this chapter claims no proof; it claims a funnel within her window.
One horse is AD 30: Friday 7 April. The other is AD 31: Wednesday 25 April. And the third year the consensus cherishes — AD 33, with its famous "Good Friday" of 3 April — turns out to be a limping horse: it reaches the line only if the calendar rule grants it a day and a half of leeway. That is not my judgement; it is measurable, and the measure is the distance of the March moon from the spring equinox.
Dossierthe seven years, and what the race rests on
The reproduction. As early as 1944, R. A. Parker — the same Parker as the standard work Babylonian Chronology — computed the first-crescent dates for Jerusalem for AD 28 to 34, and he notes expressly that his Babylonian tables hold for Jerusalem as well, provided one "subtracts 37 minutes" for the difference in longitude (Parker, 1944, pp. 173–174). I have independently recomputed his seven conjunctions with modern ephemerides; the two series agree to the day in all seven years. One honesty belongs with this: Parker's Jerusalem table (1944) and the standard work Babylonian Chronology (1946) are the same Babylonian computational method from the same hand — they confirm each other, but they are not independent. The genuinely independent check is the modern ephemeris, which coincides with Parker's Babylonian series in all seven years. And a wholly different method confirms that same April month: Grace Amadon — herself a defender of AD 31 — likewise computed an April Passover in 1942 on barley-harvest grounds, by a route that shares nothing with Parker's Babylonian reckoning; though within it she forced a Friday (27 April) that lies two days after the earliest visible crescent and is astronomically untenable (Amadon, 1942). Her independent result, then, is the April month; her Friday founders on precisely the crescent from which the Wednesday hangs. The figures, year by year:
year
1 Nisan
14 Nisan
weekday
AD 28
15 Apr
28 Apr
Wednesday
AD 29
5 Apr
18 Apr
Monday
AD 30
25 Mar
7 Apr
Friday
AD 31
12 Apr
25 Apr
Wednesday
AD 32
1 Apr
14 Apr
Monday
AD 33
19 Apr
2 May
Saturday
AD 34
9 Apr
22 Apr
Thursday
*(AD 33 stands here on Saturday 2 May — the outcome under the strict equinox rule; under the observing barley threshold its 14 Nisan falls on Friday 3 April. See appendix A.2 against A.3. AD 31 too depends on the choice of rule — Wednesday 25 April or else Tuesday 27 March — but with this difference, and it is the difference this chapter turns on: with AD 33 the choice decides whether there is a Friday, with AD 31 it decides only which midweek day it becomes.)*
The crescent of 11 April AD 31 against the criteria. The three standard visibility criteria in the professional literature — Schoch's, Yallop's (1997) and Odeh's (2004) — differ only in borderline cases of a few tenths of a degree. The crescent of 11 April AD 31 is no borderline case: with an elongation of ~13.7°, a lag of ~61 minutes between sunset and moonset, and an age of ~28 hours, it lies well above the threshold of each of the three. All three: visible. (The March crescent of 12 March is a borderline case — but that does not matter, for both outcomes of that borderline case lead, thirty days on, to the same 1 Nisan = 12 April.)
