All the pieces lie on the table. The sign has been measured, the week reconstructed, the year computed through, the strongest rival built up and the reigning date laid against its own measure. Now the weighing itself — and it begins with a temptation I refuse. The temptation is to multiply all the lines into one small number, a composite p-value that would create the appearance of a proof. I shall not do that, and the refusal is not modesty but honesty. For the lines are not independent enough to multiply; whoever does so anyway builds a scaffold of sham precision. The weighing of this book does not count towards a number. It shows where the year stands, and where it tips.
Let me first say, then, what can actually be measured, for it is less than it seems and it is deliberately less. There is one real, measured base rate in the whole case: the double construction "three days and three nights" is never, in the entire corpus searched, counted inclusively (ch. 3). That is not an impression but a count, and it has been confirmed out-of-sample. And there is one named calendar premise: the intercalary month of AD 31 (ch. 8). Around these lie not ten independent proofs, but roughly two and a half effective axes — for much of what looks like separate lines in fact measures the same thing. One measured base rate, one premise, two and a half axes: that is the hard core, and more I do not claim.
Dossierwhat can be measured, and what cannot
The measured base rate. In the corpus of biblical and rabbinic texts the double construction "N days and N nights" never counts inclusively; every narrative case is full/continuous (ch. 3). Testing that as a binomial — the chance that a series of independent cases all come out full by accident while the form was supposedly elastic — the outcome lies between about p ≈ 0.008 (on a cautious count) and p ≈ 0.0005 (on the generous one). Honest limit: this is strong against the null hypothesis "the double construction is elastic by default", but it is silent against a weak null, "the form is a tendency, not a law". The base rate thus measures much, but not everything.
The weekday base rate. A Wednesday 14 Nisan falls in the window AD 26–36 twice in eleven, a frequency of 0.18, at the chance level of an arbitrary weekday (1/7 ≈ 0.14) (ch. 7). A Wednesday "by itself" is thus chance-level. The strength lies not in the Wednesday alone, but in the crossing of the measured day-axis with that rare weekday.
No composite p-value. The day-axis and the year-axis share assumptions (the calendar, the reading of the text); they are not independent, and their probabilities may not be multiplied. The roughly two and a half effective axes were counted after deducting that overlap: the day-axis (the sign as measure) counts as one, the year-axis (the waxing moon and the calendar towards the Wednesday) as one, and the chronological envelope (Tiberius, the forty-six years, Daniel) — which largely coincides with the year-axis — as at most a half. Whoever presents one composite number here is not measuring — he is building scaffolding.
What remains, and what the actual weighing is, is a qualitative assessment: how heavily does each point weigh, and against which assumption? To make that checkable, a weighted model was built — eleven criteria, five dimensions, each with a weight and a score — and it was computed through under four different weightings. What is revealing is not one outcome but the comparison: the winner shifts with the weight, and that is shown rather than hidden.
Expertthe hinge is the reading of the sign
Beneath all the weightings lies one choice, and it determines the rest: does one read the sign of Jonah as a measure (three nights literally → a Wednesday) or as an idiom (a manner of speaking → "on the third day" → a Friday)? That is not a matter of arithmetic but a hermeneutical choice, and the year-question is derived from it. Whoever reads the sign as a measure is driven towards AD 31; whoever reads it as a manner of speaking may stay with the Friday of AD 30 or AD 33. The weighing makes that dependence visible instead of blurring it — and therein lies its honesty.