The thirty-day cap — and why cloud buys no Friday. One objection to the crescent of 11 April remains open, and it is the strongest available from the other side: the sky may have been clouded that evening, and no computation can exclude weather. Granted. What the objection cannot do is produce a Friday. On an observational calendar a month caps at thirty days — the rule with which this chapter began — and Adar II began on 13 or 14 March, from the crescent of the borderline evening above or from that of 13 March, which set ninety-four minutes after the sun and lies far inside zone A. From either start the cap puts 1 Nisan no later than 13 April, and 14 Nisan no later than Thursday 26 April — which is the very shift Jeremias must plead for, and it delivers him a Friday only as the fifteenth (ch. 1). Friday 27 April as 14 Nisan would need 1 Nisan on 14 April, that is an Adar II of thirty-one days, and the cap forbids it. Cloud on 11 April and 12 April therefore changes nothing here: the cap does the work, not the crescent. To reach a Friday 14 Nisan the court would have to have missed the crescent of 13 March as well. Every Friday available in AD 31 is a Nisan 15 Friday. [CALCULATED — own reproduction, ephem 4.2.1: conjunctions 11 March 22:21 UT and 10 April 11:35 UT, weekdays from the Julian day number]
The base rate. Across the eleven years AD 26–36, 14 Nisan falls on a Wednesday only twice: AD 28 and AD 31 — two in eleven — a frequency of 0.18, and thus at the chance level of a random weekday (1/7 ≈ 0.14). (One edge case belongs here honestly, and it cuts against this book's own case: the March alternative of AD 34 gives a Wednesday on 24 March as well. That 24 March lies — contrary to what is quickly said — two days after the equinox (22 March), precisely the profile of AD 31's March option that chapter 8 takes seriously: the month begins before the equinox, the fourteenth falls after it. Under the strict "feast-after-equinox" rule that Anatolius reports from the ancients AD 34's Wednesday survives just as well as AD 31's Tuesday beneath it survives; under Anatolius' month rule it falls away and the April reading of AD 34 remains, which gives a Thursday. If one counts the March option, the Wednesday base rate becomes 3/11 and the funnel gains a second address — that AD 34 nonetheless falls away is then not a matter of the equinox but of the ministry window: a baptism in 27 admits no crucifixion in 34 (ch. 9). The measure must lie equal for both years.) A Wednesday "by itself" is therefore at chance level. The strength lies not in the Wednesday alone, but in its crossing with the day axis (see the main text). [CALCULATED · appendix A]
The eclipse — image, not argument. In the night that ushers in 15 Nisan, 25 April AD 31, a partial lunar eclipse (± 35 %) stood high in the sky above Jerusalem — at its start, 21:35 local time, thirty-four degrees above the horizon, climbing past forty; it appears, remarkably enough, in the very table of the chief defenders of the rival year (Humphreys & Waddington, 1992, p. 345). I name it here once, and after that never again as an argument. It is image, not proof; a partial eclipse does not turn the moon blood-red, and whoever hangs weight on it measures with a double standard. As devotion: striking. As evidence: nil.
Why is AD 33 the lame one, and AD 30 and AD 31 the horses? Because the crescent that counts falls in March, and the question each time is whether that March moon stands before or after the spring equinox (22/23 March). Measure that distance, and the asymmetry is hard. In AD 30 the conjunction fell almost exactly on the spring equinox (0.2 day before it), and the first crescent a good day after it — so every observing spring rule keeps Nisan in March; 14 Nisan = Friday 7 April (under the generous crescent threshold Thursday 6 April), and under no such rule a Wednesday. In AD 31 the March moon stood nine days before the equinox (the crescent; the conjunction eleven days), deep in winter. There too the rule still chooses: under Anatolius' month rule that winter moon is rejected and 14 Nisan falls on Wednesday 25 April; under the strict rule that Anatolius reports from the ancients, which demands only the feast after the equinox, it stands and the fourteenth comes on Tuesday 27 March (ch. 8, where that rule is weighed as an attribution). But — and here lies the whole matter — neither is a Friday. The choice of rule decides, in AD 31, which midweek day it becomes; it yields no Friday under any reading. Compare that with AD 33, where the March moon stood a day and a half before the equinox: there that same choice decides not between two midweek days but over the existence of the Friday itself — if the equinox is a hard lower bound, AD 33 moves to May and the Good Friday vanishes; if it is not, Friday 3 April AD 33 stays alive.
See what stands there, and see it honestly. Astronomy does not deliver the Wednesday as the winner. It delivers a two-horse race — AD 30-Friday or AD 31-Wednesday — and it weakens only the third horse, the traditional Good Friday of AD 33. Whoever reads "therefore Wednesday" out of that measures with two standards, for the same measurement that makes AD 33 totter leaves AD 30 no Wednesday. Between the two horses the sky does not choose. That choice falls on the other axis — with the day, with the length of the ministry, with the heaviest verse — and not here.
And here comes the funnel. For suppose now that the other axis has already done its work. Suppose that the text — the sign of Jonah, the two sabbaths, the fourteenth of Nisan — has led you to a Wednesday. Then walk the whole field and ask: which Wednesday? In the eleven years AD 26–36, 14 Nisan has fallen on a Wednesday exactly twice: AD 28 and AD 31. And AD 28 falls outside the window of the ministry — too early, before everything that the baptism and the temple-building and the years of John allow (ch. 9). One address remains.
Whoever arrives at a wednesday on textual grounds hasunder the observational calendar — in the whole field only one address: 25 April AD 31.
That is the shape of the whole book in one sentence: the measurement of the day times the funnel of the year. Not astronomy spitting out the date, but astronomy which, given the day, leaves exactly one year standing.