Dossierfour weightings, one scale (weighted total 0–100)
Sceptical (sign = idiom, convergence removed, calendar liberty symmetrical)
80.1
75.5
76.6
30 > 33 ≈ 31
Recomputable; the scoring model and the weights stand in appendix F. The calendar liberty is applied symmetrically: whoever grants it to AD 33 (Stern's "quite frequently") grants it to AD 31 as well — neither year is pressed down or inflated. ⚠ One cell in that model stands a point above its own ground: C3, the ministry window, which gives AD 31 the top mark although the four-Passover reading is carried as support and not as compulsion and the count behind the baptism of 27 as unattested and not refuted. Lowering it to 2 would cost AD 31 2.7, 3.0, 3.2 and 3.4 points and would change the order of no row. It has not been done, and the reason is the same symmetry: AD 30 and AD 33 stand at 2 on that criterion on grounds that have not yet been measured against AD 31's, and whoever lowers only his own cell measures with two standards in the other direction. Appendix F.3 gives the figures both ways. [CALCULATED]
Read that table honestly, for it is the core of this chapter. Under the weighting this book itself lets weigh heaviest — exact fulfilment, in which the sign is a measure and God delivers to the day — AD 31 wins convincingly, with over ninety out of a hundred. Under the balanced default it stays in front. But under the idiom reading, academic or sceptical, that lead vanishes: then not AD 33 but AD 30 — the astronomically robust Friday — rises to the top. And read the second place honestly too, for it does not fall to AD 31: in the academic weighting AD 33 follows close behind AD 30 and AD 31 ends last, 6.1 points behind AD 33; in the sceptical, AD 33 follows AD 30 and AD 31 lies 1.1 points behind AD 33 — a photo finish. Under the idiom reading, then, AD 31 does not win: it ends at the back, in the sceptical weighting by the width of a hair. That is no defeat I am hiding; it is exactly what an honest weighing ought to show, and appendix F shows it figure by figure. AD 31 is the best-supported outcome under the presupposition that the sign is a measure — and that presupposition, not a sum, is what separates the one from the other.
Two things the table lays bare in passing deserve to be noted, for they cut against expectation. The first: the best Friday is not the reigning date. Under every idiom weighting AD 30 leads, not AD 33 — the consensus, then, does not even cherish the strongest Friday year (ch. 15). The second: nowhere, under no weighting, does AD 33 clearly win. The date that counts as the answer is, under the strictest measure, at best a draw, and under the measure of this book the weakest of the three.
Excursusthe foundation: why exactness is no pious surmise
The weighting that puts AD 31 first rests on one assumption: that God keeps His appointments exactly on time — not approximately, but to the day. That sounds devotional, but it can be made evidential. God fixed His own feasts to the day (Leviticus 23): the Passover on 14 Nisan, the firstfruits sheaf after it, the feast of Pentecost exactly fifty days later. And the New Testament presents redemption as the fulfilment of those feasts, on their date: "for our Passover lamb has been slain: Christ" (1 Corinthians 5:7); He was raised as the Firstfruits (1 Corinthians 15:20); the Spirit was poured out "when the day of Pentecost was fulfilled" (Acts 2:1). To expect exactness is thus no wish but an inference: where God set a date, He demonstrably delivered on the day. Whoever takes that pattern seriously reads the sign of Jonah as a measure. And the pivot rests on more than typology: four legs stand free of every typological and prophetic taste — the plain reading of the text, the internal consistency of a single year-measure, the primary-source empiricism of the observational calendar, and the astronomical check. Strike out all typology and the 457 count, and those four remain standing, and they point to the same year.
Now the honesty that carries this weighing, for without it it is a plea and not a test: the open flanks, as a standing list, each with what would break it.
Dossierthe open flanks, with their falsifiers
flank
the weak spot
what would break it
the intercalary month (ch. 8)
the Adar II of AD 31 is not documented
a primary source recording that AD 31 was not intercalated
the crescent premise (ch. 8)
crescent-or-conjunction, and the equinox rule, are a choice
evidence that first-century Judea bindingly followed one rule that excludes the April moon
the observe-or-compute premise (ch. 8, ch. 14)
the weekday hangs on observation, not the molad reckoning: observation → AD 31 = Wed; molad → AD 30 = Wed 5 April, AD 31 = Mon 26 March
evidence that the first-century court computed rather than observed
the sign as measure (ch. 3)
the whole Wednesday hangs on this reading
compelling evidence that "three days and three nights" is exclusively a manner of speaking
Luke 24:21 (ch. 13)
the softest text, the better-bargain price
— already weighed; absorbable, not solvable
the ministry length (ch. 9)
the AD 27 baptism rests on a construed way of counting — registered as unattested, not refuted, not probable but not impossible: no document attests the autumn-inclusive count for Tiberius, and the coin corpora cannot test it. And the fourth Passover that turns that baptism into AD 31 rather than AD 30 is carried as support and not as compulsion, on a test registered in advance of which two of the three criteria for it were not met (appendix E.3)
compelling evidence that Luke's fifteenth year demands the sole-reign count. ⚠ Both weakenings run in one direction, and the criterion that scores this flank in the model (C3) still gives AD 31 a 3. Appendix F.3 computes what lowering it to 2 would cost and states why that has not been done: AD 30 and AD 33 carry a 2 there on grounds that have not yet been measured against AD 31's, and lowering only one's own cell is a double standard in the other direction
the chronology of Paul (Acts and Galatians)
the book does not measure this front, and that is a choice that must be accounted for
a dated, independent fixed point in Acts 1–9 that pins the conversion before or after AD 31
That seventh row is new and it deserves a word, for it is the flank on which an opponent pushes hardest. The reasoning runs: Galatians gives a timeline (conversion → three years → fourteen years), the second Jerusalem visit is the famine journey of Acts 11:30, which falls in AD 44 because the death of Agrippa I brackets it, so the conversion falls in 31, so the cross in 30. That front has been measured — the full dossier lies in the underpinning work — and the outcome is that it measures nothing.
The reason is one quantity: the length of Acts 1 to 9, between the ascension and the conversion of Paul. It is nowhere given. The text offers no time-marker at all, and both parties fill it in with an assumption without a witness — the one a ceiling of nine months, the other a floor of years. For every candidate year between 29 and 33 there is a length that closes the chain. The two scholars who deploy this chain against AD 31 moreover contradict each other on the key points: where the one needs the bracket of AD 44, the other demolishes it, and where the other moves the journey to the year 48, the grammar of Acts 11:30 breaks him. No shared construction is left from which a crucifixion year follows.
And here follows the self-binding, for without it this would be a trick. Whoever declares an instrument blunt against his own year may not reload it against another man's year. This front therefore counts nowhere in this weighing — not as relief for AD 31, and not as an objection to AD 33. That on some reconstructions the chain sits tighter for AD 33 than for AD 31 is true, and it is here expressly not used. A front that measures the length of Acts 1 to 9 and not the cross measures for nobody.
One datum within this front seemed the exception, and it has been measured separately: the escape from Damascus under the governor of king Aretas (2 Corinthians 11:32). It counted as the one thing that can date the chain at the near end without running through the length of Acts 1 to 9. On measurement it cannot, and the reason is worth the trouble.
Paul does not say that Aretas held Damascus. What stands there is "the ethnarch of king Aretas" — and the same sentence does say "the city of the Damascenes" when it means the city. Josephus does not name Damascus once in the whole Aretas episode. The supposed Nabataean rule over the city rests on one argument, and that argument is a gap in the coin series: the last city coin of Damascus dates from AD 33/34, the next from AD 65/66.
That gap is real — and it refutes the inference drawn from it. It runs on until twenty-five years after the death of Aretas; the same series has an unexplained gap of twenty-three years in the middle of undisputed Roman times; and Damascus struck coin in only about one year in nine. Decisive is the control: when an earlier Aretas really did hold the city, he struck coin there at once, with his own royal title and the city goddess on it. Of Paul's Aretas not one Damascene coin is known — in those years he strikes at Petra.
And that verdict is not mine alone; the curious thing is that the two camps which fight each other over this verse say the same thing here. F. F. Bruce, who defends the non-territorial reading, names the coin gap and concludes that for it "there is no positive evidence". G. W. Bowersock, who does assume that Aretas held the city for a short while, disarms it from the other side — if that occupation lasted less than a year, the absence of coinage "would not be very disturbing". A defender and an opponent, and both reckon the gap worth nothing. Then it is worth nothing.
And with that the datum closes like a pincer. Under the reading that leaves the year open it dates nothing — the event could then fall in any year of half a century. Under the reading that makes an anchor of it, it points to 36 to 38, and that is later than this reconstruction needs. There is no reading in which it both dates and lands where this book would want it.