Countercasethe two rivals at their strongest
Two years deserve to be defended here at full strength, and neither is weak.
AD 33, the reigning date. The classic case is that of Humphreys and Waddington: 14 Nisan fell on Friday 3 April AD 33, and on the evening of that day a moon rose eclipsed above Jerusalem — the "blood moon" Peter cites in Acts 2:20 (Humphreys & Waddington, 1983; 1992). Astronomically AD 33 hangs on the weak equinox rule — and that rule is not simply invented: Sacha Stern, the standard work on the Jewish calendar, states expressly that people did not always wait for the equinox — "if both crops and fruits were sufficiently ripe, no intercalation would be made, even if this meant Passover occurring before the equinox. This may have happened quite frequently" (Stern, 2001, p. 70). Under that lenient rule Friday 3 April AD 33 lives. This is the strongest rival, and it is not written off here.
AD 30, the other horse. Astronomically stronger still is AD 30: its 14 Nisan falls on Friday 7 April, and that stands — as said — under every calendar rule, for the March moon stood after the equinox. AD 30 is not excluded by the sky. If it falls, it falls on the other axis: on the day arguments (chs. 3–6), on the length of the ministry and the forty-six years of John 2:20 (ch. 9), and on the heaviest verse (ch. 13). Astronomically AD 30-Friday is no more excluded by the sky than AD 31-Wednesday — though its firmness lies in "no Wednesday", not in a rock-solid Friday (the threshold evening is a borderline case Thursday/Friday; appendix C.2). That must be said, and it stands.
Excursus"computational tradition", not "measurement", and the circle of the consensus
Two honesties belong here, and the first cuts into this book's own source. The table of Parker and Dubberstein is a reconstruction of the Babylonian calendar — a fixed nineteen-year cycle. The Jewish calendar of the first century was not that; it was observational, a decision of the court upon eyewitness testimony. That 1 Nisan AD 31 fell on 12 April in Babylon therefore does not prove that it fell on 12 April in Judaea; it shows that the natural, seasonally correct placement of Nisan lay in April. For that reason I consistently call P&D a computational tradition, never a measurement of what Judaea did. The bridge from Babylon to Jerusalem is a separate case, and it belongs in Chapter 8.
The second honesty cuts into the opposing party, and it is sharp. The consensus rejects an intercalary month in AD 31 — the month that shifts Nisan to April — with a curious argument. Humphreys and Waddington themselves concede that "we possess no historical reports as to the proclamation of leap-months in the years AD 26–36" (1992, p. 336), and they dismiss the April variant with the sentence that, if Nisan was one month later, "Nisan 14 would not fall on a Friday in any year". Read that sentence again: the intercalary month is rejected because it yields no Friday. And they do retain it at the one place where it does yield one — Nisan 15 on 23 April AD 34, the single intercalary case left standing in their own Table 2. The same instrument, then, held on to where it delivers a Friday and let go where it does not. But that the crucifixion fell on a Friday is precisely what was to be demonstrated. The conclusion has become the premise. That is not a measurement of the calendar; that is a majority, smuggled in as a measure. And it is the reason AD 31 never stood on a shortlist — not because it had been weighed and found wanting, but because the axiom excluded it before the weighing (ch. 1).
Where does the inquiry stand, then, at the end of the first half of the year axis? Not at an established year. It stands at a two-horse race in which the sky does not choose, and at a funnel which, given the day, leaves one address. Astronomy does not exclude AD 31 — it supplies it, Wednesday and all. It does not exclude AD 30 — that must fall elsewhere. And it undermines only the Friday of AD 33, and even that not entirely.
But all of this rested, as you will have noticed, on one word I kept postponing: the equinox. Was Nisan permitted to begin only after the spring equinox, or not? Under the strict rule AD 33 falls away and the Wednesday of AD 31 stands hard; under the lenient rule AD 33 lives again. The whole funnel therefore stands or falls with the calendar of the temple — with the question of which rule Jerusalem actually followed. And that is the next question.
HeartHEART
"Let there be lights in the vault of heaven, to divide the day from the night; and let them be for signs and for appointed times" (Genesis 1:14). The moon that the commission stood awaiting on the hill above Jerusalem was no arbitrary lamp. It had been set — to mark the times, to carry the feasts, to date, without knowing it, the night in which the Lamb would be slain. Its positions are recomputed with machines the priests did not have. But it is the same moon, above the same city, and it kept the same time. There is something moving about a God who did not let His greatest deed take place in hiding, but hung it on a crescent everyone could see.