The honest conclusion, and it stands here as it is: 2 Corinthians 11:32 does not exclude AD 31, but neither does it support it. The datum is chronologically indeterminate, and this book does not adduce it as an argument.What does still press is not a coin but a mechanism: in the year this reconstruction asks for, there is no event by which the arm of Aretas reaches as far as Damascus — while for a few years later there is one. That is no refutation; it is an unexplained crease, and it belongs on the record. [the coin series DOCUMENTED; the weighing CONSTRUED]
Two of these seven are sharp breaking points: an excavation or a tablet ruling out an intercalary month in AD 31, or one clear pre-modern use of "three days and three nights" — in the corpus of the idiom itself, not in a later exegetical harmonisation about it (appendix B §B.3c weighs the candidates that have been put forward) — for demonstrably less than three full twenty-four-hour days. Both are concrete and testable. That the model names them itself is not its weakness but its credential.
There is one rule that keeps this weighing clean, and I write it out. The scales count only the load-bearing core: the measured day-axis, the two-horse race and the funnel, the calendar premise, the ministry window. The corroborating rings — the rabbinic front (ch. 5), the earliest voices (ch. 6), the birth-ring at the far end of the clock (ch. 10, which tests the baptism of 27 but does not pin it), the earth as witness (ch. 12), the calibration against AD 37 — appear in the weighing only under the heading "corroboration, each with its own base rate", and they do not carry the weight. The references run in one direction: a ring may cite the core, the core never a ring. Thus no weak strut can nest itself unnoticed in the load-bearing structure.
Excursusthe firewall sentence, and what did not count
And here the hinge of the whole book, out loud. In this weighing, the prophetic clock of Daniel — the seventy weeks in their full elaboration, the world-week, the jubilee years — did not count as a load-bearing beam, and that is by design. The 457 line was weighed in as one convergence line, expressly marked "measurable on a construed premise"; and precisely for that reason it was left out of the sceptical weighting. See what that yields: even without one trace of the prophetic clock — in the sceptical column, where it has been removed — AD 31 stays in the race, albeit last of the three (the scores stand in the weighing table above). The measuring core, then, does not lean on the construed clock. Whoever rejects the historicist framework of Daniel leaves the whole weighing standing; whoever accepts it gains a second, measurable witness. That is the meaning of the order: what comes in the next part is an independent confirmation for whoever shares its premise, not a hidden foundation beneath the measurement.
Where does the inquiry stand, then, after everything? Not at a proof — that word does not fit here, and it has not fitted anywhere in this book. It stands at a weighed surmise, and it is this: Wednesday 25 April of the year 31 is the best-supported outcome under the presupposition that the sign of Jonah is a measure and that God delivers on the day. Under that presupposition it wins convincingly; under its opposite it falls back into a three-way race in which no year clearly wins — and in which not it but the Friday of AD 30 leads slightly. More than this the book does not claim, and less it does not offer. A firm hand on the structure, a loose hand on the year — and an open hand towards the reader, who may weigh for himself which presupposition he brings.
And only now, after the weighing and not before it, may a second clock be given the floor — the seventy weeks of Daniel, which on entirely different grounds arrive at that same year. It has not carried in the weighing; it comes as a witness beside the measurement, not beneath it. That is the next, and last, movement.
HeartHEART
It takes courage to weigh your own case honestly — to let the opponent speak at his strongest, to show the three-way race instead of leaving it out, to say: under this weighting I win, under that one I do not. But it is also a form of reverence. For the truth needs no thumb on the scales; it is not made greater by exaggeration and not made smaller by honesty. And so the posture that carries the whole book remains: firm on the structure, light on the year. AD 31 is commended, not imposed — carried by the presupposition that God is precise with His times, and released to whoever does not share that presupposition. That is no weakness of the argument. It is the reason it can be trusted.
Note on the sources
The measured base rate (the double construction) and the weekday base rate are recomputable (ch. 3, ch. 7; appendices B and A). The weighted scoring model, the eleven criteria, the weights and the four weightings have been independently computed through and stand in appendix F; the calendar liberty is applied symmetrically, so that no year is pressed down or inflated. One cell — C3, the ministry window — stands a point above its own ground since the four-Passover reading was reduced to support and not compulsion; appendix F.3 sets out what lowering it from 3 to 2 would cost (2.7, 3.0, 3.2 and 3.4 points, no order altered) and why it has not been done: AD 30 and AD 33 carry a 2 on that same criterion on grounds not yet measured against AD 31's. This chapter delivers no composite p-value and no proof; it shows that AD 31 is the best-supported outcome under the premise that the sign is a measure, and that under the opposite premise it falls back into a three-way race in which AD 30 leads slightly and AD 31 ends last. The prophetic clock has not carried in the weighing; the 457 line counted only as a premise-dependent convergence line and was left out of the sceptical weighting. This chapter claims no proof.
Literature (APA)
Stern, S. (2001). Calendar and community: A history of the Jewish calendar, 2nd century BCE–10th century CE. Oxford University Press.
Primary texts: Leviticus 23 · 1 Corinthians 5:7; 15:20 · Acts 2:1. The scoring model and the base-rate computations are the author's own, reproducible work (appendices B and F); the weighing builds on the measurements of ch. 3, 7, 8, 9, 13, 14 and 15.
All the pieces lie on the table. The sign has been measured, the week reconstructed, the year computed through, the strongest rival built up and the reigning date laid against its own measure. Now the weighing itself — and it begins with a temptation I refuse. The temptation is to multiply all the lines into one small number, a composite p-value that would create the appearance of a proof. I shall not do that, and the refusal is not modesty but honesty. For the lines are not independent enough to multiply; whoever does so anyway builds a scaffold of sham precision. The weighing of this book does not count towards a number. It shows where the year stands, and where it tips.
Let me first say, then, what can actually be measured, for it is less than it seems and it is deliberately less. There is one real, measured base rate in the whole case: the double construction "three days and three nights" is never, in the entire corpus searched, counted inclusively (ch. 3). That is not an impression but a count, and it has been confirmed out-of-sample. And there is one named calendar premise: the intercalary month of AD 31 (ch. 8). Around these lie not ten independent proofs, but roughly two and a half effective axes — for much of what looks like separate lines in fact measures the same thing. One measured base rate, one premise, two and a half axes: that is the hard core, and more I do not claim.
Dossierwhat can be measured, and what cannot
The measured base rate. In the corpus of biblical and rabbinic texts the double construction "N days and N nights" never counts inclusively; every narrative case is full/continuous (ch. 3). Testing that as a binomial — the chance that a series of independent cases all come out full by accident while the form was supposedly elastic — the outcome lies between about p ≈ 0.008 (on a cautious count) and p ≈ 0.0005 (on the generous one). Honest limit: this is strong against the null hypothesis "the double construction is elastic by default", but it is silent against a weak null, "the form is a tendency, not a law". The base rate thus measures much, but not everything.
The weekday base rate. A Wednesday 14 Nisan falls in the window AD 26–36 twice in eleven, a frequency of 0.18, at the chance level of an arbitrary weekday (1/7 ≈ 0.14) (ch. 7). A Wednesday "by itself" is thus chance-level. The strength lies not in the Wednesday alone, but in the crossing of the measured day-axis with that rare weekday.
No composite p-value. The day-axis and the year-axis share assumptions (the calendar, the reading of the text); they are not independent, and their probabilities may not be multiplied. The roughly two and a half effective axes were counted after deducting that overlap: the day-axis (the sign as measure) counts as one, the year-axis (the waxing moon and the calendar towards the Wednesday) as one, and the chronological envelope (Tiberius, the forty-six years, Daniel) — which largely coincides with the year-axis — as at most a half. Whoever presents one composite number here is not measuring — he is building scaffolding.
What remains, and what the actual weighing is, is a qualitative assessment: how heavily does each point weigh, and against which assumption? To make that checkable, a weighted model was built — eleven criteria, five dimensions, each with a weight and a score — and it was computed through under four different weightings. What is revealing is not one outcome but the comparison: the winner shifts with the weight, and that is shown rather than hidden.