Note on the sources
The crescent and conjunction computations can be verified with standard ephemerides (ephem/Swiss Ephemeris) and are calibrated against an independent anchor point; the thirty-day cap in the Dossier is calendar arithmetic on top of them — the month lengths and the weekdays follow from the Julian day number and require no visibility model at all, which is why that step survives any dispute over the crescent; the reproduction of Parker's seven Jerusalem dates (1944) agrees to the day in all seven years; the article itself has been read at first hand, and his Jerusalem tabulation is quoted from it directly. The visibility criteria are the published criteria of Schoch, Yallop (1997) and Odeh (2004); the conclusion for 11 April AD 31 hangs not on an implementation of my own but on the fact that that crescent lies well above every published threshold. The eclipse of 25 April AD 31 carries no argumentative weight whatever in this chapter and is named solely as an image. Where "Wednesday 25 April AD 31" is heard, that is a computation under a documented rule — not proof, and it is not presented as proof either.
Literature (APA)
Amadon, G. (1942). The Johannine-synoptic argument. Journal of Biblical Literature, 61(4), 227–280.
Beckwith, R. T. (1996). Calendar and chronology, Jewish and Christian: Biblical, intertestamental and patristic studies. Brill. (p. 289) [consulted, not cited in this chapter]
Bond, H. K. (2013). Dating the death of Jesus: Memory and the religious imagination. New Testament Studies, 59(4), 461–475.
Humphreys, C. J., & Waddington, W. G. (1983). Dating the crucifixion. Nature, 306(5945), 743–746.
Humphreys, C. J., & Waddington, W. G. (1992). The Jewish calendar, a lunar eclipse and the date of Christ's crucifixion. Tyndale Bulletin, 43(2), 331–351. (pp. 336, 345)
Odeh, M. Sh. (2004). New criterion for lunar crescent visibility. Experimental Astronomy, 18, 39–64. https://doi.org/10.1007/s10686-005-9002-5
Parker, R. A. (1944). Ancient Jewish calendation: A criticism. Journal of Biblical Literature, 63(2), 173–176.
Parker, R. A., & Dubberstein, W. H. (1946). Babylonian chronology 626 B.C.–A.D. 45 (Studies in Ancient Oriental Civilization 24, p. 46). University of Chicago Press. [3rd ed. 1956, to A.D. 75.]
Stern, S. (2001). Calendar and community: A history of the Jewish calendar, 2nd century BCE–10th century CE (p. 70). Oxford University Press.
Yallop, B. D. (1997). A method for predicting the first sighting of the new crescent moon (NAO Technical Note No. 69). HM Nautical Almanac Office.
Scripture texts: Genesis 1:14 · Acts 2:20. My own working translation. Astronomical computations: own reproduction (ephem 4.2.1), calibrated against an independent anchor point; full tables in appendix A.
Thus far the day has spoken. The sign asks for three nights, the week carries two sabbaths for it, the fourteenth of Nisan is the stronger day of death, and the tradition that says "Friday" turned out to have no independent memory. Everything points to the middle of the week. But a day is not a date: Wednesday — which Wednesday? There are fifty-two of them in any year. To turn a weekday into a date the year is needed, and that stands not in the text but in the sky. So the inquiry follows the moon that the previous chapter pointed to, and asks: which years permit a Passover on the right weekday at all?
And let me say at once where this chapter does not end up, for that is the core of its honesty. The stars do not deliver the Wednesday. Whoever hopes that astronomy will drop "25 April AD 31" out of the sky will be disappointed. What astronomy delivers is something more modest and something stronger at once: a two-horse race, and a funnel.
Begin with the month, for that is where everything begins. The Jewish month was not calculated but seen. At the end of the old month the calendar commission convened and waited for witnesses who could report the first lunar crescent under oath; if they saw it, the new month began the next day; if they did not, the old month received a thirtieth day and the new one began after that. A month could count no fewer than twenty-nine and no more than thirty days. That is not a theory; that is how it worked. And it means that the date of 1 Nisan in any given year is not a matter of preference but of celestial mechanics: when, after the last new moon before spring, was the first crescent visible above Jerusalem?
For the year 31 that answer is sharp. The moon stood new — in exact conjunction with the sun, invisible — on 10 April, around midday. That evening, 10 April, the crescent was still too young and too close to the sun to be seen. The following evening, 11 April, it stood well above the horizon: nearly thirty hours old, thirteen degrees from the sun, setting a good hour after it. That is no borderline case. That is a crescent every observer sees — one that clears the thresholds of all the standard visibility criteria with room to spare (see the Workbench). Visible on the evening of 11 April means: 1 Nisan = 12 April. And 1 Nisan on 12 April means 14 Nisan — the day of the Passover lamb, the day of death — on 25 April. And 25 April AD 31 is a Wednesday.
Expertthe month is seen, not counted
The hinge beneath everything is the distinction between the conjunction (the astronomical new moon, when moon and sun share the same ecliptic longitude — invisible) and the first crescent (the "new light", the narrow arc of light that first becomes visible one or two evenings later). The Jewish month began not at the conjunction but at the crescent — at observation, not at calculation. That difference of one or two days is not trivial: it determines on which day 1 Nisan falls, and therefore on which weekday 14 Nisan lands. Whether the month was also bound to a second requirement — that Nisan might begin only after the spring equinox — is the narrowest link in the whole case, and it receives a chapter of its own (ch. 8). Here only one thing counts: the crescent of 11 April AD 31 was there, beyond dispute.
Now the second step, and it is the heart of this chapter. For recomputing one year is not enough; you must see the whole field. Lay the crescent computation across all the years in which the crucifixion can reasonably fall — roughly AD 26 to 36 — and see on which weekday 14 Nisan falls each time. Do that, and no "Wednesday of its own accord" appears anywhere. What appears is a race between two horses.
That window does not belong to the Wednesday school. The field itself keeps it just as wide: Helen Bond closes her survey of the dating with the finding that "the precise date can no longer be recovered. All that can be claimed [...] is that Jesus died some time around Passover [...] between 29 and 34 CE" (Bond, 2013). Within that window AD 31 is an ordinary candidate, no outlier; that the year appears on no shortlist is due not to the window but to the Friday assumed before the weighing (ch. 1). And the measure holds in return: whoever cites Bond's window also takes along her conclusion that the date can no longer be recovered from the available data. Against her this chapter claims no proof; it claims a funnel within her window.
One horse is AD 30: Friday 7 April. The other is AD 31: Wednesday 25 April. And the third year the consensus cherishes — AD 33, with its famous "Good Friday" of 3 April — turns out to be a limping horse: it reaches the line only if the calendar rule grants it a day and a half of leeway. That is not my judgement; it is measurable, and the measure is the distance of the March moon from the spring equinox.
Dossierthe seven years, and what the race rests on
The reproduction. As early as 1944, R. A. Parker — the same Parker as the standard work Babylonian Chronology — computed the first-crescent dates for Jerusalem for AD 28 to 34, and he notes expressly that his Babylonian tables hold for Jerusalem as well, provided one "subtracts 37 minutes" for the difference in longitude (Parker, 1944, pp. 173–174). I have independently recomputed his seven conjunctions with modern ephemerides; the two series agree to the day in all seven years. One honesty belongs with this: Parker's Jerusalem table (1944) and the standard work Babylonian Chronology (1946) are the same Babylonian computational method from the same hand — they confirm each other, but they are not independent. The genuinely independent check is the modern ephemeris, which coincides with Parker's Babylonian series in all seven years. And a wholly different method confirms that same April month: Grace Amadon — herself a defender of AD 31 — likewise computed an April Passover in 1942 on barley-harvest grounds, by a route that shares nothing with Parker's Babylonian reckoning; though within it she forced a Friday (27 April) that lies two days after the earliest visible crescent and is astronomically untenable (Amadon, 1942). Her independent result, then, is the April month; her Friday founders on precisely the crescent from which the Wednesday hangs. The figures, year by year:
*(AD 33 stands here on Saturday 2 May — the outcome under the strict equinox rule; under the observing barley threshold its 14 Nisan falls on Friday 3 April. See appendix A.2 against A.3. AD 31 too depends on the choice of rule — Wednesday 25 April or else Tuesday 27 March — but with this difference, and it is the difference this chapter turns on: with AD 33 the choice decides whether there is a Friday, with AD 31 it decides only which midweek day it becomes.)*
The crescent of 11 April AD 31 against the criteria. The three standard visibility criteria in the professional literature — Schoch's, Yallop's (1997) and Odeh's (2004) — differ only in borderline cases of a few tenths of a degree. The crescent of 11 April AD 31 is no borderline case: with an elongation of ~13.7°, a lag of ~61 minutes between sunset and moonset, and an age of ~28 hours, it lies well above the threshold of each of the three. All three: visible. (The March crescent of 12 March is a borderline case — but that does not matter, for both outcomes of that borderline case lead, thirty days on, to the same 1 Nisan = 12 April.)