Expertthe hinge is the reading of the sign
Beneath all the weightings lies one choice, and it determines the rest: does one read the sign of Jonah as a measure (three nights literally → a Wednesday) or as an idiom (a manner of speaking → "on the third day" → a Friday)? That is not a matter of arithmetic but a hermeneutical choice, and the year-question is derived from it. Whoever reads the sign as a measure is driven towards AD 31; whoever reads it as a manner of speaking may stay with the Friday of AD 30 or AD 33. The weighing makes that dependence visible instead of blurring it — and therein lies its honesty.
Dossierfour weightings, one scale (weighted total 0–100)
Recomputable; the scoring model and the weights stand in appendix F. The calendar liberty is applied symmetrically: whoever grants it to AD 33 (Stern's "quite frequently") grants it to AD 31 as well — neither year is pressed down or inflated. ⚠ One cell in that model stands a point above its own ground: C3, the ministry window, which gives AD 31 the top mark although the four-Passover reading is carried as support and not as compulsion and the count behind the baptism of 27 as unattested and not refuted. Lowering it to 2 would cost AD 31 2.7, 3.0, 3.2 and 3.4 points and would change the order of no row. It has not been done, and the reason is the same symmetry: AD 30 and AD 33 stand at 2 on that criterion on grounds that have not yet been measured against AD 31's, and whoever lowers only his own cell measures with two standards in the other direction. Appendix F.3 gives the figures both ways. [CALCULATED]
Read that table honestly, for it is the core of this chapter. Under the weighting this book itself lets weigh heaviest — exact fulfilment, in which the sign is a measure and God delivers to the day — AD 31 wins convincingly, with over ninety out of a hundred. Under the balanced default it stays in front. But under the idiom reading, academic or sceptical, that lead vanishes: then not AD 33 but AD 30 — the astronomically robust Friday — rises to the top. And read the second place honestly too, for it does not fall to AD 31: in the academic weighting AD 33 follows close behind AD 30 and AD 31 ends last, 6.1 points behind AD 33; in the sceptical, AD 33 follows AD 30 and AD 31 lies 1.1 points behind AD 33 — a photo finish. Under the idiom reading, then, AD 31 does not win: it ends at the back, in the sceptical weighting by the width of a hair. That is no defeat I am hiding; it is exactly what an honest weighing ought to show, and appendix F shows it figure by figure. AD 31 is the best-supported outcome under the presupposition that the sign is a measure — and that presupposition, not a sum, is what separates the one from the other.
Two things the table lays bare in passing deserve to be noted, for they cut against expectation. The first: the best Friday is not the reigning date. Under every idiom weighting AD 30 leads, not AD 33 — the consensus, then, does not even cherish the strongest Friday year (ch. 15). The second: nowhere, under no weighting, does AD 33 clearly win. The date that counts as the answer is, under the strictest measure, at best a draw, and under the measure of this book the weakest of the three.
Excursusthe foundation: why exactness is no pious surmise
The weighting that puts AD 31 first rests on one assumption: that God keeps His appointments exactly on time — not approximately, but to the day. That sounds devotional, but it can be made evidential. God fixed His own feasts to the day (Leviticus 23): the Passover on 14 Nisan, the firstfruits sheaf after it, the feast of Pentecost exactly fifty days later. And the New Testament presents redemption as the fulfilment of those feasts, on their date: "for our Passover lamb has been slain: Christ" (1 Corinthians 5:7); He was raised as the Firstfruits (1 Corinthians 15:20); the Spirit was poured out "when the day of Pentecost was fulfilled" (Acts 2:1). To expect exactness is thus no wish but an inference: where God set a date, He demonstrably delivered on the day. Whoever takes that pattern seriously reads the sign of Jonah as a measure. And the pivot rests on more than typology: four legs stand free of every typological and prophetic taste — the plain reading of the text, the internal consistency of a single year-measure, the primary-source empiricism of the observational calendar, and the astronomical check. Strike out all typology and the 457 count, and those four remain standing, and they point to the same year.
Now the honesty that carries this weighing, for without it it is a plea and not a test: the open flanks, as a standing list, each with what would break it.
Dossierthe open flanks, with their falsifiers
That seventh row is new and it deserves a word, for it is the flank on which an opponent pushes hardest. The reasoning runs: Galatians gives a timeline (conversion → three years → fourteen years), the second Jerusalem visit is the famine journey of Acts 11:30, which falls in AD 44 because the death of Agrippa I brackets it, so the conversion falls in 31, so the cross in 30. That front has been measured — the full dossier lies in the underpinning work — and the outcome is that it measures nothing.