The thirty-day cap — and why cloud buys no Friday. One objection to the crescent of 11 April remains open, and it is the strongest available from the other side: the sky may have been clouded that evening, and no computation can exclude weather. Granted. What the objection cannot do is produce a Friday. On an observational calendar a month caps at thirty days — the rule with which this chapter began — and Adar II began on 13 or 14 March, from the crescent of the borderline evening above or from that of 13 March, which set ninety-four minutes after the sun and lies far inside zone A. From either start the cap puts 1 Nisan no later than 13 April, and 14 Nisan no later than Thursday 26 April — which is the very shift Jeremias must plead for, and it delivers him a Friday only as the fifteenth (ch. 1). Friday 27 April as 14 Nisan would need 1 Nisan on 14 April, that is an Adar II of thirty-one days, and the cap forbids it. Cloud on 11 April and 12 April therefore changes nothing here: the cap does the work, not the crescent. To reach a Friday 14 Nisan the court would have to have missed the crescent of 13 March as well. Every Friday available in AD 31 is a Nisan 15 Friday. [CALCULATED — own reproduction, ephem 4.2.1: conjunctions 11 March 22:21 UT and 10 April 11:35 UT, weekdays from the Julian day number]
The base rate. Across the eleven years AD 26–36, 14 Nisan falls on a Wednesday only twice: AD 28 and AD 31 — two in eleven — a frequency of 0.18, and thus at the chance level of a random weekday (1/7 ≈ 0.14). (One edge case belongs here honestly, and it cuts against this book's own case: the March alternative of AD 34 gives a Wednesday on 24 March as well. That 24 March lies — contrary to what is quickly said — two days after the equinox (22 March), precisely the profile of AD 31's March option that chapter 8 takes seriously: the month begins before the equinox, the fourteenth falls after it. Under the strict "feast-after-equinox" rule that Anatolius reports from the ancients AD 34's Wednesday survives just as well as AD 31's Tuesday beneath it survives; under Anatolius' month rule it falls away and the April reading of AD 34 remains, which gives a Thursday. If one counts the March option, the Wednesday base rate becomes 3/11 and the funnel gains a second address — that AD 34 nonetheless falls away is then not a matter of the equinox but of the ministry window: a baptism in 27 admits no crucifixion in 34 (ch. 9). The measure must lie equal for both years.) A Wednesday "by itself" is therefore at chance level. The strength lies not in the Wednesday alone, but in its crossing with the day axis (see the main text). [CALCULATED · appendix A]
The eclipse — image, not argument. In the night that ushers in 15 Nisan, 25 April AD 31, a partial lunar eclipse (± 35 %) stood high in the sky above Jerusalem — at its start, 21:35 local time, thirty-four degrees above the horizon, climbing past forty; it appears, remarkably enough, in the very table of the chief defenders of the rival year (Humphreys & Waddington, 1992, p. 345). I name it here once, and after that never again as an argument. It is image, not proof; a partial eclipse does not turn the moon blood-red, and whoever hangs weight on it measures with a double standard. As devotion: striking. As evidence: nil.
Why is AD 33 the lame one, and AD 30 and AD 31 the horses? Because the crescent that counts falls in March, and the question each time is whether that March moon stands before or after the spring equinox (22/23 March). Measure that distance, and the asymmetry is hard. In AD 30 the conjunction fell almost exactly on the spring equinox (0.2 day before it), and the first crescent a good day after it — so every observing spring rule keeps Nisan in March; 14 Nisan = Friday 7 April (under the generous crescent threshold Thursday 6 April), and under no such rule a Wednesday. In AD 31 the March moon stood nine days before the equinox (the crescent; the conjunction eleven days), deep in winter. There too the rule still chooses: under Anatolius' month rule that winter moon is rejected and 14 Nisan falls on Wednesday 25 April; under the strict rule that Anatolius reports from the ancients, which demands only the feast after the equinox, it stands and the fourteenth comes on Tuesday 27 March (ch. 8, where that rule is weighed as an attribution). But — and here lies the whole matter — neither is a Friday. The choice of rule decides, in AD 31, which midweek day it becomes; it yields no Friday under any reading. Compare that with AD 33, where the March moon stood a day and a half before the equinox: there that same choice decides not between two midweek days but over the existence of the Friday itself — if the equinox is a hard lower bound, AD 33 moves to May and the Good Friday vanishes; if it is not, Friday 3 April AD 33 stays alive.