The reason is one quantity: the length of Acts 1 to 9, between the ascension and the conversion of Paul. It is nowhere given. The text offers no time-marker at all, and both parties fill it in with an assumption without a witness — the one a ceiling of nine months, the other a floor of years. For every candidate year between 29 and 33 there is a length that closes the chain. The two scholars who deploy this chain against AD 31 moreover contradict each other on the key points: where the one needs the bracket of AD 44, the other demolishes it, and where the other moves the journey to the year 48, the grammar of Acts 11:30 breaks him. No shared construction is left from which a crucifixion year follows.
And here follows the self-binding, for without it this would be a trick. Whoever declares an instrument blunt against his own year may not reload it against another man's year. This front therefore counts nowhere in this weighing — not as relief for AD 31, and not as an objection to AD 33. That on some reconstructions the chain sits tighter for AD 33 than for AD 31 is true, and it is here expressly not used. A front that measures the length of Acts 1 to 9 and not the cross measures for nobody.
One datum within this front seemed the exception, and it has been measured separately: the escape from Damascus under the governor of king Aretas (2 Corinthians 11:32). It counted as the one thing that can date the chain at the near end without running through the length of Acts 1 to 9. On measurement it cannot, and the reason is worth the trouble.
Paul does not say that Aretas held Damascus. What stands there is "the ethnarch of king Aretas" — and the same sentence does say "the city of the Damascenes" when it means the city. Josephus does not name Damascus once in the whole Aretas episode. The supposed Nabataean rule over the city rests on one argument, and that argument is a gap in the coin series: the last city coin of Damascus dates from AD 33/34, the next from AD 65/66.
That gap is real — and it refutes the inference drawn from it. It runs on until twenty-five years after the death of Aretas; the same series has an unexplained gap of twenty-three years in the middle of undisputed Roman times; and Damascus struck coin in only about one year in nine. Decisive is the control: when an earlier Aretas really did hold the city, he struck coin there at once, with his own royal title and the city goddess on it. Of Paul's Aretas not one Damascene coin is known — in those years he strikes at Petra.
And that verdict is not mine alone; the curious thing is that the two camps which fight each other over this verse say the same thing here. F. F. Bruce, who defends the non-territorial reading, names the coin gap and concludes that for it "there is no positive evidence". G. W. Bowersock, who does assume that Aretas held the city for a short while, disarms it from the other side — if that occupation lasted less than a year, the absence of coinage "would not be very disturbing". A defender and an opponent, and both reckon the gap worth nothing. Then it is worth nothing.
And with that the datum closes like a pincer. Under the reading that leaves the year open it dates nothing — the event could then fall in any year of half a century. Under the reading that makes an anchor of it, it points to 36 to 38, and that is later than this reconstruction needs. There is no reading in which it both dates and lands where this book would want it.
The honest conclusion, and it stands here as it is: 2 Corinthians 11:32 does not exclude AD 31, but neither does it support it. The datum is chronologically indeterminate, and this book does not adduce it as an argument. What does still press is not a coin but a mechanism: in the year this reconstruction asks for, there is no event by which the arm of Aretas reaches as far as Damascus — while for a few years later there is one. That is no refutation; it is an unexplained crease, and it belongs on the record. [the coin series DOCUMENTED; the weighing CONSTRUED]
Two of these seven are sharp breaking points: an excavation or a tablet ruling out an intercalary month in AD 31, or one clear pre-modern use of "three days and three nights" — in the corpus of the idiom itself, not in a later exegetical harmonisation about it (appendix B §B.3c weighs the candidates that have been put forward) — for demonstrably less than three full twenty-four-hour days. Both are concrete and testable. That the model names them itself is not its weakness but its credential.
There is one rule that keeps this weighing clean, and I write it out. The scales count only the load-bearing core: the measured day-axis, the two-horse race and the funnel, the calendar premise, the ministry window. The corroborating rings — the rabbinic front (ch. 5), the earliest voices (ch. 6), the birth-ring at the far end of the clock (ch. 10, which tests the baptism of 27 but does not pin it), the earth as witness (ch. 12), the calibration against AD 37 — appear in the weighing only under the heading "corroboration, each with its own base rate", and they do not carry the weight. The references run in one direction: a ring may cite the core, the core never a ring. Thus no weak strut can nest itself unnoticed in the load-bearing structure.
Excursusthe firewall sentence, and what did not count
And here the hinge of the whole book, out loud. In this weighing, the prophetic clock of Daniel — the seventy weeks in their full elaboration, the world-week, the jubilee years — did not count as a load-bearing beam, and that is by design. The 457 line was weighed in as one convergence line, expressly marked "measurable on a construed premise"; and precisely for that reason it was left out of the sceptical weighting. See what that yields: even without one trace of the prophetic clock — in the sceptical column, where it has been removed — AD 31 stays in the race, albeit last of the three (the scores stand in the weighing table above). The measuring core, then, does not lean on the construed clock. Whoever rejects the historicist framework of Daniel leaves the whole weighing standing; whoever accepts it gains a second, measurable witness. That is the meaning of the order: what comes in the next part is an independent confirmation for whoever shares its premise, not a hidden foundation beneath the measurement.
Where does the inquiry stand, then, after everything? Not at a proof — that word does not fit here, and it has not fitted anywhere in this book. It stands at a weighed surmise, and it is this: Wednesday 25 April of the year 31 is the best-supported outcome under the presupposition that the sign of Jonah is a measure and that God delivers on the day. Under that presupposition it wins convincingly; under its opposite it falls back into a three-way race in which no year clearly wins — and in which not it but the Friday of AD 30 leads slightly. More than this the book does not claim, and less it does not offer. A firm hand on the structure, a loose hand on the year — and an open hand towards the reader, who may weigh for himself which presupposition he brings.
And only now, after the weighing and not before it, may a second clock be given the floor — the seventy weeks of Daniel, which on entirely different grounds arrive at that same year. It has not carried in the weighing; it comes as a witness beside the measurement, not beneath it. That is the next, and last, movement.
HeartHEART
It takes courage to weigh your own case honestly — to let the opponent speak at his strongest, to show the three-way race instead of leaving it out, to say: under this weighting I win, under that one I do not. But it is also a form of reverence. For the truth needs no thumb on the scales; it is not made greater by exaggeration and not made smaller by honesty. And so the posture that carries the whole book remains: firm on the structure, light on the year. AD 31 is commended, not imposed — carried by the presupposition that God is precise with His times, and released to whoever does not share that presupposition. That is no weakness of the argument. It is the reason it can be trusted.
Note on the sources
The measured base rate (the double construction) and the weekday base rate are recomputable (ch. 3, ch. 7; appendices B and A). The weighted scoring model, the eleven criteria, the weights and the four weightings have been independently computed through and stand in appendix F; the calendar liberty is applied symmetrically, so that no year is pressed down or inflated. One cell — C3, the ministry window — stands a point above its own ground since the four-Passover reading was reduced to support and not compulsion; appendix F.3 sets out what lowering it from 3 to 2 would cost (2.7, 3.0, 3.2 and 3.4 points, no order altered) and why it has not been done: AD 30 and AD 33 carry a 2 on that same criterion on grounds not yet measured against AD 31's. This chapter delivers no composite p-value and no proof; it shows that AD 31 is the best-supported outcome under the premise that the sign is a measure, and that under the opposite premise it falls back into a three-way race in which AD 30 leads slightly and AD 31 ends last. The prophetic clock has not carried in the weighing; the 457 line counted only as a premise-dependent convergence line and was left out of the sceptical weighting. This chapter claims no proof.
Literature (APA)
Stern, S. (2001). Calendar and community: A history of the Jewish calendar, 2nd century BCE–10th century CE. Oxford University Press.
Primary texts: Leviticus 23 · 1 Corinthians 5:7; 15:20 · Acts 2:1. The scoring model and the base-rate computations are the author's own, reproducible work (appendices B and F); the weighing builds on the measurements of ch. 3, 7, 8, 9, 13, 14 and 15.