See what stands there, and see it honestly. Astronomy does not deliver the Wednesday as the winner. It delivers a two-horse race — AD 30-Friday or AD 31-Wednesday — and it weakens only the third horse, the traditional Good Friday of AD 33. Whoever reads "therefore Wednesday" out of that measures with two standards, for the same measurement that makes AD 33 totter leaves AD 30 no Wednesday. Between the two horses the sky does not choose. That choice falls on the other axis — with the day, with the length of the ministry, with the heaviest verse — and not here.
And here comes the funnel. For suppose now that the other axis has already done its work. Suppose that the text — the sign of Jonah, the two sabbaths, the fourteenth of Nisan — has led you to a Wednesday. Then walk the whole field and ask: which Wednesday? In the eleven years AD 26–36, 14 Nisan has fallen on a Wednesday exactly twice: AD 28 and AD 31. And AD 28 falls outside the window of the ministry — too early, before everything that the baptism and the temple-building and the years of John allow (ch. 9). One address remains.
Whoever arrives at a wednesday on textual grounds hasunder the observational calendar — in the whole field only one address: 25 April AD 31.
That is the shape of the whole book in one sentence: the measurement of the day times the funnel of the year. Not astronomy spitting out the date, but astronomy which, given the day, leaves exactly one year standing.
Countercasethe two rivals at their strongest
Two years deserve to be defended here at full strength, and neither is weak.
AD 33, the reigning date. The classic case is that of Humphreys and Waddington: 14 Nisan fell on Friday 3 April AD 33, and on the evening of that day a moon rose eclipsed above Jerusalem — the "blood moon" Peter cites in Acts 2:20 (Humphreys & Waddington, 1983; 1992). Astronomically AD 33 hangs on the weak equinox rule — and that rule is not simply invented: Sacha Stern, the standard work on the Jewish calendar, states expressly that people did not always wait for the equinox — "if both crops and fruits were sufficiently ripe, no intercalation would be made, even if this meant Passover occurring before the equinox. This may have happened quite frequently" (Stern, 2001, p. 70). Under that lenient rule Friday 3 April AD 33 lives. This is the strongest rival, and it is not written off here.
AD 30, the other horse. Astronomically stronger still is AD 30: its 14 Nisan falls on Friday 7 April, and that stands — as said — under every calendar rule, for the March moon stood after the equinox. AD 30 is not excluded by the sky. If it falls, it falls on the other axis: on the day arguments (chs. 3–6), on the length of the ministry and the forty-six years of John 2:20 (ch. 9), and on the heaviest verse (ch. 13). Astronomically AD 30-Friday is no more excluded by the sky than AD 31-Wednesday — though its firmness lies in "no Wednesday", not in a rock-solid Friday (the threshold evening is a borderline case Thursday/Friday; appendix C.2). That must be said, and it stands.
Excursus"computational tradition", not "measurement", and the circle of the consensus
Two honesties belong here, and the first cuts into this book's own source. The table of Parker and Dubberstein is a reconstruction of the Babylonian calendar — a fixed nineteen-year cycle. The Jewish calendar of the first century was not that; it was observational, a decision of the court upon eyewitness testimony. That 1 Nisan AD 31 fell on 12 April in Babylon therefore does not prove that it fell on 12 April in Judaea; it shows that the natural, seasonally correct placement of Nisan lay in April. For that reason I consistently call P&D a computational tradition, never a measurement of what Judaea did. The bridge from Babylon to Jerusalem is a separate case, and it belongs in Chapter 8.
The second honesty cuts into the opposing party, and it is sharp. The consensus rejects an intercalary month in AD 31 — the month that shifts Nisan to April — with a curious argument. Humphreys and Waddington themselves concede that "we possess no historical reports as to the proclamation of leap-months in the years AD 26–36" (1992, p. 336), and they dismiss the April variant with the sentence that, if Nisan was one month later, "Nisan 14 would not fall on a Friday in any year". Read that sentence again: the intercalary month is rejected because it yields no Friday. And they do retain it at the one place where it does yield one — Nisan 15 on 23 April AD 34, the single intercalary case left standing in their own Table 2. The same instrument, then, held on to where it delivers a Friday and let go where it does not. But that the crucifixion fell on a Friday is precisely what was to be demonstrated. The conclusion has become the premise. That is not a measurement of the calendar; that is a majority, smuggled in as a measure. And it is the reason AD 31 never stood on a shortlist — not because it had been weighed and found wanting, but because the axiom excluded it before the weighing (ch. 1).
Where does the inquiry stand, then, at the end of the first half of the year axis? Not at an established year. It stands at a two-horse race in which the sky does not choose, and at a funnel which, given the day, leaves one address. Astronomy does not exclude AD 31 — it supplies it, Wednesday and all. It does not exclude AD 30 — that must fall elsewhere. And it undermines only the Friday of AD 33, and even that not entirely.
But all of this rested, as you will have noticed, on one word I kept postponing: the equinox. Was Nisan permitted to begin only after the spring equinox, or not? Under the strict rule AD 33 falls away and the Wednesday of AD 31 stands hard; under the lenient rule AD 33 lives again. The whole funnel therefore stands or falls with the calendar of the temple — with the question of which rule Jerusalem actually followed. And that is the next question.
HeartHEART
"Let there be lights in the vault of heaven, to divide the day from the night; and let them be for signs and for appointed times" (Genesis 1:14). The moon that the commission stood awaiting on the hill above Jerusalem was no arbitrary lamp. It had been set — to mark the times, to carry the feasts, to date, without knowing it, the night in which the Lamb would be slain. Its positions are recomputed with machines the priests did not have. But it is the same moon, above the same city, and it kept the same time. There is something moving about a God who did not let His greatest deed take place in hiding, but hung it on a crescent everyone could see.
Note on the sources
The crescent and conjunction computations can be verified with standard ephemerides (ephem/Swiss Ephemeris) and are calibrated against an independent anchor point; the thirty-day cap in the Dossier is calendar arithmetic on top of them — the month lengths and the weekdays follow from the Julian day number and require no visibility model at all, which is why that step survives any dispute over the crescent; the reproduction of Parker's seven Jerusalem dates (1944) agrees to the day in all seven years; the article itself has been read at first hand, and his Jerusalem tabulation is quoted from it directly. The visibility criteria are the published criteria of Schoch, Yallop (1997) and Odeh (2004); the conclusion for 11 April AD 31 hangs not on an implementation of my own but on the fact that that crescent lies well above every published threshold. The eclipse of 25 April AD 31 carries no argumentative weight whatever in this chapter and is named solely as an image. Where "Wednesday 25 April AD 31" is heard, that is a computation under a documented rule — not proof, and it is not presented as proof either.
Literature (APA)
Amadon, G. (1942). The Johannine-synoptic argument. Journal of Biblical Literature, 61(4), 227–280.
Beckwith, R. T. (1996). Calendar and chronology, Jewish and Christian: Biblical, intertestamental and patristic studies. Brill. (p. 289) [consulted, not cited in this chapter]
Bond, H. K. (2013). Dating the death of Jesus: Memory and the religious imagination. New Testament Studies, 59(4), 461–475.
Humphreys, C. J., & Waddington, W. G. (1983). Dating the crucifixion. Nature, 306(5945), 743–746.
Humphreys, C. J., & Waddington, W. G. (1992). The Jewish calendar, a lunar eclipse and the date of Christ's crucifixion. Tyndale Bulletin, 43(2), 331–351. (pp. 336, 345)
Odeh, M. Sh. (2004). New criterion for lunar crescent visibility. Experimental Astronomy, 18, 39–64. https://doi.org/10.1007/s10686-005-9002-5
Parker, R. A. (1944). Ancient Jewish calendation: A criticism. Journal of Biblical Literature, 63(2), 173–176.
Parker, R. A., & Dubberstein, W. H. (1946). Babylonian chronology 626 B.C.–A.D. 45 (Studies in Ancient Oriental Civilization 24, p. 46). University of Chicago Press. [3rd ed. 1956, to A.D. 75.]
Stern, S. (2001). Calendar and community: A history of the Jewish calendar, 2nd century BCE–10th century CE (p. 70). Oxford University Press.
Yallop, B. D. (1997). A method for predicting the first sighting of the new crescent moon (NAO Technical Note No. 69). HM Nautical Almanac Office.
Scripture texts: Genesis 1:14 · Acts 2:20. My own working translation. Astronomical computations: own reproduction (ephem 4.2.1), calibrated against an independent anchor point; full tables in appendix A.